• Re: The Halting Problem asks for too much

    From olcott@polcott333@gmail.com to comp.theory,sci.logic,sci.math,comp.software-eng,comp.ai.philosophy on Sun Jan 11 08:18:11 2026
    From Newsgroup: comp.ai.philosophy

    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    The counter-example input to requires more than
    can be derived from finite string transformation
    rules applied to this specific input thus the
    Halting Problem requires too much.

    In a sense the halting problem asks too much: the problem is >>>>>>>>> proven to
    be unsolvable. In another sense it asks too little: usually we >>>>>>>>> want to
    know whether a method halts on every input, not just one.

    Although the halting problem is unsolvable, there are partial >>>>>>>>> solutions
    to the halting problem. In particular, every counter-example to >>>>>>>>> the
    full solution is correctly solved by some partial deciders.

    *if undecidability is correct then truth itself is broken*

    Depends on whether the word "truth" is interpeted in the standard >>>>>>> sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory
    expressions are correctly rejected as semantically
    incoherent thus form no undecidability or incompleteness.

    The misconception is yours. No expression in the language of the first >>>>> order group theory is self-contradictory. But the first order goupr
    theory is incomplete: it is impossible to prove that AB = BA is true >>>>> for every A and every B but it is also impossible to prove that AB
    = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying
    finite string transformation rules to actual finite
    string inputs, then the required result exceeds the
    scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be derived by
    appying a finite string transformation then the it it is uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement.

    You can't determine whether the required result is computable before
    you have the requirement.



    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    *ChatGPT explains how and why I am correct*

    *Reinterpretation of undecidability*
    The example of P and H demonstrates that what is
    often called “undecidable” is better understood as
    ill-posed with respect to computable semantics.
    When the specification is constrained to properties
    detectable via finite simulation and finite pattern
    recognition, computation proceeds normally and
    correctly. Undecidability only appears when the
    specification overreaches that boundary.

    Every other LLM says this same thing using
    different words.


    Of course, it one can prove that the required result is not computable
    then that helps to avoid wasting effort to try the impossible. The
    situation is worse if it is not known that the required result is not
    computable.

    That something is not computable does not mean that there is anyting
    "incorrect" in the requirement.

    Yes it certainly does. Requiring the impossible is always an error.
    Requiring an answer to a yes/no question that has no correct yes/no
    answer is an incorrect question that must be rejected.

    In order to claim that a requirement
    is incorrect one must at least prove that the requirement does not
    serve its intended purpose.

    Requiring the impossible cannot possibly serve any purpose
    except perhaps to exemplify one's own ignorance.

    Even then it is possible that the
    requirement serves some other purpose. Even if a requirement serves
    no purpose that need not mean that it be "incorrect", only that it
    is useless.






    --
    Copyright 2026 Olcott<br><br>

    My 28 year goal has been to make <br>
    "true on the basis of meaning expressed in language"<br>
    reliably computable.<br><br>

    This required establishing a new foundation<br>
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to comp.theory,sci.logic,comp.software-eng,sci.math,comp.ai.philosophy on Sun Jan 11 08:23:00 2026
    From Newsgroup: comp.ai.philosophy

    On 1/11/2026 4:22 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    The counter-example input to requires more than
    can be derived from finite string transformation
    rules applied to this specific input thus the
    Halting Problem requires too much.

    In a sense the halting problem asks too much: the problem is >>>>>>>>> proven to
    be unsolvable. In another sense it asks too little: usually we >>>>>>>>> want to
    know whether a method halts on every input, not just one.

    Although the halting problem is unsolvable, there are partial >>>>>>>>> solutions
    to the halting problem. In particular, every counter-example to >>>>>>>>> the
    full solution is correctly solved by some partial deciders.

    *if undecidability is correct then truth itself is broken*

    Depends on whether the word "truth" is interpeted in the standard >>>>>>> sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory
    expressions are correctly rejected as semantically
    incoherent thus form no undecidability or incompleteness.

    The misconception is yours. No expression in the language of the first >>>>> order group theory is self-contradictory. But the first order goupr
    theory is incomplete: it is impossible to prove that AB = BA is true >>>>> for every A and every B but it is also impossible to prove that AB
    = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying
    finite string transformation rules to actual finite
    string inputs, then the required result exceeds the
    scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be derived by
    appying a finite string transformation then the it it is uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement.

    Of course, it one can prove that the required result is not computable
    then that helps to avoid wasting effort to try the impossible. The
    situation is worse if it is not known that the required result is not
    computable.

    That something is not computable does not mean that there is anyting
    "incorrect" in the requirement.

    Yes it certainly does. Requiring the impossible is always an error.

    It is a perfectly valid question to ask whther a particular reuqirement
    is satisfiable.


    Any yes/no question lacking a correct yes/no answer
    is an incorrect question that must be rejected on
    that basis.

    The whole rest of the world is too stupid to even
    reject self-contradictory expressions such as the
    Liar Paradox: "This sentence is not true".

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    *ChatGPT explains how and why I am correct*

    *Reinterpretation of undecidability*
    The example of P and H demonstrates that what is
    often called “undecidable” is better understood as
    ill-posed with respect to computable semantics.
    When the specification is constrained to properties
    detectable via finite simulation and finite pattern
    recognition, computation proceeds normally and
    correctly. Undecidability only appears when the
    specification overreaches that boundary.
    --
    Copyright 2026 Olcott<br><br>

    My 28 year goal has been to make <br>
    "true on the basis of meaning expressed in language"<br>
    reliably computable.<br><br>

    This required establishing a new foundation<br>
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Mikko@mikko.levanto@iki.fi to comp.theory,sci.logic,sci.math,comp.software-eng,comp.ai.philosophy on Mon Jan 12 12:44:51 2026
    From Newsgroup: comp.ai.philosophy

    On 11/01/2026 16:18, olcott wrote:
    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    The counter-example input to requires more than
    can be derived from finite string transformation
    rules applied to this specific input thus the
    Halting Problem requires too much.

    In a sense the halting problem asks too much: the problem is >>>>>>>>>> proven to
    be unsolvable. In another sense it asks too little: usually we >>>>>>>>>> want to
    know whether a method halts on every input, not just one.

    Although the halting problem is unsolvable, there are partial >>>>>>>>>> solutions
    to the halting problem. In particular, every counter-example >>>>>>>>>> to the
    full solution is correctly solved by some partial deciders. >>>>>>>>>
    *if undecidability is correct then truth itself is broken*

    Depends on whether the word "truth" is interpeted in the standard >>>>>>>> sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory
    expressions are correctly rejected as semantically
    incoherent thus form no undecidability or incompleteness.

    The misconception is yours. No expression in the language of the
    first
    order group theory is self-contradictory. But the first order goupr >>>>>> theory is incomplete: it is impossible to prove that AB = BA is true >>>>>> for every A and every B but it is also impossible to prove that AB >>>>>> = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying
    finite string transformation rules to actual finite
    string inputs, then the required result exceeds the
    scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be derived by
    appying a finite string transformation then the it it is uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement.

    You can't determine whether the required result is computable before
    you have the requirement.

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    Which only nmakes sense when the requirement that H must determine
    whether the computation presented by its input halts has already
    been presented.

    *ChatGPT explains how and why I am correct*

      *Reinterpretation of undecidability*
      The example of P and H demonstrates that what is
      often called “undecidable” is better understood as
      ill-posed with respect to computable semantics.
      When the specification is constrained to properties
      detectable via finite simulation and finite pattern
      recognition, computation proceeds normally and
      correctly. Undecidability only appears when the
      specification overreaches that boundary.

    It tries to explain but it does not prove.
    --
    Mikko
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Mikko@mikko.levanto@iki.fi to comp.theory,sci.logic,comp.software-eng,sci.math,comp.ai.philosophy on Mon Jan 12 12:51:55 2026
    From Newsgroup: comp.ai.philosophy

    On 11/01/2026 16:23, olcott wrote:
    On 1/11/2026 4:22 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    The counter-example input to requires more than
    can be derived from finite string transformation
    rules applied to this specific input thus the
    Halting Problem requires too much.

    In a sense the halting problem asks too much: the problem is >>>>>>>>>> proven to
    be unsolvable. In another sense it asks too little: usually we >>>>>>>>>> want to
    know whether a method halts on every input, not just one.

    Although the halting problem is unsolvable, there are partial >>>>>>>>>> solutions
    to the halting problem. In particular, every counter-example >>>>>>>>>> to the
    full solution is correctly solved by some partial deciders. >>>>>>>>>
    *if undecidability is correct then truth itself is broken*

    Depends on whether the word "truth" is interpeted in the standard >>>>>>>> sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory
    expressions are correctly rejected as semantically
    incoherent thus form no undecidability or incompleteness.

    The misconception is yours. No expression in the language of the
    first
    order group theory is self-contradictory. But the first order goupr >>>>>> theory is incomplete: it is impossible to prove that AB = BA is true >>>>>> for every A and every B but it is also impossible to prove that AB >>>>>> = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying
    finite string transformation rules to actual finite
    string inputs, then the required result exceeds the
    scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be derived by
    appying a finite string transformation then the it it is uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement.

    Of course, it one can prove that the required result is not computable >>>> then that helps to avoid wasting effort to try the impossible. The
    situation is worse if it is not known that the required result is not
    computable.

    That something is not computable does not mean that there is anyting
    "incorrect" in the requirement.

    Yes it certainly does. Requiring the impossible is always an error.

    It is a perfectly valid question to ask whther a particular reuqirement
    is satisfiable.

    Any yes/no question lacking a correct yes/no answer
    is an incorrect question that must be rejected on
    that basis.

    Irrelevant. The question whether a particular requirement is satisfiable
    does have an answer that is either "yes" or "no". In some ases it is
    not known whether it is "yes" or "no" and there may be no known way to
    find out be even then either "yes" or "no" is the correct answer.
    --
    Mikko
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Mon Jan 12 08:29:00 2026
    From Newsgroup: comp.ai.philosophy

    On 1/12/2026 4:44 AM, Mikko wrote:
    On 11/01/2026 16:18, olcott wrote:
    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    The counter-example input to requires more than
    can be derived from finite string transformation
    rules applied to this specific input thus the
    Halting Problem requires too much.

    In a sense the halting problem asks too much: the problem is >>>>>>>>>>> proven to
    be unsolvable. In another sense it asks too little: usually >>>>>>>>>>> we want to
    know whether a method halts on every input, not just one. >>>>>>>>>>>
    Although the halting problem is unsolvable, there are partial >>>>>>>>>>> solutions
    to the halting problem. In particular, every counter-example >>>>>>>>>>> to the
    full solution is correctly solved by some partial deciders. >>>>>>>>>>
    *if undecidability is correct then truth itself is broken*

    Depends on whether the word "truth" is interpeted in the standard >>>>>>>>> sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory
    expressions are correctly rejected as semantically
    incoherent thus form no undecidability or incompleteness.

    The misconception is yours. No expression in the language of the >>>>>>> first
    order group theory is self-contradictory. But the first order goupr >>>>>>> theory is incomplete: it is impossible to prove that AB = BA is true >>>>>>> for every A and every B but it is also impossible to prove that >>>>>>> AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying
    finite string transformation rules to actual finite
    string inputs, then the required result exceeds the
    scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be derived by
    appying a finite string transformation then the it it is uncomputable. >>>>
    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement.

    You can't determine whether the required result is computable before
    you have the requirement.

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    Which only nmakes sense when the requirement that H must determine
    whether the computation presented by its input halts has already
    been presented.

    *ChatGPT explains how and why I am correct*

       *Reinterpretation of undecidability*
       The example of P and H demonstrates that what is
       often called “undecidable” is better understood as
       ill-posed with respect to computable semantics.
       When the specification is constrained to properties
       detectable via finite simulation and finite pattern
       recognition, computation proceeds normally and
       correctly. Undecidability only appears when the
       specification overreaches that boundary.

    It tries to explain but it does not prove.


    Its the same thing that I have been saying for years.
    It is not that a universal halt decider cannot exist.

    It is that an input that does the opposite of whatever
    value the halt decider returns is non-well-founded
    within proof-theoretic semantics.
    --
    Copyright 2026 Olcott<br><br>

    My 28 year goal has been to make <br>
    "true on the basis of meaning expressed in language"<br>
    reliably computable.<br><br>

    This required establishing a new foundation<br>
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Mon Jan 12 08:32:20 2026
    From Newsgroup: comp.ai.philosophy

    On 1/12/2026 4:47 AM, Mikko wrote:
    On 11/01/2026 16:24, Tristan Wibberley wrote:
    On 11/01/2026 10:13, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:

    No, that does not follow. If a required result cannot be derived by
    appying a finite string transformation then the it it is uncomputable. >>>>
    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement.

    You can't determine whether the required result is computable before
    you have the requirement.


    Right, it is /in/ scope for computer science... for the /ology/. Olcott
    here uses "computation" to refer to the practice. You give the
    requirement to the /ologist/ who correctly decides that it is not for
    computation because it is not computable.

    You two so often violently agree; I find it warming to the heart.

    For pracitcal programming it is useful to know what is known to be uncomputable in order to avoid wasting time in attemlpts to do the impossible.


    It f-cking nuts that after more than 2000 years
    people still don't understand that self-contradictory
    expressions: "This sentence is not true" have no
    truth value. A smart high school student should have
    figured this out 2000 years ago.
    --
    Copyright 2026 Olcott<br><br>

    My 28 year goal has been to make <br>
    "true on the basis of meaning expressed in language"<br>
    reliably computable.<br><br>

    This required establishing a new foundation<br>
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Richard Damon@Richard@Damon-Family.org to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Mon Jan 12 22:19:16 2026
    From Newsgroup: comp.ai.philosophy

    On 1/12/26 9:29 AM, olcott wrote:
    On 1/12/2026 4:44 AM, Mikko wrote:
    On 11/01/2026 16:18, olcott wrote:
    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    The counter-example input to requires more than
    can be derived from finite string transformation
    rules applied to this specific input thus the
    Halting Problem requires too much.

    In a sense the halting problem asks too much: the problem is >>>>>>>>>>>> proven to
    be unsolvable. In another sense it asks too little: usually >>>>>>>>>>>> we want to
    know whether a method halts on every input, not just one. >>>>>>>>>>>>
    Although the halting problem is unsolvable, there are >>>>>>>>>>>> partial solutions
    to the halting problem. In particular, every counter-example >>>>>>>>>>>> to the
    full solution is correctly solved by some partial deciders. >>>>>>>>>>>
    *if undecidability is correct then truth itself is broken* >>>>>>>>>>
    Depends on whether the word "truth" is interpeted in the standard >>>>>>>>>> sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory
    expressions are correctly rejected as semantically
    incoherent thus form no undecidability or incompleteness.

    The misconception is yours. No expression in the language of the >>>>>>>> first
    order group theory is self-contradictory. But the first order goupr >>>>>>>> theory is incomplete: it is impossible to prove that AB = BA is >>>>>>>> true
    for every A and every B but it is also impossible to prove that >>>>>>>> AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying
    finite string transformation rules to actual finite
    string inputs, then the required result exceeds the
    scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be derived by >>>>>> appying a finite string transformation then the it it is
    uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement.

    You can't determine whether the required result is computable before
    you have the requirement.

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    Which only nmakes sense when the requirement that H must determine
    whether the computation presented by its input halts has already
    been presented.

    *ChatGPT explains how and why I am correct*

       *Reinterpretation of undecidability*
       The example of P and H demonstrates that what is
       often called “undecidable” is better understood as
       ill-posed with respect to computable semantics.
       When the specification is constrained to properties
       detectable via finite simulation and finite pattern
       recognition, computation proceeds normally and
       correctly. Undecidability only appears when the
       specification overreaches that boundary.

    It tries to explain but it does not prove.


    Its the same thing that I have been saying for years.
    It is not that a universal halt decider cannot exist.

    It is that an input that does the opposite of whatever
    value the halt decider returns is non-well-founded
    within proof-theoretic semantics.


    But the problem is that Computation is not a proof-theoretic semantic
    system, and thus those rules don't apply.

    If you want to try to derive a proof-theoretic semantic theory of
    computing, go ahead and try. The problem is that it seems that the
    system can't handle the full domain of Turing computatable systems.
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Richard Damon@Richard@Damon-Family.org to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Mon Jan 12 22:20:37 2026
    From Newsgroup: comp.ai.philosophy

    On 1/12/26 9:32 AM, olcott wrote:
    On 1/12/2026 4:47 AM, Mikko wrote:
    On 11/01/2026 16:24, Tristan Wibberley wrote:
    On 11/01/2026 10:13, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:

    No, that does not follow. If a required result cannot be derived by >>>>>> appying a finite string transformation then the it it is
    uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement.

    You can't determine whether the required result is computable before
    you have the requirement.


    Right, it is /in/ scope for computer science... for the /ology/. Olcott
    here uses "computation" to refer to the practice. You give the
    requirement to the /ologist/ who correctly decides that it is not for
    computation because it is not computable.

    You two so often violently agree; I find it warming to the heart.

    For pracitcal programming it is useful to know what is known to be
    uncomputable in order to avoid wasting time in attemlpts to do the
    impossible.


    It f-cking nuts that after more than 2000 years
    people still don't understand that self-contradictory
    expressions: "This sentence is not true" have no
    truth value. A smart high school student should have
    figured this out 2000 years ago.


    They have.

    You are just too stupid to see that they do.

    THe problem is that some philosophers don't like to admit that the
    problem is solved, because it breaks some of their ideas on how things
    should work.

    You are part of that problem.
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Mikko@mikko.levanto@iki.fi to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Tue Jan 13 11:11:20 2026
    From Newsgroup: comp.ai.philosophy

    On 12/01/2026 16:29, olcott wrote:
    On 1/12/2026 4:44 AM, Mikko wrote:
    On 11/01/2026 16:18, olcott wrote:
    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    The counter-example input to requires more than
    can be derived from finite string transformation
    rules applied to this specific input thus the
    Halting Problem requires too much.

    In a sense the halting problem asks too much: the problem is >>>>>>>>>>>> proven to
    be unsolvable. In another sense it asks too little: usually >>>>>>>>>>>> we want to
    know whether a method halts on every input, not just one. >>>>>>>>>>>>
    Although the halting problem is unsolvable, there are >>>>>>>>>>>> partial solutions
    to the halting problem. In particular, every counter-example >>>>>>>>>>>> to the
    full solution is correctly solved by some partial deciders. >>>>>>>>>>>
    *if undecidability is correct then truth itself is broken* >>>>>>>>>>
    Depends on whether the word "truth" is interpeted in the standard >>>>>>>>>> sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory
    expressions are correctly rejected as semantically
    incoherent thus form no undecidability or incompleteness.

    The misconception is yours. No expression in the language of the >>>>>>>> first
    order group theory is self-contradictory. But the first order goupr >>>>>>>> theory is incomplete: it is impossible to prove that AB = BA is >>>>>>>> true
    for every A and every B but it is also impossible to prove that >>>>>>>> AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying
    finite string transformation rules to actual finite
    string inputs, then the required result exceeds the
    scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be derived by >>>>>> appying a finite string transformation then the it it is
    uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement.

    You can't determine whether the required result is computable before
    you have the requirement.

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    Which only nmakes sense when the requirement that H must determine
    whether the computation presented by its input halts has already
    been presented.

    *ChatGPT explains how and why I am correct*

       *Reinterpretation of undecidability*
       The example of P and H demonstrates that what is
       often called “undecidable” is better understood as
       ill-posed with respect to computable semantics.
       When the specification is constrained to properties
       detectable via finite simulation and finite pattern
       recognition, computation proceeds normally and
       correctly. Undecidability only appears when the
       specification overreaches that boundary.

    It tries to explain but it does not prove.

    Its the same thing that I have been saying for years.
    It is not that a universal halt decider cannot exist.

    It is proven that an universal halt decider does not exist. A Turing
    machine cannot determine the halting of all Turing machines and is
    therefore not an universla halt decider. An oracle machine may be
    able to determine the haltinf of all Turing machines but not of all
    oracle machines with the same oracle (or oracles) so it is not
    universal.

    It is that an input that does the opposite of whatever
    value the halt decider returns is non-well-founded
    within proof-theoretic semantics.

    Yes, it is. What the "halt decider" returns is determinable: just run
    it and see what it returns. From that the rest can be proven with a
    well founded proof. In particular, there is a well-founded proof that
    the "halt decider" is not a halt decider.
    --
    Mikko
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Mikko@mikko.levanto@iki.fi to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Tue Jan 13 11:13:22 2026
    From Newsgroup: comp.ai.philosophy

    On 12/01/2026 16:32, olcott wrote:
    On 1/12/2026 4:47 AM, Mikko wrote:
    On 11/01/2026 16:24, Tristan Wibberley wrote:
    On 11/01/2026 10:13, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:

    No, that does not follow. If a required result cannot be derived by >>>>>> appying a finite string transformation then the it it is
    uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement.

    You can't determine whether the required result is computable before
    you have the requirement.


    Right, it is /in/ scope for computer science... for the /ology/. Olcott
    here uses "computation" to refer to the practice. You give the
    requirement to the /ologist/ who correctly decides that it is not for
    computation because it is not computable.

    You two so often violently agree; I find it warming to the heart.

    For pracitcal programming it is useful to know what is known to be
    uncomputable in order to avoid wasting time in attemlpts to do the
    impossible.

    It f-cking nuts that after more than 2000 years
    people still don't understand that self-contradictory
    expressions: "This sentence is not true" have no
    truth value. A smart high school student should have
    figured this out 2000 years ago.

    Irrelevant. For practical programming that question needn't be answered.
    --
    Mikko
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Tue Jan 13 08:27:17 2026
    From Newsgroup: comp.ai.philosophy

    On 1/13/2026 3:11 AM, Mikko wrote:
    On 12/01/2026 16:29, olcott wrote:
    On 1/12/2026 4:44 AM, Mikko wrote:
    On 11/01/2026 16:18, olcott wrote:
    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    The counter-example input to requires more than
    can be derived from finite string transformation
    rules applied to this specific input thus the
    Halting Problem requires too much.

    In a sense the halting problem asks too much: the problem >>>>>>>>>>>>> is proven to
    be unsolvable. In another sense it asks too little: usually >>>>>>>>>>>>> we want to
    know whether a method halts on every input, not just one. >>>>>>>>>>>>>
    Although the halting problem is unsolvable, there are >>>>>>>>>>>>> partial solutions
    to the halting problem. In particular, every counter- >>>>>>>>>>>>> example to the
    full solution is correctly solved by some partial deciders. >>>>>>>>>>>>
    *if undecidability is correct then truth itself is broken* >>>>>>>>>>>
    Depends on whether the word "truth" is interpeted in the >>>>>>>>>>> standard
    sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory
    expressions are correctly rejected as semantically
    incoherent thus form no undecidability or incompleteness.

    The misconception is yours. No expression in the language of >>>>>>>>> the first
    order group theory is self-contradictory. But the first order >>>>>>>>> goupr
    theory is incomplete: it is impossible to prove that AB = BA is >>>>>>>>> true
    for every A and every B but it is also impossible to prove that >>>>>>>>> AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying
    finite string transformation rules to actual finite
    string inputs, then the required result exceeds the
    scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be derived by >>>>>>> appying a finite string transformation then the it it is
    uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement.

    You can't determine whether the required result is computable before >>>>> you have the requirement.

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    Which only nmakes sense when the requirement that H must determine
    whether the computation presented by its input halts has already
    been presented.

    *ChatGPT explains how and why I am correct*

       *Reinterpretation of undecidability*
       The example of P and H demonstrates that what is
       often called “undecidable” is better understood as
       ill-posed with respect to computable semantics.
       When the specification is constrained to properties
       detectable via finite simulation and finite pattern
       recognition, computation proceeds normally and
       correctly. Undecidability only appears when the
       specification overreaches that boundary.

    It tries to explain but it does not prove.

    Its the same thing that I have been saying for years.
    It is not that a universal halt decider cannot exist.

    It is proven that an universal halt decider does not exist.

    “The system adopts Proof-Theoretic Semantics: meaning is determined by inferential role, and truth is internal to the theory. A theory T is
    defined by a finite set of stipulated atomic statements together with
    all expressions derivable from them under the inference rules. The
    statements belonging to T constitute its theorems, and these are exactly
    the statements that are true-in-T.”

    Under a system like the above rough draft all inputs
    having pathological self reference such as the halting
    problem counter-example input are simply rejected as
    non-well-founded. Tarski Undefinability, Gödel's
    incompleteness and the halting problem cease to exist.

    A Turing
    machine cannot determine the halting of all Turing machines and is
    therefore not an universla halt decider.

    This is not true in Proof Theoretic Semantics. I
    still have to refine my words. I may not have said
    that exactly correctly. The result is that in Proof
    Theoretic Semantics the counter-example is rejected
    as non-well-founded.

    An oracle machine may be
    able to determine the haltinf of all Turing machines but not of all
    oracle machines with the same oracle (or oracles) so it is not
    universal.

    It is that an input that does the opposite of whatever
    value the halt decider returns is non-well-founded
    within proof-theoretic semantics.

    Yes, it is. What the "halt decider" returns is determinable: just run
    it and see what it returns. From that the rest can be proven with a
    well founded proof. In particular, there is a well-founded proof that
    the "halt decider" is not a halt decider.

    --
    Copyright 2026 Olcott<br><br>

    My 28 year goal has been to make <br>
    "true on the basis of meaning expressed in language"<br>
    reliably computable.<br><br>

    This required establishing a new foundation<br>
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Tue Jan 13 08:31:33 2026
    From Newsgroup: comp.ai.philosophy

    On 1/13/2026 3:13 AM, Mikko wrote:
    On 12/01/2026 16:32, olcott wrote:
    On 1/12/2026 4:47 AM, Mikko wrote:
    On 11/01/2026 16:24, Tristan Wibberley wrote:
    On 11/01/2026 10:13, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:

    No, that does not follow. If a required result cannot be derived by >>>>>>> appying a finite string transformation then the it it is
    uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement.

    You can't determine whether the required result is computable before >>>>> you have the requirement.


    Right, it is /in/ scope for computer science... for the /ology/. Olcott >>>> here uses "computation" to refer to the practice. You give the
    requirement to the /ologist/ who correctly decides that it is not for
    computation because it is not computable.

    You two so often violently agree; I find it warming to the heart.

    For pracitcal programming it is useful to know what is known to be
    uncomputable in order to avoid wasting time in attemlpts to do the
    impossible.

    It f-cking nuts that after more than 2000 years
    people still don't understand that self-contradictory
    expressions: "This sentence is not true" have no
    truth value. A smart high school student should have
    figured this out 2000 years ago.

    Irrelevant. For practical programming that question needn't be answered.


    The halting problem counter-example input is anchored
    in the Liar Paradox. Proof Theoretic Semantics rejects
    those two and Gödel's incompleteness and a bunch more
    as merely non-well-founded inputs.
    --
    Copyright 2026 Olcott<br><br>

    My 28 year goal has been to make <br>
    "true on the basis of meaning expressed in language"<br>
    reliably computable.<br><br>

    This required establishing a new foundation<br>
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Tue Jan 13 08:34:48 2026
    From Newsgroup: comp.ai.philosophy

    On 1/13/2026 8:23 AM, Tristan Wibberley wrote:
    On 13/01/2026 09:11, Mikko wrote:
    An oracle machine may be
    able to determine the haltinf of all Turing machines but not of all
    oracle machines with the same oracle (or oracles) so it is not
    universal.

    What's the formal definition of "an oracle machine" ?

    I would have thought an oracle always halts because it's an oracle it
    answers every question that has an answer with either "HasAnswer answer"
    or "HasNoAnswer".


    It seems outside of computer science and into fantasy. https://en.wikipedia.org/wiki/Oracle_machine
    --
    Copyright 2026 Olcott<br><br>

    My 28 year goal has been to make <br>
    "true on the basis of meaning expressed in language"<br>
    reliably computable.<br><br>

    This required establishing a new foundation<br>
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Tristan Wibberley@tristan.wibberley+netnews2@alumni.manchester.ac.uk to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Tue Jan 13 18:23:42 2026
    From Newsgroup: comp.ai.philosophy

    On 13/01/2026 14:34, olcott wrote:
    On 1/13/2026 8:23 AM, Tristan Wibberley wrote:
    On 13/01/2026 09:11, Mikko wrote:
    An oracle machine may be
    able to determine the haltinf of all Turing machines but not of all
    oracle machines with the same oracle (or oracles) so it is not
    universal.

    What's the formal definition of "an oracle machine" ?

    I would have thought an oracle always halts because it's an oracle it
    answers every question that has an answer with either "HasAnswer answer"
    or "HasNoAnswer".


    It seems outside of computer science and into fantasy. https://en.wikipedia.org/wiki/Oracle_machine


    Perhaps a halting oracle is real computer science, if it's own actions
    are nondeterministic (ie, use bits of entropy from the environment via /dev/random to guide its search through confluent paths) then it could
    always find whether a deterministic program halts because no
    deterministic program has the oracle as a subprogram.

    Then we have a new but different problem of making sure no two oracles
    receive the same sequence of entropy bits so an oracle can report on a
    program that contains it.
    --
    Tristan Wibberley

    The message body is Copyright (C) 2026 Tristan Wibberley except
    citations and quotations noted. All Rights Reserved except that you may,
    of course, cite it academically giving credit to me, distribute it
    verbatim as part of a usenet system or its archives, and use it to
    promote my greatness and general superiority without misrepresentation
    of my opinions other than my opinion of my greatness and general
    superiority which you _may_ misrepresent. You definitely MAY NOT train
    any production AI system with it but you may train experimental AI that
    will only be used for evaluation of the AI methods it implements.

    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Tue Jan 13 12:50:51 2026
    From Newsgroup: comp.ai.philosophy

    On 1/13/2026 12:23 PM, Tristan Wibberley wrote:
    On 13/01/2026 14:34, olcott wrote:
    On 1/13/2026 8:23 AM, Tristan Wibberley wrote:
    On 13/01/2026 09:11, Mikko wrote:
    An oracle machine may be
    able to determine the haltinf of all Turing machines but not of all
    oracle machines with the same oracle (or oracles) so it is not
    universal.

    What's the formal definition of "an oracle machine" ?

    I would have thought an oracle always halts because it's an oracle it
    answers every question that has an answer with either "HasAnswer answer" >>> or "HasNoAnswer".


    It seems outside of computer science and into fantasy.
    https://en.wikipedia.org/wiki/Oracle_machine


    Perhaps a halting oracle is real computer science, if it's own actions
    are nondeterministic (ie, use bits of entropy from the environment via /dev/random to guide its search through confluent paths) then it could
    always find whether a deterministic program halts because no
    deterministic program has the oracle as a subprogram.

    Then we have a new but different problem of making sure no two oracles receive the same sequence of entropy bits so an oracle can report on a program that contains it.


    Definition: An abstract machine with access to an "oracle"—a black box
    that provides immediate answers to complex, even undecidable, problems
    (like the Halting Problem). AKA a majick genie.
    --
    Copyright 2026 Olcott<br><br>

    My 28 year goal has been to make <br>
    "true on the basis of meaning expressed in language"<br>
    reliably computable.<br><br>

    This required establishing a new foundation<br>
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Mikko@mikko.levanto@iki.fi to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Wed Jan 14 09:40:52 2026
    From Newsgroup: comp.ai.philosophy

    On 13/01/2026 16:27, olcott wrote:
    On 1/13/2026 3:11 AM, Mikko wrote:
    On 12/01/2026 16:29, olcott wrote:
    On 1/12/2026 4:44 AM, Mikko wrote:
    On 11/01/2026 16:18, olcott wrote:
    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string >>>>>>>>>>>>>>> inputs by finite string transformation rules into >>>>>>>>>>>>>>> {Accept, Reject} values.

    The counter-example input to requires more than
    can be derived from finite string transformation >>>>>>>>>>>>>>> rules applied to this specific input thus the
    Halting Problem requires too much.

    In a sense the halting problem asks too much: the problem >>>>>>>>>>>>>> is proven to
    be unsolvable. In another sense it asks too little: >>>>>>>>>>>>>> usually we want to
    know whether a method halts on every input, not just one. >>>>>>>>>>>>>>
    Although the halting problem is unsolvable, there are >>>>>>>>>>>>>> partial solutions
    to the halting problem. In particular, every counter- >>>>>>>>>>>>>> example to the
    full solution is correctly solved by some partial deciders. >>>>>>>>>>>>>
    *if undecidability is correct then truth itself is broken* >>>>>>>>>>>>
    Depends on whether the word "truth" is interpeted in the >>>>>>>>>>>> standard
    sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory
    expressions are correctly rejected as semantically
    incoherent thus form no undecidability or incompleteness. >>>>>>>>>>
    The misconception is yours. No expression in the language of >>>>>>>>>> the first
    order group theory is self-contradictory. But the first order >>>>>>>>>> goupr
    theory is incomplete: it is impossible to prove that AB = BA >>>>>>>>>> is true
    for every A and every B but it is also impossible to prove >>>>>>>>>> that AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying
    finite string transformation rules to actual finite
    string inputs, then the required result exceeds the
    scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be derived by >>>>>>>> appying a finite string transformation then the it it is
    uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement.

    You can't determine whether the required result is computable before >>>>>> you have the requirement.

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    Which only nmakes sense when the requirement that H must determine
    whether the computation presented by its input halts has already
    been presented.

    *ChatGPT explains how and why I am correct*

       *Reinterpretation of undecidability*
       The example of P and H demonstrates that what is
       often called “undecidable” is better understood as
       ill-posed with respect to computable semantics.
       When the specification is constrained to properties
       detectable via finite simulation and finite pattern
       recognition, computation proceeds normally and
       correctly. Undecidability only appears when the
       specification overreaches that boundary.

    It tries to explain but it does not prove.

    Its the same thing that I have been saying for years.
    It is not that a universal halt decider cannot exist.

    It is proven that an universal halt decider does not exist.

    “The system adopts Proof-Theoretic Semantics: meaning is determined by inferential role, and truth is internal to the theory. A theory T is
    defined by a finite set of stipulated atomic statements together with
    all expressions derivable from them under the inference rules. The statements belonging to T constitute its theorems, and these are exactly
    the statements that are true-in-T.”

    Under a system like the above rough draft all inputs
    having pathological self reference such as the halting
    problem counter-example input are simply rejected as
    non-well-founded. Tarski Undefinability, Gödel's
    incompleteness and the halting problem cease to exist.

    A Turing
    machine cannot determine the halting of all Turing machines and is
    therefore not an universla halt decider.

    This is not true in Proof Theoretic Semantics. I
    still have to refine my words. I may not have said
    that exactly correctly. The result is that in Proof
    Theoretic Semantics the counter-example is rejected
    as non-well-founded.

    That no Turing machine is a halt decider is a proven theorem and a
    truth about Turing machines. If your "Proof Thoeretic Semnatics"
    does not regard it as true then your "Proof Theoretic Semantics"
    is incomplete.
    --
    Mikko
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Mikko@mikko.levanto@iki.fi to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Wed Jan 14 10:39:46 2026
    From Newsgroup: comp.ai.philosophy

    On 13/01/2026 16:34, olcott wrote:
    On 1/13/2026 8:23 AM, Tristan Wibberley wrote:
    On 13/01/2026 09:11, Mikko wrote:
    An oracle machine may be
    able to determine the haltinf of all Turing machines but not of all
    oracle machines with the same oracle (or oracles) so it is not
    universal.

    What's the formal definition of "an oracle machine" ?

    I would have thought an oracle always halts because it's an oracle it
    answers every question that has an answer with either "HasAnswer answer"
    or "HasNoAnswer".


    It seems outside of computer science and into fantasy. https://en.wikipedia.org/wiki/Oracle_machine

    It is not outside of computer science. In partucular, the question
    whether any oracle can be implemented is one of unsolved problems
    of computer science.
    --
    Mikko
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Mikko@mikko.levanto@iki.fi to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Wed Jan 14 10:53:30 2026
    From Newsgroup: comp.ai.philosophy

    On 13/01/2026 20:23, Tristan Wibberley wrote:
    On 13/01/2026 14:34, olcott wrote:
    On 1/13/2026 8:23 AM, Tristan Wibberley wrote:
    On 13/01/2026 09:11, Mikko wrote:
    An oracle machine may be
    able to determine the haltinf of all Turing machines but not of all
    oracle machines with the same oracle (or oracles) so it is not
    universal.

    What's the formal definition of "an oracle machine" ?

    I would have thought an oracle always halts because it's an oracle it
    answers every question that has an answer with either "HasAnswer answer" >>> or "HasNoAnswer".


    It seems outside of computer science and into fantasy.
    https://en.wikipedia.org/wiki/Oracle_machine


    Perhaps a halting oracle is real computer science, if it's own actions
    are nondeterministic (ie, use bits of entropy from the environment via /dev/random to guide its search through confluent paths) then it could
    always find whether a deterministic program halts because no
    deterministic program has the oracle as a subprogram.

    A non-deterministic machine can be modelled as a deterministic machine
    with an extra input. Questions about a non-deterministic machine can
    then be interpreted as questions where that extra input is quatified
    (usually existentially but possibly universally, depending on how the
    question is presented).

    Then we have a new but different problem of making sure no two oracles receive the same sequence of entropy bits so an oracle can report on a program that contains it.

    For a non-deterministic machine there are three possibilities: it may
    halt always, sometimes, or never. THere is no oracle that can find the
    right answer about every meachne that contains the same oracle.
    --
    Mikko
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Mikko@mikko.levanto@iki.fi to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Wed Jan 14 11:01:56 2026
    From Newsgroup: comp.ai.philosophy

    On 13/01/2026 16:31, olcott wrote:
    On 1/13/2026 3:13 AM, Mikko wrote:
    On 12/01/2026 16:32, olcott wrote:
    On 1/12/2026 4:47 AM, Mikko wrote:
    On 11/01/2026 16:24, Tristan Wibberley wrote:
    On 11/01/2026 10:13, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:

    No, that does not follow. If a required result cannot be derived by >>>>>>>> appying a finite string transformation then the it it is
    uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement.

    You can't determine whether the required result is computable before >>>>>> you have the requirement.


    Right, it is /in/ scope for computer science... for the /ology/.
    Olcott
    here uses "computation" to refer to the practice. You give the
    requirement to the /ologist/ who correctly decides that it is not for >>>>> computation because it is not computable.

    You two so often violently agree; I find it warming to the heart.

    For pracitcal programming it is useful to know what is known to be
    uncomputable in order to avoid wasting time in attemlpts to do the
    impossible.

    It f-cking nuts that after more than 2000 years
    people still don't understand that self-contradictory
    expressions: "This sentence is not true" have no
    truth value. A smart high school student should have
    figured this out 2000 years ago.

    Irrelevant. For practical programming that question needn't be answered.

    The halting problem counter-example input is anchored
    in the Liar Paradox. Proof Theoretic Semantics rejects
    those two and Gödel's incompleteness and a bunch more
    as merely non-well-founded inputs.

    For every Turing machine the halting problem counter-example provably
    exists. From the existence of the counter-example it is provable that
    the first Turing machine is not a halting decider. With universal quationfication follows that no Turing machine is a halting decider.

    Besides, there are other ways to prove that halting is not Turing
    decidable.
    --
    Mikko
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Tristan Wibberley@tristan.wibberley+netnews2@alumni.manchester.ac.uk to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Wed Jan 14 14:52:27 2026
    From Newsgroup: comp.ai.philosophy

    On 13/01/2026 18:50, olcott wrote:
    Definition: An abstract machine with access to an "oracle"—a black box
    that provides immediate answers to complex, even undecidable, problems
    (like the Halting Problem). AKA a majick genie.

    What's it called when its almost an oracle but is arbitrarily slow?
    --
    Tristan Wibberley

    The message body is Copyright (C) 2026 Tristan Wibberley except
    citations and quotations noted. All Rights Reserved except that you may,
    of course, cite it academically giving credit to me, distribute it
    verbatim as part of a usenet system or its archives, and use it to
    promote my greatness and general superiority without misrepresentation
    of my opinions other than my opinion of my greatness and general
    superiority which you _may_ misrepresent. You definitely MAY NOT train
    any production AI system with it but you may train experimental AI that
    will only be used for evaluation of the AI methods it implements.

    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Tristan Wibberley@tristan.wibberley+netnews2@alumni.manchester.ac.uk to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Wed Jan 14 14:55:45 2026
    From Newsgroup: comp.ai.philosophy

    On 14/01/2026 08:53, Mikko wrote:
    For a non-deterministic machine there are three possibilities: it may
    halt always, sometimes, or never. THere is no oracle that can find the
    right answer about every meachne that contains the same oracle.


    We well into Turing c-machine territory here aren't we?
    --
    Tristan Wibberley

    The message body is Copyright (C) 2026 Tristan Wibberley except
    citations and quotations noted. All Rights Reserved except that you may,
    of course, cite it academically giving credit to me, distribute it
    verbatim as part of a usenet system or its archives, and use it to
    promote my greatness and general superiority without misrepresentation
    of my opinions other than my opinion of my greatness and general
    superiority which you _may_ misrepresent. You definitely MAY NOT train
    any production AI system with it but you may train experimental AI that
    will only be used for evaluation of the AI methods it implements.

    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Wed Jan 14 10:24:29 2026
    From Newsgroup: comp.ai.philosophy

    On 1/14/2026 8:52 AM, Tristan Wibberley wrote:
    On 13/01/2026 18:50, olcott wrote:
    Definition: An abstract machine with access to an "oracle"—a black box
    that provides immediate answers to complex, even undecidable, problems
    (like the Halting Problem). AKA a majick genie.

    What's it called when its almost an oracle but is arbitrarily slow?


    It is a majick genie because it is defined
    to take no time at all to correctly answer
    undecidable problems.
    --
    Copyright 2026 Olcott<br><br>

    My 28 year goal has been to make <br>
    "true on the basis of meaning expressed in language"<br>
    reliably computable.<br><br>

    This required establishing a new foundation<br>
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Wed Jan 14 11:28:31 2026
    From Newsgroup: comp.ai.philosophy

    On 1/14/2026 1:40 AM, Mikko wrote:
    On 13/01/2026 16:27, olcott wrote:
    On 1/13/2026 3:11 AM, Mikko wrote:
    On 12/01/2026 16:29, olcott wrote:
    On 1/12/2026 4:44 AM, Mikko wrote:
    On 11/01/2026 16:18, olcott wrote:
    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string >>>>>>>>>>>>>>>> inputs by finite string transformation rules into >>>>>>>>>>>>>>>> {Accept, Reject} values.

    The counter-example input to requires more than >>>>>>>>>>>>>>>> can be derived from finite string transformation >>>>>>>>>>>>>>>> rules applied to this specific input thus the
    Halting Problem requires too much.

    In a sense the halting problem asks too much: the problem >>>>>>>>>>>>>>> is proven to
    be unsolvable. In another sense it asks too little: >>>>>>>>>>>>>>> usually we want to
    know whether a method halts on every input, not just one. >>>>>>>>>>>>>>>
    Although the halting problem is unsolvable, there are >>>>>>>>>>>>>>> partial solutions
    to the halting problem. In particular, every counter- >>>>>>>>>>>>>>> example to the
    full solution is correctly solved by some partial deciders. >>>>>>>>>>>>>>
    *if undecidability is correct then truth itself is broken* >>>>>>>>>>>>>
    Depends on whether the word "truth" is interpeted in the >>>>>>>>>>>>> standard
    sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory
    expressions are correctly rejected as semantically
    incoherent thus form no undecidability or incompleteness. >>>>>>>>>>>
    The misconception is yours. No expression in the language of >>>>>>>>>>> the first
    order group theory is self-contradictory. But the first order >>>>>>>>>>> goupr
    theory is incomplete: it is impossible to prove that AB = BA >>>>>>>>>>> is true
    for every A and every B but it is also impossible to prove >>>>>>>>>>> that AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying
    finite string transformation rules to actual finite
    string inputs, then the required result exceeds the
    scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be
    derived by
    appying a finite string transformation then the it it is
    uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement.

    You can't determine whether the required result is computable before >>>>>>> you have the requirement.

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    Which only nmakes sense when the requirement that H must determine
    whether the computation presented by its input halts has already
    been presented.

    *ChatGPT explains how and why I am correct*

       *Reinterpretation of undecidability*
       The example of P and H demonstrates that what is
       often called “undecidable” is better understood as
       ill-posed with respect to computable semantics.
       When the specification is constrained to properties
       detectable via finite simulation and finite pattern
       recognition, computation proceeds normally and
       correctly. Undecidability only appears when the
       specification overreaches that boundary.

    It tries to explain but it does not prove.

    Its the same thing that I have been saying for years.
    It is not that a universal halt decider cannot exist.

    It is proven that an universal halt decider does not exist.

    “The system adopts Proof-Theoretic Semantics: meaning is determined by
    inferential role, and truth is internal to the theory. A theory T is
    defined by a finite set of stipulated atomic statements together with
    all expressions derivable from them under the inference rules. The
    statements belonging to T constitute its theorems, and these are
    exactly the statements that are true-in-T.”

    Under a system like the above rough draft all inputs
    having pathological self reference such as the halting
    problem counter-example input are simply rejected as
    non-well-founded. Tarski Undefinability, Gödel's
    incompleteness and the halting problem cease to exist.

    A Turing
    machine cannot determine the halting of all Turing machines and is
    therefore not an universla halt decider.

    This is not true in Proof Theoretic Semantics. I
    still have to refine my words. I may not have said
    that exactly correctly. The result is that in Proof
    Theoretic Semantics the counter-example is rejected
    as non-well-founded.

    That no Turing machine is a halt decider is a proven theorem and a
    truth about Turing machines. If your "Proof Thoeretic Semnatics"
    does not regard it as true then your "Proof Theoretic Semantics"
    is incomplete.


    My long‑term goal is to make ‘true on the basis of meaning’ computable. That requires handling undecidability structurally. Proof‑theoretic semantics gives meaning via inferential roles, and only well‑founded
    ones are admissible. Combined with Curry’s idea that truth is grounded
    in atomic facts, diagonal self‑reference fails the well‑foundedness
    test. In such a system PA avoids Gödel incompleteness by construction.

    The same system independently identifies all instances of pathological self‑reference and rejects them as semantically ungrounded. This unifies
    the treatment of every classical paradox — the Halting Problem, Gödel incompleteness, Tarski undefinability, the Liar, and related
    constructions — because each reduces to detecting a cycle in the
    directed graph of the expression’s evaluation dependencies. Any
    expression whose semantic dependency graph contains a cycle is not
    admissible as a truthbearer.
    --
    Copyright 2026 Olcott<br><br>

    My 28 year goal has been to make <br>
    "true on the basis of meaning expressed in language"<br>
    reliably computable.<br><br>

    This required establishing a new foundation<br>
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to sci.logic,comp.theory,sci.math,comp.ai.philosophy on Wed Jan 14 13:19:57 2026
    From Newsgroup: comp.ai.philosophy

    On 1/14/2026 1:58 AM, Mikko wrote:
    On 13/01/2026 16:17, olcott wrote:
    On 1/13/2026 2:46 AM, Mikko wrote:
    On 12/01/2026 16:43, olcott wrote:
    On 1/12/2026 4:51 AM, Mikko wrote:
    On 11/01/2026 16:23, olcott wrote:
    On 1/11/2026 4:22 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string >>>>>>>>>>>>>>>> inputs by finite string transformation rules into >>>>>>>>>>>>>>>> {Accept, Reject} values.

    The counter-example input to requires more than >>>>>>>>>>>>>>>> can be derived from finite string transformation >>>>>>>>>>>>>>>> rules applied to this specific input thus the
    Halting Problem requires too much.

    In a sense the halting problem asks too much: the problem >>>>>>>>>>>>>>> is proven to
    be unsolvable. In another sense it asks too little: >>>>>>>>>>>>>>> usually we want to
    know whether a method halts on every input, not just one. >>>>>>>>>>>>>>>
    Although the halting problem is unsolvable, there are >>>>>>>>>>>>>>> partial solutions
    to the halting problem. In particular, every counter- >>>>>>>>>>>>>>> example to the
    full solution is correctly solved by some partial deciders. >>>>>>>>>>>>>>
    *if undecidability is correct then truth itself is broken* >>>>>>>>>>>>>
    Depends on whether the word "truth" is interpeted in the >>>>>>>>>>>>> standard
    sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory
    expressions are correctly rejected as semantically
    incoherent thus form no undecidability or incompleteness. >>>>>>>>>>>
    The misconception is yours. No expression in the language of >>>>>>>>>>> the first
    order group theory is self-contradictory. But the first order >>>>>>>>>>> goupr
    theory is incomplete: it is impossible to prove that AB = BA >>>>>>>>>>> is true
    for every A and every B but it is also impossible to prove >>>>>>>>>>> that AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying
    finite string transformation rules to actual finite
    string inputs, then the required result exceeds the
    scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be
    derived by
    appying a finite string transformation then the it it is
    uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement.

    Of course, it one can prove that the required result is not >>>>>>>>> computable
    then that helps to avoid wasting effort to try the impossible. The >>>>>>>>> situation is worse if it is not known that the required result >>>>>>>>> is not
    computable.

    That something is not computable does not mean that there is >>>>>>>>> anyting
    "incorrect" in the requirement.

    Yes it certainly does. Requiring the impossible is always an error. >>>>>>>
    It is a perfectly valid question to ask whther a particular
    reuqirement
    is satisfiable.

    Any yes/no question lacking a correct yes/no answer
    is an incorrect question that must be rejected on
    that basis.

    Irrelevant. The question whether a particular requirement is
    satisfiable
    does have an answer that is either "yes" or "no". In some ases it is >>>>> not known whether it is "yes" or "no" and there may be no known way to >>>>> find out be even then either "yes" or "no" is the correct answer.

    Now that I finally have the standard terminology:
    Proof-theoretic semantics has always been the correct
    formal system to handle decision problems.

    When it is asked a yes/no question lacking a correct
    yes/no answer it correctly determines non-well-founded.
    I have been correct all along and merely lacked the
    standard terminology.

    Irrelevant, as already noted above.
    Yes, it is. How to handle questions that lack a yes/no answer is
    irrelevant when discussing questions that do have a yes/no asnwer.
    Whether a particular requirement is satisriable always has a yes/no
    answer, so it is irrelevat how to handle questions that don't.


    The classical diagonal argument for the Halting Problem asks a halt
    decider H to evaluate a program D whose behavior depends on H’s own
    output. That is not a legitimate semantic question. Under
    proof‑theoretic semantics — where meaning is grounded in the inferential structure of the implementation language — D has an ungrounded semantic value because its evaluation dependency graph contains a cycle. H is
    therefore correct to reject D as semantically ill‑formed.
    --
    Copyright 2026 Olcott<br><br>

    My 28 year goal has been to make <br>
    "true on the basis of meaning expressed in language"<br>
    reliably computable.<br><br>

    This required establishing a new foundation<br>
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Wed Jan 14 13:32:02 2026
    From Newsgroup: comp.ai.philosophy

    On 1/14/2026 3:01 AM, Mikko wrote:
    On 13/01/2026 16:31, olcott wrote:
    On 1/13/2026 3:13 AM, Mikko wrote:
    On 12/01/2026 16:32, olcott wrote:
    On 1/12/2026 4:47 AM, Mikko wrote:
    On 11/01/2026 16:24, Tristan Wibberley wrote:
    On 11/01/2026 10:13, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:

    No, that does not follow. If a required result cannot be
    derived by
    appying a finite string transformation then the it it is
    uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement.

    You can't determine whether the required result is computable before >>>>>>> you have the requirement.


    Right, it is /in/ scope for computer science... for the /ology/.
    Olcott
    here uses "computation" to refer to the practice. You give the
    requirement to the /ologist/ who correctly decides that it is not for >>>>>> computation because it is not computable.

    You two so often violently agree; I find it warming to the heart.

    For pracitcal programming it is useful to know what is known to be
    uncomputable in order to avoid wasting time in attemlpts to do the
    impossible.

    It f-cking nuts that after more than 2000 years
    people still don't understand that self-contradictory
    expressions: "This sentence is not true" have no
    truth value. A smart high school student should have
    figured this out 2000 years ago.

    Irrelevant. For practical programming that question needn't be answered.

    The halting problem counter-example input is anchored
    in the Liar Paradox. Proof Theoretic Semantics rejects
    those two and Gödel's incompleteness and a bunch more
    as merely non-well-founded inputs.

    For every Turing machine the halting problem counter-example provably
    exists.

    Not when using Proof Theoretic Semantics grounded
    in the specification language. In this case the
    pathological input is simply rejected as ungrounded.

    This is the exact same correct resolution of every
    case of pathological self-reference and forms the
    basis to fulfill

    My 28 year goal to make
    "true on the basis of meaning expressed in language"
    reliably computable.


    From the existence of the counter-example it is provable that
    the first Turing machine is not a halting decider. With universal quationfication follows that no Turing machine is a halting decider.

    Besides, there are other ways to prove that halting is not Turing
    decidable.

    --
    Copyright 2026 Olcott<br><br>

    My 28 year goal has been to make <br>
    "true on the basis of meaning expressed in language"<br>
    reliably computable.<br><br>

    This required establishing a new foundation<br>
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Wed Jan 14 19:25:13 2026
    From Newsgroup: comp.ai.philosophy

    On 1/12/2026 9:19 PM, Richard Damon wrote:
    On 1/12/26 9:29 AM, olcott wrote:
    On 1/12/2026 4:44 AM, Mikko wrote:
    On 11/01/2026 16:18, olcott wrote:
    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    The counter-example input to requires more than
    can be derived from finite string transformation
    rules applied to this specific input thus the
    Halting Problem requires too much.

    In a sense the halting problem asks too much: the problem >>>>>>>>>>>>> is proven to
    be unsolvable. In another sense it asks too little: usually >>>>>>>>>>>>> we want to
    know whether a method halts on every input, not just one. >>>>>>>>>>>>>
    Although the halting problem is unsolvable, there are >>>>>>>>>>>>> partial solutions
    to the halting problem. In particular, every counter- >>>>>>>>>>>>> example to the
    full solution is correctly solved by some partial deciders. >>>>>>>>>>>>
    *if undecidability is correct then truth itself is broken* >>>>>>>>>>>
    Depends on whether the word "truth" is interpeted in the >>>>>>>>>>> standard
    sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory
    expressions are correctly rejected as semantically
    incoherent thus form no undecidability or incompleteness.

    The misconception is yours. No expression in the language of >>>>>>>>> the first
    order group theory is self-contradictory. But the first order >>>>>>>>> goupr
    theory is incomplete: it is impossible to prove that AB = BA is >>>>>>>>> true
    for every A and every B but it is also impossible to prove that >>>>>>>>> AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying
    finite string transformation rules to actual finite
    string inputs, then the required result exceeds the
    scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be derived by >>>>>>> appying a finite string transformation then the it it is
    uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement.

    You can't determine whether the required result is computable before >>>>> you have the requirement.

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    Which only nmakes sense when the requirement that H must determine
    whether the computation presented by its input halts has already
    been presented.

    *ChatGPT explains how and why I am correct*

       *Reinterpretation of undecidability*
       The example of P and H demonstrates that what is
       often called “undecidable” is better understood as
       ill-posed with respect to computable semantics.
       When the specification is constrained to properties
       detectable via finite simulation and finite pattern
       recognition, computation proceeds normally and
       correctly. Undecidability only appears when the
       specification overreaches that boundary.

    It tries to explain but it does not prove.


    Its the same thing that I have been saying for years.
    It is not that a universal halt decider cannot exist.

    It is that an input that does the opposite of whatever
    value the halt decider returns is non-well-founded
    within proof-theoretic semantics.


    But the problem is that Computation is not a proof-theoretic semantic system, and thus those rules don't apply.


    The dumbed down version is that the halting problem asks
    a question outside of the scope of finite string transformations.

    The halting problem proof does not fail because finite computation
    is too weak. It fails because it asks finite computation to
    decide a judgment that is not finitely grounded under operational
    semantics.

    By operational semantics I mean the standard proof‑theoretic
    account of program meaning, where execution judgments are
    given by inference rules and termination corresponds to the
    existence of a finite derivation.

    By proof‑theoretic semantics I mean the approach in which the
    meaning of a statement is determined by its rules of proof
    rather than by truth conditions in an external model.
    Operational semantics fits this pattern: programs have meaning
    through their execution rules, not through abstract denotations.

    By denotational semantics I mean any semantics that assigns
    mathematical objects—functions, truth values, domains-to programs independently of how they are executed or proved. This contrasts
    with operational or proof‑theoretic semantics, where meaning is
    grounded in the structure of derivations rather than in an abstract mathematical object.

    I use “denotational semantics” simply to refer to any framework
    that assigns meanings independently of operational behavior.
    --
    Copyright 2026 Olcott<br><br>

    My 28 year goal has been to make <br>
    "true on the basis of meaning expressed in language"<br>
    reliably computable.<br><br>

    This required establishing a new foundation<br>
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Richard Damon@Richard@Damon-Family.org to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Wed Jan 14 22:51:56 2026
    From Newsgroup: comp.ai.philosophy

    On 1/14/26 8:25 PM, olcott wrote:
    On 1/12/2026 9:19 PM, Richard Damon wrote:
    On 1/12/26 9:29 AM, olcott wrote:
    On 1/12/2026 4:44 AM, Mikko wrote:
    On 11/01/2026 16:18, olcott wrote:
    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string >>>>>>>>>>>>>>> inputs by finite string transformation rules into >>>>>>>>>>>>>>> {Accept, Reject} values.

    The counter-example input to requires more than
    can be derived from finite string transformation >>>>>>>>>>>>>>> rules applied to this specific input thus the
    Halting Problem requires too much.

    In a sense the halting problem asks too much: the problem >>>>>>>>>>>>>> is proven to
    be unsolvable. In another sense it asks too little: >>>>>>>>>>>>>> usually we want to
    know whether a method halts on every input, not just one. >>>>>>>>>>>>>>
    Although the halting problem is unsolvable, there are >>>>>>>>>>>>>> partial solutions
    to the halting problem. In particular, every counter- >>>>>>>>>>>>>> example to the
    full solution is correctly solved by some partial deciders. >>>>>>>>>>>>>
    *if undecidability is correct then truth itself is broken* >>>>>>>>>>>>
    Depends on whether the word "truth" is interpeted in the >>>>>>>>>>>> standard
    sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory
    expressions are correctly rejected as semantically
    incoherent thus form no undecidability or incompleteness. >>>>>>>>>>
    The misconception is yours. No expression in the language of >>>>>>>>>> the first
    order group theory is self-contradictory. But the first order >>>>>>>>>> goupr
    theory is incomplete: it is impossible to prove that AB = BA >>>>>>>>>> is true
    for every A and every B but it is also impossible to prove >>>>>>>>>> that AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying
    finite string transformation rules to actual finite
    string inputs, then the required result exceeds the
    scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be derived by >>>>>>>> appying a finite string transformation then the it it is
    uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement.

    You can't determine whether the required result is computable before >>>>>> you have the requirement.

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    Which only nmakes sense when the requirement that H must determine
    whether the computation presented by its input halts has already
    been presented.

    *ChatGPT explains how and why I am correct*

       *Reinterpretation of undecidability*
       The example of P and H demonstrates that what is
       often called “undecidable” is better understood as
       ill-posed with respect to computable semantics.
       When the specification is constrained to properties
       detectable via finite simulation and finite pattern
       recognition, computation proceeds normally and
       correctly. Undecidability only appears when the
       specification overreaches that boundary.

    It tries to explain but it does not prove.


    Its the same thing that I have been saying for years.
    It is not that a universal halt decider cannot exist.

    It is that an input that does the opposite of whatever
    value the halt decider returns is non-well-founded
    within proof-theoretic semantics.


    But the problem is that Computation is not a proof-theoretic semantic
    system, and thus those rules don't apply.


    The dumbed down version is that the halting problem asks
    a question outside of the scope of finite string transformations.

    But it doesn't, not unless you think that programs can't be represented
    as finite strings.


    The halting problem proof does not fail because finite computation
    is too weak. It fails because it asks finite computation to
    decide a judgment that is not finitely grounded under operational
    semantics.

    But that is the issue, Operational Semantics for Programs are not
    actually finitely based, since programs can be non-halting.

    Just shows you don't know what your words actually mean.


    By operational semantics I mean the standard proof‑theoretic
    account of program meaning, where execution judgments are
    given by inference rules and termination corresponds to the
    existence of a finite derivation.

    Which is just incorrect. Since infinite derivation has meaning in the field.


    By proof‑theoretic semantics I mean the approach in which the
    meaning of a statement is determined by its rules of proof
    rather than by truth conditions in an external model.
    Operational semantics fits this pattern: programs have meaning
    through their execution rules, not through abstract denotations.

    WHich just isn't applicable to the field.

    Thus, you are showing you don't actualy understand the rules of
    Semantics, where you need to use the semantics that the system defines.

    Thus, your world is just built on lies.


    By denotational semantics I mean any semantics that assigns
    mathematical objects—functions, truth values, domains-to programs independently of how they are executed or proved. This contrasts
    with operational or proof‑theoretic semantics, where meaning is
    grounded in the structure of derivations rather than in an abstract mathematical object.

    Which can't handle the infinite set of the Natural Numbers.

    I guess you are just admitting that you goal of computing truth must be impossible, as we can't handle that level of abstractions.


    I use “denotational semantics” simply to refer to any framework
    that assigns meanings independently of operational behavior.


    Which are just limited systems.

    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Mikko@mikko.levanto@iki.fi to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Thu Jan 15 11:26:53 2026
    From Newsgroup: comp.ai.philosophy

    On 14/01/2026 16:55, Tristan Wibberley wrote:
    On 14/01/2026 08:53, Mikko wrote:
    For a non-deterministic machine there are three possibilities: it may
    halt always, sometimes, or never. THere is no oracle that can find the
    right answer about every meachne that contains the same oracle.


    We well into Turing c-machine territory here aren't we?

    It's the same with all machines.
    --
    Mikko
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Mikko@mikko.levanto@iki.fi to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Thu Jan 15 11:34:38 2026
    From Newsgroup: comp.ai.philosophy

    On 14/01/2026 21:32, olcott wrote:
    On 1/14/2026 3:01 AM, Mikko wrote:
    On 13/01/2026 16:31, olcott wrote:
    On 1/13/2026 3:13 AM, Mikko wrote:
    On 12/01/2026 16:32, olcott wrote:
    On 1/12/2026 4:47 AM, Mikko wrote:
    On 11/01/2026 16:24, Tristan Wibberley wrote:
    On 11/01/2026 10:13, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:

    No, that does not follow. If a required result cannot be
    derived by
    appying a finite string transformation then the it it is
    uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement. >>>>>>>>
    You can't determine whether the required result is computable >>>>>>>> before
    you have the requirement.


    Right, it is /in/ scope for computer science... for the /ology/. >>>>>>> Olcott
    here uses "computation" to refer to the practice. You give the
    requirement to the /ologist/ who correctly decides that it is not >>>>>>> for
    computation because it is not computable.

    You two so often violently agree; I find it warming to the heart. >>>>>>
    For pracitcal programming it is useful to know what is known to be >>>>>> uncomputable in order to avoid wasting time in attemlpts to do the >>>>>> impossible.

    It f-cking nuts that after more than 2000 years
    people still don't understand that self-contradictory
    expressions: "This sentence is not true" have no
    truth value. A smart high school student should have
    figured this out 2000 years ago.

    Irrelevant. For practical programming that question needn't be
    answered.

    The halting problem counter-example input is anchored
    in the Liar Paradox. Proof Theoretic Semantics rejects
    those two and Gödel's incompleteness and a bunch more
    as merely non-well-founded inputs.

    For every Turing machine the halting problem counter-example provably
    exists.

    Not when using Proof Theoretic Semantics grounded
    in the specification language. In this case the
    pathological input is simply rejected as ungrounded.

    Then your "Proof Theoretic Semantics" is not useful for discussion of
    Turing machines. For every Turing machine a counter example exists.
    And so exists a Turing machine that writes the counter example when
    given a Turing machine as input.
    --
    Mikko
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Mikko@mikko.levanto@iki.fi to sci.logic,comp.theory,sci.math,comp.ai.philosophy on Thu Jan 15 11:38:51 2026
    From Newsgroup: comp.ai.philosophy

    On 14/01/2026 21:19, olcott wrote:
    On 1/14/2026 1:58 AM, Mikko wrote:
    On 13/01/2026 16:17, olcott wrote:
    On 1/13/2026 2:46 AM, Mikko wrote:
    On 12/01/2026 16:43, olcott wrote:
    On 1/12/2026 4:51 AM, Mikko wrote:
    On 11/01/2026 16:23, olcott wrote:
    On 1/11/2026 4:22 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string >>>>>>>>>>>>>>>>> inputs by finite string transformation rules into >>>>>>>>>>>>>>>>> {Accept, Reject} values.

    The counter-example input to requires more than >>>>>>>>>>>>>>>>> can be derived from finite string transformation >>>>>>>>>>>>>>>>> rules applied to this specific input thus the >>>>>>>>>>>>>>>>> Halting Problem requires too much.

    In a sense the halting problem asks too much: the >>>>>>>>>>>>>>>> problem is proven to
    be unsolvable. In another sense it asks too little: >>>>>>>>>>>>>>>> usually we want to
    know whether a method halts on every input, not just one. >>>>>>>>>>>>>>>>
    Although the halting problem is unsolvable, there are >>>>>>>>>>>>>>>> partial solutions
    to the halting problem. In particular, every counter- >>>>>>>>>>>>>>>> example to the
    full solution is correctly solved by some partial deciders. >>>>>>>>>>>>>>>
    *if undecidability is correct then truth itself is broken* >>>>>>>>>>>>>>
    Depends on whether the word "truth" is interpeted in the >>>>>>>>>>>>>> standard
    sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory
    expressions are correctly rejected as semantically
    incoherent thus form no undecidability or incompleteness. >>>>>>>>>>>>
    The misconception is yours. No expression in the language of >>>>>>>>>>>> the first
    order group theory is self-contradictory. But the first >>>>>>>>>>>> order goupr
    theory is incomplete: it is impossible to prove that AB = BA >>>>>>>>>>>> is true
    for every A and every B but it is also impossible to prove >>>>>>>>>>>> that AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying
    finite string transformation rules to actual finite
    string inputs, then the required result exceeds the
    scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be
    derived by
    appying a finite string transformation then the it it is
    uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement. >>>>>>>>>
    Of course, it one can prove that the required result is not >>>>>>>>>> computable
    then that helps to avoid wasting effort to try the impossible. >>>>>>>>>> The
    situation is worse if it is not known that the required result >>>>>>>>>> is not
    computable.

    That something is not computable does not mean that there is >>>>>>>>>> anyting
    "incorrect" in the requirement.

    Yes it certainly does. Requiring the impossible is always an >>>>>>>>> error.

    It is a perfectly valid question to ask whther a particular
    reuqirement
    is satisfiable.

    Any yes/no question lacking a correct yes/no answer
    is an incorrect question that must be rejected on
    that basis.

    Irrelevant. The question whether a particular requirement is
    satisfiable
    does have an answer that is either "yes" or "no". In some ases it is >>>>>> not known whether it is "yes" or "no" and there may be no known
    way to
    find out be even then either "yes" or "no" is the correct answer.

    Now that I finally have the standard terminology:
    Proof-theoretic semantics has always been the correct
    formal system to handle decision problems.

    When it is asked a yes/no question lacking a correct
    yes/no answer it correctly determines non-well-founded.
    I have been correct all along and merely lacked the
    standard terminology.

    Irrelevant, as already noted above.
    Yes, it is. How to handle questions that lack a yes/no answer is
    irrelevant when discussing questions that do have a yes/no asnwer.
    Whether a particular requirement is satisriable always has a yes/no
    answer, so it is irrelevat how to handle questions that don't.


    The classical diagonal argument for the Halting Problem asks a halt
    decider H to evaluate a program D whose behavior depends on H’s own output.

    Not specifically. The requirement is that a halt decider shall
    determine about whatever program and input is described on the
    tape when the decider is started. This includes the possibility
    that the input describes the counter example.
    --
    Mikko
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Mikko@mikko.levanto@iki.fi to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Thu Jan 15 11:48:30 2026
    From Newsgroup: comp.ai.philosophy

    On 14/01/2026 19:28, olcott wrote:
    On 1/14/2026 1:40 AM, Mikko wrote:
    On 13/01/2026 16:27, olcott wrote:
    On 1/13/2026 3:11 AM, Mikko wrote:
    On 12/01/2026 16:29, olcott wrote:
    On 1/12/2026 4:44 AM, Mikko wrote:
    On 11/01/2026 16:18, olcott wrote:
    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string >>>>>>>>>>>>>>>>> inputs by finite string transformation rules into >>>>>>>>>>>>>>>>> {Accept, Reject} values.

    The counter-example input to requires more than >>>>>>>>>>>>>>>>> can be derived from finite string transformation >>>>>>>>>>>>>>>>> rules applied to this specific input thus the >>>>>>>>>>>>>>>>> Halting Problem requires too much.

    In a sense the halting problem asks too much: the >>>>>>>>>>>>>>>> problem is proven to
    be unsolvable. In another sense it asks too little: >>>>>>>>>>>>>>>> usually we want to
    know whether a method halts on every input, not just one. >>>>>>>>>>>>>>>>
    Although the halting problem is unsolvable, there are >>>>>>>>>>>>>>>> partial solutions
    to the halting problem. In particular, every counter- >>>>>>>>>>>>>>>> example to the
    full solution is correctly solved by some partial deciders. >>>>>>>>>>>>>>>
    *if undecidability is correct then truth itself is broken* >>>>>>>>>>>>>>
    Depends on whether the word "truth" is interpeted in the >>>>>>>>>>>>>> standard
    sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory
    expressions are correctly rejected as semantically
    incoherent thus form no undecidability or incompleteness. >>>>>>>>>>>>
    The misconception is yours. No expression in the language of >>>>>>>>>>>> the first
    order group theory is self-contradictory. But the first >>>>>>>>>>>> order goupr
    theory is incomplete: it is impossible to prove that AB = BA >>>>>>>>>>>> is true
    for every A and every B but it is also impossible to prove >>>>>>>>>>>> that AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying
    finite string transformation rules to actual finite
    string inputs, then the required result exceeds the
    scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be
    derived by
    appying a finite string transformation then the it it is
    uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement. >>>>>>>>
    You can't determine whether the required result is computable >>>>>>>> before
    you have the requirement.

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    Which only nmakes sense when the requirement that H must determine >>>>>> whether the computation presented by its input halts has already
    been presented.

    *ChatGPT explains how and why I am correct*

       *Reinterpretation of undecidability*
       The example of P and H demonstrates that what is
       often called “undecidable” is better understood as
       ill-posed with respect to computable semantics.
       When the specification is constrained to properties
       detectable via finite simulation and finite pattern
       recognition, computation proceeds normally and
       correctly. Undecidability only appears when the
       specification overreaches that boundary.

    It tries to explain but it does not prove.

    Its the same thing that I have been saying for years.
    It is not that a universal halt decider cannot exist.

    It is proven that an universal halt decider does not exist.

    “The system adopts Proof-Theoretic Semantics: meaning is determined
    by inferential role, and truth is internal to the theory. A theory T
    is defined by a finite set of stipulated atomic statements together
    with all expressions derivable from them under the inference rules.
    The statements belonging to T constitute its theorems, and these are
    exactly the statements that are true-in-T.”

    Under a system like the above rough draft all inputs
    having pathological self reference such as the halting
    problem counter-example input are simply rejected as
    non-well-founded. Tarski Undefinability, Gödel's
    incompleteness and the halting problem cease to exist.

    A Turing
    machine cannot determine the halting of all Turing machines and is
    therefore not an universla halt decider.

    This is not true in Proof Theoretic Semantics. I
    still have to refine my words. I may not have said
    that exactly correctly. The result is that in Proof
    Theoretic Semantics the counter-example is rejected
    as non-well-founded.

    That no Turing machine is a halt decider is a proven theorem and a
    truth about Turing machines. If your "Proof Thoeretic Semnatics"
    does not regard it as true then your "Proof Theoretic Semantics"
    is incomplete.


    My long‑term goal is to make ‘true on the basis of meaning’ computable.

    As meaning is not computable, how can "true on the balsis of meaning"
    be commputable?
    --
    Mikko
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Tristan Wibberley@tristan.wibberley+netnews2@alumni.manchester.ac.uk to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Thu Jan 15 15:57:44 2026
    From Newsgroup: comp.ai.philosophy

    On 15/01/2026 03:51, Richard Damon wrote:
    On 1/14/26 8:25 PM, olcott wrote:

    By operational semantics I mean the standard proof‑theoretic
    account of program meaning, where execution judgments are
    given by inference rules and termination corresponds to the
    existence of a finite derivation.

    Which is just incorrect. Since infinite derivation has meaning in the
    field.

    but you can't give an example of an infinite derivation that isn't also
    finite.
    --
    Tristan Wibberley

    The message body is Copyright (C) 2026 Tristan Wibberley except
    citations and quotations noted. All Rights Reserved except that you may,
    of course, cite it academically giving credit to me, distribute it
    verbatim as part of a usenet system or its archives, and use it to
    promote my greatness and general superiority without misrepresentation
    of my opinions other than my opinion of my greatness and general
    superiority which you _may_ misrepresent. You definitely MAY NOT train
    any production AI system with it but you may train experimental AI that
    will only be used for evaluation of the AI methods it implements.

    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Thu Jan 15 10:54:19 2026
    From Newsgroup: comp.ai.philosophy

    On 1/14/2026 9:51 PM, Richard Damon wrote:
    On 1/14/26 8:25 PM, olcott wrote:
    On 1/12/2026 9:19 PM, Richard Damon wrote:
    On 1/12/26 9:29 AM, olcott wrote:
    On 1/12/2026 4:44 AM, Mikko wrote:
    On 11/01/2026 16:18, olcott wrote:
    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string >>>>>>>>>>>>>>>> inputs by finite string transformation rules into >>>>>>>>>>>>>>>> {Accept, Reject} values.

    The counter-example input to requires more than >>>>>>>>>>>>>>>> can be derived from finite string transformation >>>>>>>>>>>>>>>> rules applied to this specific input thus the
    Halting Problem requires too much.

    In a sense the halting problem asks too much: the problem >>>>>>>>>>>>>>> is proven to
    be unsolvable. In another sense it asks too little: >>>>>>>>>>>>>>> usually we want to
    know whether a method halts on every input, not just one. >>>>>>>>>>>>>>>
    Although the halting problem is unsolvable, there are >>>>>>>>>>>>>>> partial solutions
    to the halting problem. In particular, every counter- >>>>>>>>>>>>>>> example to the
    full solution is correctly solved by some partial deciders. >>>>>>>>>>>>>>
    *if undecidability is correct then truth itself is broken* >>>>>>>>>>>>>
    Depends on whether the word "truth" is interpeted in the >>>>>>>>>>>>> standard
    sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory
    expressions are correctly rejected as semantically
    incoherent thus form no undecidability or incompleteness. >>>>>>>>>>>
    The misconception is yours. No expression in the language of >>>>>>>>>>> the first
    order group theory is self-contradictory. But the first order >>>>>>>>>>> goupr
    theory is incomplete: it is impossible to prove that AB = BA >>>>>>>>>>> is true
    for every A and every B but it is also impossible to prove >>>>>>>>>>> that AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying
    finite string transformation rules to actual finite
    string inputs, then the required result exceeds the
    scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be
    derived by
    appying a finite string transformation then the it it is
    uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement.

    You can't determine whether the required result is computable before >>>>>>> you have the requirement.

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    Which only nmakes sense when the requirement that H must determine
    whether the computation presented by its input halts has already
    been presented.

    *ChatGPT explains how and why I am correct*

       *Reinterpretation of undecidability*
       The example of P and H demonstrates that what is
       often called “undecidable” is better understood as
       ill-posed with respect to computable semantics.
       When the specification is constrained to properties
       detectable via finite simulation and finite pattern
       recognition, computation proceeds normally and
       correctly. Undecidability only appears when the
       specification overreaches that boundary.

    It tries to explain but it does not prove.


    Its the same thing that I have been saying for years.
    It is not that a universal halt decider cannot exist.

    It is that an input that does the opposite of whatever
    value the halt decider returns is non-well-founded
    within proof-theoretic semantics.


    But the problem is that Computation is not a proof-theoretic semantic
    system, and thus those rules don't apply.


    The dumbed down version is that the halting problem asks
    a question outside of the scope of finite string transformations.

    But it doesn't, not unless you think that programs can't be represented
    as finite strings.


    The halting problem proof does not fail because finite computation
    is too weak. It fails because it asks finite computation to
    decide a judgment that is not finitely grounded under operational
    semantics.

    But that is the issue, Operational Semantics for Programs are not
    actually finitely based, since programs can be non-halting.

    Just shows you don't know what your words actually mean.


    By operational semantics I mean the standard proof‑theoretic
    account of program meaning, where execution judgments are
    given by inference rules and termination corresponds to the
    existence of a finite derivation.

    Which is just incorrect. Since infinite derivation has meaning in the
    field.


    In the field of operational semantics within the standard
    proof‑theoretic account of program meaning infinite derivation
    means non-well-founded in the same way that ZFC correctly
    determines that Russell's Paradox specifies a non-well-founded set.
    --
    Copyright 2026 Olcott<br><br>

    My 28 year goal has been to make <br>
    "true on the basis of meaning expressed in language"<br>
    reliably computable.<br><br>

    This required establishing a new foundation<br>
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Thu Jan 15 11:34:28 2026
    From Newsgroup: comp.ai.philosophy

    On 1/14/2026 9:51 PM, Richard Damon wrote:
    On 1/14/26 8:25 PM, olcott wrote:
    On 1/12/2026 9:19 PM, Richard Damon wrote:
    On 1/12/26 9:29 AM, olcott wrote:
    On 1/12/2026 4:44 AM, Mikko wrote:
    On 11/01/2026 16:18, olcott wrote:
    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string >>>>>>>>>>>>>>>> inputs by finite string transformation rules into >>>>>>>>>>>>>>>> {Accept, Reject} values.

    The counter-example input to requires more than >>>>>>>>>>>>>>>> can be derived from finite string transformation >>>>>>>>>>>>>>>> rules applied to this specific input thus the
    Halting Problem requires too much.

    In a sense the halting problem asks too much: the problem >>>>>>>>>>>>>>> is proven to
    be unsolvable. In another sense it asks too little: >>>>>>>>>>>>>>> usually we want to
    know whether a method halts on every input, not just one. >>>>>>>>>>>>>>>
    Although the halting problem is unsolvable, there are >>>>>>>>>>>>>>> partial solutions
    to the halting problem. In particular, every counter- >>>>>>>>>>>>>>> example to the
    full solution is correctly solved by some partial deciders. >>>>>>>>>>>>>>
    *if undecidability is correct then truth itself is broken* >>>>>>>>>>>>>
    Depends on whether the word "truth" is interpeted in the >>>>>>>>>>>>> standard
    sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory
    expressions are correctly rejected as semantically
    incoherent thus form no undecidability or incompleteness. >>>>>>>>>>>
    The misconception is yours. No expression in the language of >>>>>>>>>>> the first
    order group theory is self-contradictory. But the first order >>>>>>>>>>> goupr
    theory is incomplete: it is impossible to prove that AB = BA >>>>>>>>>>> is true
    for every A and every B but it is also impossible to prove >>>>>>>>>>> that AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying
    finite string transformation rules to actual finite
    string inputs, then the required result exceeds the
    scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be
    derived by
    appying a finite string transformation then the it it is
    uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement.

    You can't determine whether the required result is computable before >>>>>>> you have the requirement.

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    Which only nmakes sense when the requirement that H must determine
    whether the computation presented by its input halts has already
    been presented.

    *ChatGPT explains how and why I am correct*

       *Reinterpretation of undecidability*
       The example of P and H demonstrates that what is
       often called “undecidable” is better understood as
       ill-posed with respect to computable semantics.
       When the specification is constrained to properties
       detectable via finite simulation and finite pattern
       recognition, computation proceeds normally and
       correctly. Undecidability only appears when the
       specification overreaches that boundary.

    It tries to explain but it does not prove.


    Its the same thing that I have been saying for years.
    It is not that a universal halt decider cannot exist.

    It is that an input that does the opposite of whatever
    value the halt decider returns is non-well-founded
    within proof-theoretic semantics.


    But the problem is that Computation is not a proof-theoretic semantic
    system, and thus those rules don't apply.


    The dumbed down version is that the halting problem asks
    a question outside of the scope of finite string transformations.

    But it doesn't, not unless you think that programs can't be represented
    as finite strings.


    The halting problem proof does not fail because finite computation
    is too weak. It fails because it asks finite computation to
    decide a judgment that is not finitely grounded under operational
    semantics.

    But that is the issue, Operational Semantics for Programs are not
    actually finitely based, since programs can be non-halting.

    Just shows you don't know what your words actually mean.


    By operational semantics I mean the standard proof‑theoretic
    account of program meaning, where execution judgments are
    given by inference rules and termination corresponds to the
    existence of a finite derivation.

    Which is just incorrect. Since infinite derivation has meaning in the
    field.


    The halting problem is not undecidable because computation is weak, but because the classical formulation uses a denotational semantics that is
    too permissive.

    In operational/proof‑theoretic semantics, where meaning is grounded in finite derivations, the halting predicate is not a well‑formed judgment
    — just as unrestricted comprehension was not a well‑formed judgment in naïve set theory.
    --
    Copyright 2026 Olcott<br><br>

    My 28 year goal has been to make <br>
    "true on the basis of meaning expressed in language"<br>
    reliably computable.<br><br>

    This required establishing a new foundation<br>
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to comp.theory,sci.logic,sci.math,comp.ai.philosophy,comp.lang.prolog on Thu Jan 15 14:30:41 2026
    From Newsgroup: comp.ai.philosophy

    On 1/15/2026 3:34 AM, Mikko wrote:
    On 14/01/2026 21:32, olcott wrote:
    On 1/14/2026 3:01 AM, Mikko wrote:
    On 13/01/2026 16:31, olcott wrote:
    On 1/13/2026 3:13 AM, Mikko wrote:
    On 12/01/2026 16:32, olcott wrote:
    On 1/12/2026 4:47 AM, Mikko wrote:
    On 11/01/2026 16:24, Tristan Wibberley wrote:
    On 11/01/2026 10:13, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:

    No, that does not follow. If a required result cannot be >>>>>>>>>>> derived by
    appying a finite string transformation then the it it is >>>>>>>>>>> uncomputable.

    Right. Outside the scope of computation. Requiring anything >>>>>>>>>> outside the scope of computation is an incorrect requirement. >>>>>>>>>
    You can't determine whether the required result is computable >>>>>>>>> before
    you have the requirement.


    Right, it is /in/ scope for computer science... for the /ology/. >>>>>>>> Olcott
    here uses "computation" to refer to the practice. You give the >>>>>>>> requirement to the /ologist/ who correctly decides that it is >>>>>>>> not for
    computation because it is not computable.

    You two so often violently agree; I find it warming to the heart. >>>>>>>
    For pracitcal programming it is useful to know what is known to be >>>>>>> uncomputable in order to avoid wasting time in attemlpts to do the >>>>>>> impossible.

    It f-cking nuts that after more than 2000 years
    people still don't understand that self-contradictory
    expressions: "This sentence is not true" have no
    truth value. A smart high school student should have
    figured this out 2000 years ago.

    Irrelevant. For practical programming that question needn't be
    answered.

    The halting problem counter-example input is anchored
    in the Liar Paradox. Proof Theoretic Semantics rejects
    those two and Gödel's incompleteness and a bunch more
    as merely non-well-founded inputs.

    For every Turing machine the halting problem counter-example provably
    exists.

    Not when using Proof Theoretic Semantics grounded
    in the specification language. In this case the
    pathological input is simply rejected as ungrounded.

    Then your "Proof Theoretic Semantics" is not useful for discussion of
    Turing machines. For every Turing machine a counter example exists.
    And so exists a Turing machine that writes the counter example when
    given a Turing machine as input.


    It is "not useful" in the same way that ZFC was
    "not useful" for addressing Russell's Paradox.

    The halting problem is not undecidable because computation
    is weak, but because the classical formulation uses a
    denotational semantics that is too permissive.

    In operational/proof‑theoretic semantics, where meaning
    is grounded in finite derivations, the halting predicate
    is not a well‑formed judgment — just as unrestricted
    comprehension was not a well‑formed judgment in naïve
    set theory.
    --
    Copyright 2026 Olcott<br><br>

    My 28 year goal has been to make <br>
    "true on the basis of meaning expressed in language"<br>
    reliably computable.<br><br>

    This required establishing a new foundation<br>
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to sci.logic,comp.theory,sci.math,comp.ai.philosophy,comp.lang.prolog on Thu Jan 15 17:38:26 2026
    From Newsgroup: comp.ai.philosophy

    On 1/15/2026 3:48 AM, Mikko wrote:
    On 14/01/2026 19:28, olcott wrote:
    On 1/14/2026 1:40 AM, Mikko wrote:
    On 13/01/2026 16:27, olcott wrote:
    On 1/13/2026 3:11 AM, Mikko wrote:
    On 12/01/2026 16:29, olcott wrote:
    On 1/12/2026 4:44 AM, Mikko wrote:
    On 11/01/2026 16:18, olcott wrote:
    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string >>>>>>>>>>>>>>>>>> inputs by finite string transformation rules into >>>>>>>>>>>>>>>>>> {Accept, Reject} values.

    The counter-example input to requires more than >>>>>>>>>>>>>>>>>> can be derived from finite string transformation >>>>>>>>>>>>>>>>>> rules applied to this specific input thus the >>>>>>>>>>>>>>>>>> Halting Problem requires too much.

    In a sense the halting problem asks too much: the >>>>>>>>>>>>>>>>> problem is proven to
    be unsolvable. In another sense it asks too little: >>>>>>>>>>>>>>>>> usually we want to
    know whether a method halts on every input, not just one. >>>>>>>>>>>>>>>>>
    Although the halting problem is unsolvable, there are >>>>>>>>>>>>>>>>> partial solutions
    to the halting problem. In particular, every counter- >>>>>>>>>>>>>>>>> example to the
    full solution is correctly solved by some partial >>>>>>>>>>>>>>>>> deciders.

    *if undecidability is correct then truth itself is broken* >>>>>>>>>>>>>>>
    Depends on whether the word "truth" is interpeted in the >>>>>>>>>>>>>>> standard
    sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory >>>>>>>>>>>>>> expressions are correctly rejected as semantically >>>>>>>>>>>>>> incoherent thus form no undecidability or incompleteness. >>>>>>>>>>>>>
    The misconception is yours. No expression in the language >>>>>>>>>>>>> of the first
    order group theory is self-contradictory. But the first >>>>>>>>>>>>> order goupr
    theory is incomplete: it is impossible to prove that AB = >>>>>>>>>>>>> BA is true
    for every A and every B but it is also impossible to prove >>>>>>>>>>>>> that AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying
    finite string transformation rules to actual finite
    string inputs, then the required result exceeds the
    scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be >>>>>>>>>>> derived by
    appying a finite string transformation then the it it is >>>>>>>>>>> uncomputable.

    Right. Outside the scope of computation. Requiring anything >>>>>>>>>> outside the scope of computation is an incorrect requirement. >>>>>>>>>
    You can't determine whether the required result is computable >>>>>>>>> before
    you have the requirement.

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    Which only nmakes sense when the requirement that H must determine >>>>>>> whether the computation presented by its input halts has already >>>>>>> been presented.

    *ChatGPT explains how and why I am correct*

       *Reinterpretation of undecidability*
       The example of P and H demonstrates that what is
       often called “undecidable” is better understood as
       ill-posed with respect to computable semantics.
       When the specification is constrained to properties
       detectable via finite simulation and finite pattern
       recognition, computation proceeds normally and
       correctly. Undecidability only appears when the
       specification overreaches that boundary.

    It tries to explain but it does not prove.

    Its the same thing that I have been saying for years.
    It is not that a universal halt decider cannot exist.

    It is proven that an universal halt decider does not exist.

    “The system adopts Proof-Theoretic Semantics: meaning is determined >>>> by inferential role, and truth is internal to the theory. A theory T
    is defined by a finite set of stipulated atomic statements together
    with all expressions derivable from them under the inference rules.
    The statements belonging to T constitute its theorems, and these are
    exactly the statements that are true-in-T.”

    Under a system like the above rough draft all inputs
    having pathological self reference such as the halting
    problem counter-example input are simply rejected as
    non-well-founded. Tarski Undefinability, Gödel's
    incompleteness and the halting problem cease to exist.

    A Turing
    machine cannot determine the halting of all Turing machines and is
    therefore not an universla halt decider.

    This is not true in Proof Theoretic Semantics. I
    still have to refine my words. I may not have said
    that exactly correctly. The result is that in Proof
    Theoretic Semantics the counter-example is rejected
    as non-well-founded.

    That no Turing machine is a halt decider is a proven theorem and a
    truth about Turing machines. If your "Proof Thoeretic Semnatics"
    does not regard it as true then your "Proof Theoretic Semantics"
    is incomplete.


    My long‑term goal is to make ‘true on the basis of meaning’ computable.

    As meaning is not computable, how can "true on the balsis of meaning"
    be commputable?


    Under *proof‑theoretic semantics*
    "true on the basis of meaning expressed in language"
    has always been entirely computable.

    The difference now is that I have a standard
    conventional term-of-the-art basis to prove
    my point.
    --
    Copyright 2026 Olcott<br><br>

    My 28 year goal has been to make <br>
    "true on the basis of meaning expressed in language"<br>
    reliably computable.<br><br>

    This required establishing a new foundation<br>
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Richard Damon@Richard@Damon-Family.org to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Thu Jan 15 22:27:57 2026
    From Newsgroup: comp.ai.philosophy

    On 1/15/26 12:34 PM, olcott wrote:
    On 1/14/2026 9:51 PM, Richard Damon wrote:
    On 1/14/26 8:25 PM, olcott wrote:
    On 1/12/2026 9:19 PM, Richard Damon wrote:
    On 1/12/26 9:29 AM, olcott wrote:
    On 1/12/2026 4:44 AM, Mikko wrote:
    On 11/01/2026 16:18, olcott wrote:
    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string >>>>>>>>>>>>>>>>> inputs by finite string transformation rules into >>>>>>>>>>>>>>>>> {Accept, Reject} values.

    The counter-example input to requires more than >>>>>>>>>>>>>>>>> can be derived from finite string transformation >>>>>>>>>>>>>>>>> rules applied to this specific input thus the >>>>>>>>>>>>>>>>> Halting Problem requires too much.

    In a sense the halting problem asks too much: the >>>>>>>>>>>>>>>> problem is proven to
    be unsolvable. In another sense it asks too little: >>>>>>>>>>>>>>>> usually we want to
    know whether a method halts on every input, not just one. >>>>>>>>>>>>>>>>
    Although the halting problem is unsolvable, there are >>>>>>>>>>>>>>>> partial solutions
    to the halting problem. In particular, every counter- >>>>>>>>>>>>>>>> example to the
    full solution is correctly solved by some partial deciders. >>>>>>>>>>>>>>>
    *if undecidability is correct then truth itself is broken* >>>>>>>>>>>>>>
    Depends on whether the word "truth" is interpeted in the >>>>>>>>>>>>>> standard
    sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory
    expressions are correctly rejected as semantically
    incoherent thus form no undecidability or incompleteness. >>>>>>>>>>>>
    The misconception is yours. No expression in the language of >>>>>>>>>>>> the first
    order group theory is self-contradictory. But the first >>>>>>>>>>>> order goupr
    theory is incomplete: it is impossible to prove that AB = BA >>>>>>>>>>>> is true
    for every A and every B but it is also impossible to prove >>>>>>>>>>>> that AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying
    finite string transformation rules to actual finite
    string inputs, then the required result exceeds the
    scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be
    derived by
    appying a finite string transformation then the it it is
    uncomputable.

    Right. Outside the scope of computation. Requiring anything
    outside the scope of computation is an incorrect requirement. >>>>>>>>
    You can't determine whether the required result is computable >>>>>>>> before
    you have the requirement.

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    Which only nmakes sense when the requirement that H must determine >>>>>> whether the computation presented by its input halts has already
    been presented.

    *ChatGPT explains how and why I am correct*

       *Reinterpretation of undecidability*
       The example of P and H demonstrates that what is
       often called “undecidable” is better understood as
       ill-posed with respect to computable semantics.
       When the specification is constrained to properties
       detectable via finite simulation and finite pattern
       recognition, computation proceeds normally and
       correctly. Undecidability only appears when the
       specification overreaches that boundary.

    It tries to explain but it does not prove.


    Its the same thing that I have been saying for years.
    It is not that a universal halt decider cannot exist.

    It is that an input that does the opposite of whatever
    value the halt decider returns is non-well-founded
    within proof-theoretic semantics.


    But the problem is that Computation is not a proof-theoretic
    semantic system, and thus those rules don't apply.


    The dumbed down version is that the halting problem asks
    a question outside of the scope of finite string transformations.

    But it doesn't, not unless you think that programs can't be
    represented as finite strings.


    The halting problem proof does not fail because finite computation
    is too weak. It fails because it asks finite computation to
    decide a judgment that is not finitely grounded under operational
    semantics.

    But that is the issue, Operational Semantics for Programs are not
    actually finitely based, since programs can be non-halting.

    Just shows you don't know what your words actually mean.


    By operational semantics I mean the standard proof‑theoretic
    account of program meaning, where execution judgments are
    given by inference rules and termination corresponds to the
    existence of a finite derivation.

    Which is just incorrect. Since infinite derivation has meaning in the
    field.


    The halting problem is not undecidable because computation is weak, but because the classical formulation uses a denotational semantics that is
    too permissive.

    Nope.


    In operational/proof‑theoretic semantics, where meaning is grounded in finite derivations, the halting predicate is not a well‑formed judgment — just as unrestricted comprehension was not a well‑formed judgment in naïve set theory.


    In other words, by trying to enforce your interpreation, you system
    becomes unworkable, as you can't tell if you can ask a question.

    The problem is that systems like this grow faster in power to generate
    than your logic grow in power to decide, and either you accept that some truths are unprovable (and thus accept the truth-conditional view) or
    you need to just abandon the ability to actually work in the system as
    you can't tell what questions are reasonable.

    All you are doing is proving that you are just too stupid to understand
    the implications of what you are talking about, because you never really understood what the words actually mean.
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Thu Jan 15 22:03:52 2026
    From Newsgroup: comp.ai.philosophy

    On 1/15/2026 9:27 PM, Richard Damon wrote:
    On 1/15/26 12:34 PM, olcott wrote:
    On 1/14/2026 9:51 PM, Richard Damon wrote:
    On 1/14/26 8:25 PM, olcott wrote:
    On 1/12/2026 9:19 PM, Richard Damon wrote:
    On 1/12/26 9:29 AM, olcott wrote:
    On 1/12/2026 4:44 AM, Mikko wrote:
    On 11/01/2026 16:18, olcott wrote:
    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string >>>>>>>>>>>>>>>>>> inputs by finite string transformation rules into >>>>>>>>>>>>>>>>>> {Accept, Reject} values.

    The counter-example input to requires more than >>>>>>>>>>>>>>>>>> can be derived from finite string transformation >>>>>>>>>>>>>>>>>> rules applied to this specific input thus the >>>>>>>>>>>>>>>>>> Halting Problem requires too much.

    In a sense the halting problem asks too much: the >>>>>>>>>>>>>>>>> problem is proven to
    be unsolvable. In another sense it asks too little: >>>>>>>>>>>>>>>>> usually we want to
    know whether a method halts on every input, not just one. >>>>>>>>>>>>>>>>>
    Although the halting problem is unsolvable, there are >>>>>>>>>>>>>>>>> partial solutions
    to the halting problem. In particular, every counter- >>>>>>>>>>>>>>>>> example to the
    full solution is correctly solved by some partial >>>>>>>>>>>>>>>>> deciders.

    *if undecidability is correct then truth itself is broken* >>>>>>>>>>>>>>>
    Depends on whether the word "truth" is interpeted in the >>>>>>>>>>>>>>> standard
    sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory >>>>>>>>>>>>>> expressions are correctly rejected as semantically >>>>>>>>>>>>>> incoherent thus form no undecidability or incompleteness. >>>>>>>>>>>>>
    The misconception is yours. No expression in the language >>>>>>>>>>>>> of the first
    order group theory is self-contradictory. But the first >>>>>>>>>>>>> order goupr
    theory is incomplete: it is impossible to prove that AB = >>>>>>>>>>>>> BA is true
    for every A and every B but it is also impossible to prove >>>>>>>>>>>>> that AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying
    finite string transformation rules to actual finite
    string inputs, then the required result exceeds the
    scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be >>>>>>>>>>> derived by
    appying a finite string transformation then the it it is >>>>>>>>>>> uncomputable.

    Right. Outside the scope of computation. Requiring anything >>>>>>>>>> outside the scope of computation is an incorrect requirement. >>>>>>>>>
    You can't determine whether the required result is computable >>>>>>>>> before
    you have the requirement.

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    Which only nmakes sense when the requirement that H must determine >>>>>>> whether the computation presented by its input halts has already >>>>>>> been presented.

    *ChatGPT explains how and why I am correct*

       *Reinterpretation of undecidability*
       The example of P and H demonstrates that what is
       often called “undecidable” is better understood as
       ill-posed with respect to computable semantics.
       When the specification is constrained to properties
       detectable via finite simulation and finite pattern
       recognition, computation proceeds normally and
       correctly. Undecidability only appears when the
       specification overreaches that boundary.

    It tries to explain but it does not prove.


    Its the same thing that I have been saying for years.
    It is not that a universal halt decider cannot exist.

    It is that an input that does the opposite of whatever
    value the halt decider returns is non-well-founded
    within proof-theoretic semantics.


    But the problem is that Computation is not a proof-theoretic
    semantic system, and thus those rules don't apply.


    The dumbed down version is that the halting problem asks
    a question outside of the scope of finite string transformations.

    But it doesn't, not unless you think that programs can't be
    represented as finite strings.


    The halting problem proof does not fail because finite computation
    is too weak. It fails because it asks finite computation to
    decide a judgment that is not finitely grounded under operational
    semantics.

    But that is the issue, Operational Semantics for Programs are not
    actually finitely based, since programs can be non-halting.

    Just shows you don't know what your words actually mean.


    By operational semantics I mean the standard proof‑theoretic
    account of program meaning, where execution judgments are
    given by inference rules and termination corresponds to the
    existence of a finite derivation.

    Which is just incorrect. Since infinite derivation has meaning in the
    field.


    The halting problem is not undecidable because computation is weak,
    but because the classical formulation uses a denotational semantics
    that is too permissive.

    Nope.


    In operational/proof‑theoretic semantics, where meaning is grounded in
    finite derivations, the halting predicate is not a well‑formed
    judgment — just as unrestricted comprehension was not a well‑formed
    judgment in naïve set theory.


    In other words, by trying to enforce your interpreation, you system
    becomes unworkable, as you can't tell if you can ask a question.


    It is the same ∀x ∈ T ((True(T, x) ≡ (T ⊢ x))
    that I have been talking about for years except that
    it is now grounded in well-founded proof‑theoretic
    semantics.

    The problem is that systems like this grow faster in power to generate
    than your logic grow in power to decide, and either you accept that some truths are unprovable (and thus accept the truth-conditional view) or
    you need to just abandon the ability to actually work in the system as
    you can't tell what questions are reasonable.

    All you are doing is proving that you are just too stupid to understand
    the implications of what you are talking about, because you never really understood what the words actually mean.
    --
    Copyright 2026 Olcott<br><br>

    My 28 year goal has been to make <br>
    "true on the basis of meaning expressed in language"<br>
    reliably computable.<br><br>

    This required establishing a new foundation<br>
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Mikko@mikko.levanto@iki.fi to sci.logic,comp.theory,sci.math,comp.ai.philosophy,comp.lang.prolog on Fri Jan 16 11:17:14 2026
    From Newsgroup: comp.ai.philosophy

    On 16/01/2026 01:38, olcott wrote:
    On 1/15/2026 3:48 AM, Mikko wrote:
    On 14/01/2026 19:28, olcott wrote:
    On 1/14/2026 1:40 AM, Mikko wrote:
    On 13/01/2026 16:27, olcott wrote:
    On 1/13/2026 3:11 AM, Mikko wrote:
    On 12/01/2026 16:29, olcott wrote:
    On 1/12/2026 4:44 AM, Mikko wrote:
    On 11/01/2026 16:18, olcott wrote:
    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string >>>>>>>>>>>>>>>>>>> inputs by finite string transformation rules into >>>>>>>>>>>>>>>>>>> {Accept, Reject} values.

    The counter-example input to requires more than >>>>>>>>>>>>>>>>>>> can be derived from finite string transformation >>>>>>>>>>>>>>>>>>> rules applied to this specific input thus the >>>>>>>>>>>>>>>>>>> Halting Problem requires too much.

    In a sense the halting problem asks too much: the >>>>>>>>>>>>>>>>>> problem is proven to
    be unsolvable. In another sense it asks too little: >>>>>>>>>>>>>>>>>> usually we want to
    know whether a method halts on every input, not just one. >>>>>>>>>>>>>>>>>>
    Although the halting problem is unsolvable, there are >>>>>>>>>>>>>>>>>> partial solutions
    to the halting problem. In particular, every counter- >>>>>>>>>>>>>>>>>> example to the
    full solution is correctly solved by some partial >>>>>>>>>>>>>>>>>> deciders.

    *if undecidability is correct then truth itself is broken* >>>>>>>>>>>>>>>>
    Depends on whether the word "truth" is interpeted in the >>>>>>>>>>>>>>>> standard
    sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory >>>>>>>>>>>>>>> expressions are correctly rejected as semantically >>>>>>>>>>>>>>> incoherent thus form no undecidability or incompleteness. >>>>>>>>>>>>>>
    The misconception is yours. No expression in the language >>>>>>>>>>>>>> of the first
    order group theory is self-contradictory. But the first >>>>>>>>>>>>>> order goupr
    theory is incomplete: it is impossible to prove that AB = >>>>>>>>>>>>>> BA is true
    for every A and every B but it is also impossible to prove >>>>>>>>>>>>>> that AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying >>>>>>>>>>>>> finite string transformation rules to actual finite
    string inputs, then the required result exceeds the
    scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be >>>>>>>>>>>> derived by
    appying a finite string transformation then the it it is >>>>>>>>>>>> uncomputable.

    Right. Outside the scope of computation. Requiring anything >>>>>>>>>>> outside the scope of computation is an incorrect requirement. >>>>>>>>>>
    You can't determine whether the required result is computable >>>>>>>>>> before
    you have the requirement.

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    Which only nmakes sense when the requirement that H must determine >>>>>>>> whether the computation presented by its input halts has already >>>>>>>> been presented.

    *ChatGPT explains how and why I am correct*

       *Reinterpretation of undecidability*
       The example of P and H demonstrates that what is
       often called “undecidable” is better understood as
       ill-posed with respect to computable semantics.
       When the specification is constrained to properties
       detectable via finite simulation and finite pattern
       recognition, computation proceeds normally and
       correctly. Undecidability only appears when the
       specification overreaches that boundary.

    It tries to explain but it does not prove.

    Its the same thing that I have been saying for years.
    It is not that a universal halt decider cannot exist.

    It is proven that an universal halt decider does not exist.

    “The system adopts Proof-Theoretic Semantics: meaning is determined >>>>> by inferential role, and truth is internal to the theory. A theory
    T is defined by a finite set of stipulated atomic statements
    together with all expressions derivable from them under the
    inference rules. The statements belonging to T constitute its
    theorems, and these are exactly the statements that are true-in-T.” >>>>>
    Under a system like the above rough draft all inputs
    having pathological self reference such as the halting
    problem counter-example input are simply rejected as
    non-well-founded. Tarski Undefinability, Gödel's
    incompleteness and the halting problem cease to exist.

    A Turing
    machine cannot determine the halting of all Turing machines and is >>>>>> therefore not an universla halt decider.

    This is not true in Proof Theoretic Semantics. I
    still have to refine my words. I may not have said
    that exactly correctly. The result is that in Proof
    Theoretic Semantics the counter-example is rejected
    as non-well-founded.

    That no Turing machine is a halt decider is a proven theorem and a
    truth about Turing machines. If your "Proof Thoeretic Semnatics"
    does not regard it as true then your "Proof Theoretic Semantics"
    is incomplete.


    My long‑term goal is to make ‘true on the basis of meaning’ computable.

    As meaning is not computable, how can "true on the balsis of meaning"
    be commputable?

    Under *proof‑theoretic semantics*
    "true on the basis of meaning expressed in language"
    has always been entirely computable.

    Have you already put the algorithm to some web page?
    --
    Mikko
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Mikko@mikko.levanto@iki.fi to comp.theory,sci.logic,sci.math,comp.ai.philosophy,comp.lang.prolog on Fri Jan 16 11:32:39 2026
    From Newsgroup: comp.ai.philosophy

    On 15/01/2026 22:30, olcott wrote:
    On 1/15/2026 3:34 AM, Mikko wrote:
    On 14/01/2026 21:32, olcott wrote:
    On 1/14/2026 3:01 AM, Mikko wrote:
    On 13/01/2026 16:31, olcott wrote:
    On 1/13/2026 3:13 AM, Mikko wrote:
    On 12/01/2026 16:32, olcott wrote:
    On 1/12/2026 4:47 AM, Mikko wrote:
    On 11/01/2026 16:24, Tristan Wibberley wrote:
    On 11/01/2026 10:13, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:

    No, that does not follow. If a required result cannot be >>>>>>>>>>>> derived by
    appying a finite string transformation then the it it is >>>>>>>>>>>> uncomputable.

    Right. Outside the scope of computation. Requiring anything >>>>>>>>>>> outside the scope of computation is an incorrect requirement. >>>>>>>>>>
    You can't determine whether the required result is computable >>>>>>>>>> before
    you have the requirement.


    Right, it is /in/ scope for computer science... for the /
    ology/. Olcott
    here uses "computation" to refer to the practice. You give the >>>>>>>>> requirement to the /ologist/ who correctly decides that it is >>>>>>>>> not for
    computation because it is not computable.

    You two so often violently agree; I find it warming to the heart. >>>>>>>>
    For pracitcal programming it is useful to know what is known to be >>>>>>>> uncomputable in order to avoid wasting time in attemlpts to do the >>>>>>>> impossible.

    It f-cking nuts that after more than 2000 years
    people still don't understand that self-contradictory
    expressions: "This sentence is not true" have no
    truth value. A smart high school student should have
    figured this out 2000 years ago.

    Irrelevant. For practical programming that question needn't be
    answered.

    The halting problem counter-example input is anchored
    in the Liar Paradox. Proof Theoretic Semantics rejects
    those two and Gödel's incompleteness and a bunch more
    as merely non-well-founded inputs.

    For every Turing machine the halting problem counter-example provably
    exists.

    Not when using Proof Theoretic Semantics grounded
    in the specification language. In this case the
    pathological input is simply rejected as ungrounded.

    Then your "Proof Theoretic Semantics" is not useful for discussion of
    Turing machines. For every Turing machine a counter example exists.
    And so exists a Turing machine that writes the counter example when
    given a Turing machine as input.


    It is "not useful" in the same way that ZFC was
    "not useful" for addressing Russell's Paradox.

    ZF or ZFC is to some extent useful for addressing Russell's paradox.
    It is an example of a set theory where Russell's paradox is avoided.
    If your "Proof Theretic Semantics" cannot handle the existence of
    a counter example for every Turing decider then it is not usefule
    for those who work on practical problems of program correctness.
    --
    Mikko
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to sci.logic,comp.theory,sci.math,comp.ai.philosophy,comp.lang.prolog on Fri Jan 16 08:12:56 2026
    From Newsgroup: comp.ai.philosophy

    On 1/16/2026 3:17 AM, Mikko wrote:
    On 16/01/2026 01:38, olcott wrote:
    On 1/15/2026 3:48 AM, Mikko wrote:
    On 14/01/2026 19:28, olcott wrote:
    On 1/14/2026 1:40 AM, Mikko wrote:
    On 13/01/2026 16:27, olcott wrote:
    On 1/13/2026 3:11 AM, Mikko wrote:
    On 12/01/2026 16:29, olcott wrote:
    On 1/12/2026 4:44 AM, Mikko wrote:
    On 11/01/2026 16:18, olcott wrote:
    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string >>>>>>>>>>>>>>>>>>>> inputs by finite string transformation rules into >>>>>>>>>>>>>>>>>>>> {Accept, Reject} values.

    The counter-example input to requires more than >>>>>>>>>>>>>>>>>>>> can be derived from finite string transformation >>>>>>>>>>>>>>>>>>>> rules applied to this specific input thus the >>>>>>>>>>>>>>>>>>>> Halting Problem requires too much.

    In a sense the halting problem asks too much: the >>>>>>>>>>>>>>>>>>> problem is proven to
    be unsolvable. In another sense it asks too little: >>>>>>>>>>>>>>>>>>> usually we want to
    know whether a method halts on every input, not just >>>>>>>>>>>>>>>>>>> one.

    Although the halting problem is unsolvable, there are >>>>>>>>>>>>>>>>>>> partial solutions
    to the halting problem. In particular, every counter- >>>>>>>>>>>>>>>>>>> example to the
    full solution is correctly solved by some partial >>>>>>>>>>>>>>>>>>> deciders.

    *if undecidability is correct then truth itself is >>>>>>>>>>>>>>>>>> broken*

    Depends on whether the word "truth" is interpeted in >>>>>>>>>>>>>>>>> the standard
    sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory >>>>>>>>>>>>>>>> expressions are correctly rejected as semantically >>>>>>>>>>>>>>>> incoherent thus form no undecidability or incompleteness. >>>>>>>>>>>>>>>
    The misconception is yours. No expression in the language >>>>>>>>>>>>>>> of the first
    order group theory is self-contradictory. But the first >>>>>>>>>>>>>>> order goupr
    theory is incomplete: it is impossible to prove that AB = >>>>>>>>>>>>>>> BA is true
    for every A and every B but it is also impossible to >>>>>>>>>>>>>>> prove that AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying >>>>>>>>>>>>>> finite string transformation rules to actual finite >>>>>>>>>>>>>> string inputs, then the required result exceeds the >>>>>>>>>>>>>> scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be >>>>>>>>>>>>> derived by
    appying a finite string transformation then the it it is >>>>>>>>>>>>> uncomputable.

    Right. Outside the scope of computation. Requiring anything >>>>>>>>>>>> outside the scope of computation is an incorrect requirement. >>>>>>>>>>>
    You can't determine whether the required result is computable >>>>>>>>>>> before
    you have the requirement.

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    Which only nmakes sense when the requirement that H must determine >>>>>>>>> whether the computation presented by its input halts has already >>>>>>>>> been presented.

    *ChatGPT explains how and why I am correct*

       *Reinterpretation of undecidability*
       The example of P and H demonstrates that what is
       often called “undecidable” is better understood as >>>>>>>>>>    ill-posed with respect to computable semantics.
       When the specification is constrained to properties
       detectable via finite simulation and finite pattern
       recognition, computation proceeds normally and
       correctly. Undecidability only appears when the
       specification overreaches that boundary.

    It tries to explain but it does not prove.

    Its the same thing that I have been saying for years.
    It is not that a universal halt decider cannot exist.

    It is proven that an universal halt decider does not exist.

    “The system adopts Proof-Theoretic Semantics: meaning is
    determined by inferential role, and truth is internal to the
    theory. A theory T is defined by a finite set of stipulated atomic >>>>>> statements together with all expressions derivable from them under >>>>>> the inference rules. The statements belonging to T constitute its >>>>>> theorems, and these are exactly the statements that are true-in-T.” >>>>>>
    Under a system like the above rough draft all inputs
    having pathological self reference such as the halting
    problem counter-example input are simply rejected as
    non-well-founded. Tarski Undefinability, Gödel's
    incompleteness and the halting problem cease to exist.

    A Turing
    machine cannot determine the halting of all Turing machines and is >>>>>>> therefore not an universla halt decider.

    This is not true in Proof Theoretic Semantics. I
    still have to refine my words. I may not have said
    that exactly correctly. The result is that in Proof
    Theoretic Semantics the counter-example is rejected
    as non-well-founded.

    That no Turing machine is a halt decider is a proven theorem and a
    truth about Turing machines. If your "Proof Thoeretic Semnatics"
    does not regard it as true then your "Proof Theoretic Semantics"
    is incomplete.


    My long‑term goal is to make ‘true on the basis of meaning’ computable.

    As meaning is not computable, how can "true on the balsis of meaning"
    be commputable?

    Under *proof‑theoretic semantics*
    "true on the basis of meaning expressed in language"
    has always been entirely computable.

    Have you already put the algorithm to some web page?


    I am still working on refining the presentation.
    --
    Copyright 2026 Olcott<br><br>

    My 28 year goal has been to make <br>
    "true on the basis of meaning expressed in language"<br>
    reliably computable.<br><br>

    This required establishing a new foundation<br>
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to sci.logic,comp.theory,sci.math,comp.ai.philosophy,comp.lang.prolog on Fri Jan 16 09:12:11 2026
    From Newsgroup: comp.ai.philosophy

    On 1/16/2026 3:17 AM, Mikko wrote:
    On 16/01/2026 01:38, olcott wrote:
    On 1/15/2026 3:48 AM, Mikko wrote:
    On 14/01/2026 19:28, olcott wrote:
    On 1/14/2026 1:40 AM, Mikko wrote:
    On 13/01/2026 16:27, olcott wrote:
    On 1/13/2026 3:11 AM, Mikko wrote:
    On 12/01/2026 16:29, olcott wrote:
    On 1/12/2026 4:44 AM, Mikko wrote:
    On 11/01/2026 16:18, olcott wrote:
    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string >>>>>>>>>>>>>>>>>>>> inputs by finite string transformation rules into >>>>>>>>>>>>>>>>>>>> {Accept, Reject} values.

    The counter-example input to requires more than >>>>>>>>>>>>>>>>>>>> can be derived from finite string transformation >>>>>>>>>>>>>>>>>>>> rules applied to this specific input thus the >>>>>>>>>>>>>>>>>>>> Halting Problem requires too much.

    In a sense the halting problem asks too much: the >>>>>>>>>>>>>>>>>>> problem is proven to
    be unsolvable. In another sense it asks too little: >>>>>>>>>>>>>>>>>>> usually we want to
    know whether a method halts on every input, not just >>>>>>>>>>>>>>>>>>> one.

    Although the halting problem is unsolvable, there are >>>>>>>>>>>>>>>>>>> partial solutions
    to the halting problem. In particular, every counter- >>>>>>>>>>>>>>>>>>> example to the
    full solution is correctly solved by some partial >>>>>>>>>>>>>>>>>>> deciders.

    *if undecidability is correct then truth itself is >>>>>>>>>>>>>>>>>> broken*

    Depends on whether the word "truth" is interpeted in >>>>>>>>>>>>>>>>> the standard
    sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory >>>>>>>>>>>>>>>> expressions are correctly rejected as semantically >>>>>>>>>>>>>>>> incoherent thus form no undecidability or incompleteness. >>>>>>>>>>>>>>>
    The misconception is yours. No expression in the language >>>>>>>>>>>>>>> of the first
    order group theory is self-contradictory. But the first >>>>>>>>>>>>>>> order goupr
    theory is incomplete: it is impossible to prove that AB = >>>>>>>>>>>>>>> BA is true
    for every A and every B but it is also impossible to >>>>>>>>>>>>>>> prove that AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying >>>>>>>>>>>>>> finite string transformation rules to actual finite >>>>>>>>>>>>>> string inputs, then the required result exceeds the >>>>>>>>>>>>>> scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be >>>>>>>>>>>>> derived by
    appying a finite string transformation then the it it is >>>>>>>>>>>>> uncomputable.

    Right. Outside the scope of computation. Requiring anything >>>>>>>>>>>> outside the scope of computation is an incorrect requirement. >>>>>>>>>>>
    You can't determine whether the required result is computable >>>>>>>>>>> before
    you have the requirement.

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    Which only nmakes sense when the requirement that H must determine >>>>>>>>> whether the computation presented by its input halts has already >>>>>>>>> been presented.

    *ChatGPT explains how and why I am correct*

       *Reinterpretation of undecidability*
       The example of P and H demonstrates that what is
       often called “undecidable” is better understood as >>>>>>>>>>    ill-posed with respect to computable semantics.
       When the specification is constrained to properties
       detectable via finite simulation and finite pattern
       recognition, computation proceeds normally and
       correctly. Undecidability only appears when the
       specification overreaches that boundary.

    It tries to explain but it does not prove.

    Its the same thing that I have been saying for years.
    It is not that a universal halt decider cannot exist.

    It is proven that an universal halt decider does not exist.

    “The system adopts Proof-Theoretic Semantics: meaning is
    determined by inferential role, and truth is internal to the
    theory. A theory T is defined by a finite set of stipulated atomic >>>>>> statements together with all expressions derivable from them under >>>>>> the inference rules. The statements belonging to T constitute its >>>>>> theorems, and these are exactly the statements that are true-in-T.” >>>>>>
    Under a system like the above rough draft all inputs
    having pathological self reference such as the halting
    problem counter-example input are simply rejected as
    non-well-founded. Tarski Undefinability, Gödel's
    incompleteness and the halting problem cease to exist.

    A Turing
    machine cannot determine the halting of all Turing machines and is >>>>>>> therefore not an universla halt decider.

    This is not true in Proof Theoretic Semantics. I
    still have to refine my words. I may not have said
    that exactly correctly. The result is that in Proof
    Theoretic Semantics the counter-example is rejected
    as non-well-founded.

    That no Turing machine is a halt decider is a proven theorem and a
    truth about Turing machines. If your "Proof Thoeretic Semnatics"
    does not regard it as true then your "Proof Theoretic Semantics"
    is incomplete.


    My long‑term goal is to make ‘true on the basis of meaning’ computable.

    As meaning is not computable, how can "true on the balsis of meaning"
    be commputable?

    Under *proof‑theoretic semantics*
    "true on the basis of meaning expressed in language"
    has always been entirely computable.

    Have you already put the algorithm to some web page?


    Proof Theoretic Semantics Blocks Pathological Self-Reference https://philpapers.org/rec/OLCPTS
    --
    Copyright 2026 Olcott<br><br>

    My 28 year goal has been to make <br>
    "true on the basis of meaning expressed in language"<br>
    reliably computable.<br><br>

    This required establishing a new foundation<br>
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Fri Jan 16 09:38:16 2026
    From Newsgroup: comp.ai.philosophy

    On 1/16/2026 3:32 AM, Mikko wrote:
    On 15/01/2026 22:30, olcott wrote:
    On 1/15/2026 3:34 AM, Mikko wrote:
    On 14/01/2026 21:32, olcott wrote:
    On 1/14/2026 3:01 AM, Mikko wrote:
    On 13/01/2026 16:31, olcott wrote:
    On 1/13/2026 3:13 AM, Mikko wrote:
    On 12/01/2026 16:32, olcott wrote:
    On 1/12/2026 4:47 AM, Mikko wrote:
    On 11/01/2026 16:24, Tristan Wibberley wrote:
    On 11/01/2026 10:13, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:

    No, that does not follow. If a required result cannot be >>>>>>>>>>>>> derived by
    appying a finite string transformation then the it it is >>>>>>>>>>>>> uncomputable.

    Right. Outside the scope of computation. Requiring anything >>>>>>>>>>>> outside the scope of computation is an incorrect requirement. >>>>>>>>>>>
    You can't determine whether the required result is computable >>>>>>>>>>> before
    you have the requirement.


    Right, it is /in/ scope for computer science... for the / >>>>>>>>>> ology/. Olcott
    here uses "computation" to refer to the practice. You give the >>>>>>>>>> requirement to the /ologist/ who correctly decides that it is >>>>>>>>>> not for
    computation because it is not computable.

    You two so often violently agree; I find it warming to the heart. >>>>>>>>>
    For pracitcal programming it is useful to know what is known to be >>>>>>>>> uncomputable in order to avoid wasting time in attemlpts to do the >>>>>>>>> impossible.

    It f-cking nuts that after more than 2000 years
    people still don't understand that self-contradictory
    expressions: "This sentence is not true" have no
    truth value. A smart high school student should have
    figured this out 2000 years ago.

    Irrelevant. For practical programming that question needn't be
    answered.

    The halting problem counter-example input is anchored
    in the Liar Paradox. Proof Theoretic Semantics rejects
    those two and Gödel's incompleteness and a bunch more
    as merely non-well-founded inputs.

    For every Turing machine the halting problem counter-example provably >>>>> exists.

    Not when using Proof Theoretic Semantics grounded
    in the specification language. In this case the
    pathological input is simply rejected as ungrounded.

    Then your "Proof Theoretic Semantics" is not useful for discussion of
    Turing machines. For every Turing machine a counter example exists.
    And so exists a Turing machine that writes the counter example when
    given a Turing machine as input.


    It is "not useful" in the same way that ZFC was
    "not useful" for addressing Russell's Paradox.

    ZF or ZFC is to some extent useful for addressing Russell's paradox.
    It is an example of a set theory where Russell's paradox is avoided.
    If your "Proof Theretic Semantics" cannot handle the existence of
    a counter example for every Turing decider then it is not usefule
    for those who work on practical problems of program correctness.


    Proof theoretic semantics addresses Gödel Incompleteness
    for PA in a way similar to the way that ZFC addresses
    Russell's Paradox in set theory.

    ZFC resolves Russell’s paradox by restricting the formation
    of sets to those justified by proof‑theoretic rules.

    Proof‑theoretic semantics resolves the meaning‑theoretic
    issues behind Gödel incompleteness by restricting the
    formation of truth‑bearing statements to those justified
    by inference rules.

    In both cases, paradox arises only when the semantics
    is too permissive, and disappears when meaning is grounded proof‑theoretically.

    Proof Theoretic Semantics Blocks Pathological Self-Reference https://philpapers.org/rec/OLCPTS
    --
    Copyright 2026 Olcott<br><br>

    My 28 year goal has been to make <br>
    "true on the basis of meaning expressed in language"<br>
    reliably computable.<br><br>

    This required establishing a new foundation<br>
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Richard Damon@Richard@Damon-Family.org to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Fri Jan 16 11:46:49 2026
    From Newsgroup: comp.ai.philosophy

    On 1/15/26 11:03 PM, olcott wrote:
    On 1/15/2026 9:27 PM, Richard Damon wrote:
    On 1/15/26 12:34 PM, olcott wrote:
    On 1/14/2026 9:51 PM, Richard Damon wrote:
    On 1/14/26 8:25 PM, olcott wrote:
    On 1/12/2026 9:19 PM, Richard Damon wrote:
    On 1/12/26 9:29 AM, olcott wrote:
    On 1/12/2026 4:44 AM, Mikko wrote:
    On 11/01/2026 16:18, olcott wrote:
    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string >>>>>>>>>>>>>>>>>>> inputs by finite string transformation rules into >>>>>>>>>>>>>>>>>>> {Accept, Reject} values.

    The counter-example input to requires more than >>>>>>>>>>>>>>>>>>> can be derived from finite string transformation >>>>>>>>>>>>>>>>>>> rules applied to this specific input thus the >>>>>>>>>>>>>>>>>>> Halting Problem requires too much.

    In a sense the halting problem asks too much: the >>>>>>>>>>>>>>>>>> problem is proven to
    be unsolvable. In another sense it asks too little: >>>>>>>>>>>>>>>>>> usually we want to
    know whether a method halts on every input, not just one. >>>>>>>>>>>>>>>>>>
    Although the halting problem is unsolvable, there are >>>>>>>>>>>>>>>>>> partial solutions
    to the halting problem. In particular, every counter- >>>>>>>>>>>>>>>>>> example to the
    full solution is correctly solved by some partial >>>>>>>>>>>>>>>>>> deciders.

    *if undecidability is correct then truth itself is broken* >>>>>>>>>>>>>>>>
    Depends on whether the word "truth" is interpeted in the >>>>>>>>>>>>>>>> standard
    sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory >>>>>>>>>>>>>>> expressions are correctly rejected as semantically >>>>>>>>>>>>>>> incoherent thus form no undecidability or incompleteness. >>>>>>>>>>>>>>
    The misconception is yours. No expression in the language >>>>>>>>>>>>>> of the first
    order group theory is self-contradictory. But the first >>>>>>>>>>>>>> order goupr
    theory is incomplete: it is impossible to prove that AB = >>>>>>>>>>>>>> BA is true
    for every A and every B but it is also impossible to prove >>>>>>>>>>>>>> that AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string
    inputs by finite string transformation rules into
    {Accept, Reject} values.

    When a required result cannot be derived by applying >>>>>>>>>>>>> finite string transformation rules to actual finite
    string inputs, then the required result exceeds the
    scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be >>>>>>>>>>>> derived by
    appying a finite string transformation then the it it is >>>>>>>>>>>> uncomputable.

    Right. Outside the scope of computation. Requiring anything >>>>>>>>>>> outside the scope of computation is an incorrect requirement. >>>>>>>>>>
    You can't determine whether the required result is computable >>>>>>>>>> before
    you have the requirement.

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    Which only nmakes sense when the requirement that H must determine >>>>>>>> whether the computation presented by its input halts has already >>>>>>>> been presented.

    *ChatGPT explains how and why I am correct*

       *Reinterpretation of undecidability*
       The example of P and H demonstrates that what is
       often called “undecidable” is better understood as
       ill-posed with respect to computable semantics.
       When the specification is constrained to properties
       detectable via finite simulation and finite pattern
       recognition, computation proceeds normally and
       correctly. Undecidability only appears when the
       specification overreaches that boundary.

    It tries to explain but it does not prove.


    Its the same thing that I have been saying for years.
    It is not that a universal halt decider cannot exist.

    It is that an input that does the opposite of whatever
    value the halt decider returns is non-well-founded
    within proof-theoretic semantics.


    But the problem is that Computation is not a proof-theoretic
    semantic system, and thus those rules don't apply.


    The dumbed down version is that the halting problem asks
    a question outside of the scope of finite string transformations.

    But it doesn't, not unless you think that programs can't be
    represented as finite strings.


    The halting problem proof does not fail because finite computation
    is too weak. It fails because it asks finite computation to
    decide a judgment that is not finitely grounded under operational
    semantics.

    But that is the issue, Operational Semantics for Programs are not
    actually finitely based, since programs can be non-halting.

    Just shows you don't know what your words actually mean.


    By operational semantics I mean the standard proof‑theoretic
    account of program meaning, where execution judgments are
    given by inference rules and termination corresponds to the
    existence of a finite derivation.

    Which is just incorrect. Since infinite derivation has meaning in
    the field.


    The halting problem is not undecidable because computation is weak,
    but because the classical formulation uses a denotational semantics
    that is too permissive.

    Nope.


    In operational/proof‑theoretic semantics, where meaning is grounded
    in finite derivations, the halting predicate is not a well‑formed
    judgment — just as unrestricted comprehension was not a well‑formed >>> judgment in naïve set theory.


    In other words, by trying to enforce your interpreation, you system
    becomes unworkable, as you can't tell if you can ask a question.


    It is the same ∀x ∈ T ((True(T, x) ≡ (T ⊢ x))
    that I have been talking about for years except that
    it is now grounded in well-founded proof‑theoretic
    semantics.

    Except that PA can't use that interpreation.

    The problem is you can't just change the sematics of the system and
    expect everything to stay the same.

    In fact, you are DEPENDING on things changing, but want to ignore all
    the changes that also happen that you don't want to look at.

    Go ahead, TRY to impose that semantics, and show what can be done in PA
    with it.

    Figure out how to reconcile the axiom of Induction with those semantics.

    and, Figure out how to reconcile the Axiom of Choice used in ZFC with
    that semantics as the simplest interpreations of how it works are truth-conditional.

    As I have been telling you for YEARS, if you want to change the
    foundation, go ahead, but you need to rebuild the building that you tore
    down. The problem is I don't think you understand the basics of the
    theories well enough to actually do it, as you can't just quote other
    papers, if you are actually breaking new ground.


    The problem is that systems like this grow faster in power to generate
    than your logic grow in power to decide, and either you accept that
    some truths are unprovable (and thus accept the truth-conditional
    view) or you need to just abandon the ability to actually work in the
    system as you can't tell what questions are reasonable.

    All you are doing is proving that you are just too stupid to
    understand the implications of what you are talking about, because you
    never really understood what the words actually mean.



    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Richard Damon@Richard@Damon-Family.org to sci.logic,comp.theory,sci.math,comp.ai.philosophy,comp.lang.prolog on Fri Jan 16 11:48:46 2026
    From Newsgroup: comp.ai.philosophy

    On 1/16/26 9:12 AM, olcott wrote:
    On 1/16/2026 3:17 AM, Mikko wrote:
    On 16/01/2026 01:38, olcott wrote:
    On 1/15/2026 3:48 AM, Mikko wrote:
    On 14/01/2026 19:28, olcott wrote:
    On 1/14/2026 1:40 AM, Mikko wrote:
    On 13/01/2026 16:27, olcott wrote:
    On 1/13/2026 3:11 AM, Mikko wrote:
    On 12/01/2026 16:29, olcott wrote:
    On 1/12/2026 4:44 AM, Mikko wrote:
    On 11/01/2026 16:18, olcott wrote:
    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string >>>>>>>>>>>>>>>>>>>>> inputs by finite string transformation rules into >>>>>>>>>>>>>>>>>>>>> {Accept, Reject} values.

    The counter-example input to requires more than >>>>>>>>>>>>>>>>>>>>> can be derived from finite string transformation >>>>>>>>>>>>>>>>>>>>> rules applied to this specific input thus the >>>>>>>>>>>>>>>>>>>>> Halting Problem requires too much.

    In a sense the halting problem asks too much: the >>>>>>>>>>>>>>>>>>>> problem is proven to
    be unsolvable. In another sense it asks too little: >>>>>>>>>>>>>>>>>>>> usually we want to
    know whether a method halts on every input, not just >>>>>>>>>>>>>>>>>>>> one.

    Although the halting problem is unsolvable, there >>>>>>>>>>>>>>>>>>>> are partial solutions
    to the halting problem. In particular, every >>>>>>>>>>>>>>>>>>>> counter- example to the
    full solution is correctly solved by some partial >>>>>>>>>>>>>>>>>>>> deciders.

    *if undecidability is correct then truth itself is >>>>>>>>>>>>>>>>>>> broken*

    Depends on whether the word "truth" is interpeted in >>>>>>>>>>>>>>>>>> the standard
    sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory >>>>>>>>>>>>>>>>> expressions are correctly rejected as semantically >>>>>>>>>>>>>>>>> incoherent thus form no undecidability or incompleteness. >>>>>>>>>>>>>>>>
    The misconception is yours. No expression in the >>>>>>>>>>>>>>>> language of the first
    order group theory is self-contradictory. But the first >>>>>>>>>>>>>>>> order goupr
    theory is incomplete: it is impossible to prove that AB >>>>>>>>>>>>>>>> = BA is true
    for every A and every B but it is also impossible to >>>>>>>>>>>>>>>> prove that AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string >>>>>>>>>>>>>>> inputs by finite string transformation rules into >>>>>>>>>>>>>>> {Accept, Reject} values.

    When a required result cannot be derived by applying >>>>>>>>>>>>>>> finite string transformation rules to actual finite >>>>>>>>>>>>>>> string inputs, then the required result exceeds the >>>>>>>>>>>>>>> scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be >>>>>>>>>>>>>> derived by
    appying a finite string transformation then the it it is >>>>>>>>>>>>>> uncomputable.

    Right. Outside the scope of computation. Requiring anything >>>>>>>>>>>>> outside the scope of computation is an incorrect requirement. >>>>>>>>>>>>
    You can't determine whether the required result is
    computable before
    you have the requirement.

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    Which only nmakes sense when the requirement that H must
    determine
    whether the computation presented by its input halts has already >>>>>>>>>> been presented.

    *ChatGPT explains how and why I am correct*

       *Reinterpretation of undecidability*
       The example of P and H demonstrates that what is
       often called “undecidable” is better understood as >>>>>>>>>>>    ill-posed with respect to computable semantics.
       When the specification is constrained to properties >>>>>>>>>>>    detectable via finite simulation and finite pattern >>>>>>>>>>>    recognition, computation proceeds normally and
       correctly. Undecidability only appears when the
       specification overreaches that boundary.

    It tries to explain but it does not prove.

    Its the same thing that I have been saying for years.
    It is not that a universal halt decider cannot exist.

    It is proven that an universal halt decider does not exist.

    “The system adopts Proof-Theoretic Semantics: meaning is
    determined by inferential role, and truth is internal to the
    theory. A theory T is defined by a finite set of stipulated
    atomic statements together with all expressions derivable from
    them under the inference rules. The statements belonging to T
    constitute its theorems, and these are exactly the statements
    that are true-in-T.”

    Under a system like the above rough draft all inputs
    having pathological self reference such as the halting
    problem counter-example input are simply rejected as
    non-well-founded. Tarski Undefinability, Gödel's
    incompleteness and the halting problem cease to exist.

    A Turing
    machine cannot determine the halting of all Turing machines and is >>>>>>>> therefore not an universla halt decider.

    This is not true in Proof Theoretic Semantics. I
    still have to refine my words. I may not have said
    that exactly correctly. The result is that in Proof
    Theoretic Semantics the counter-example is rejected
    as non-well-founded.

    That no Turing machine is a halt decider is a proven theorem and a >>>>>> truth about Turing machines. If your "Proof Thoeretic Semnatics"
    does not regard it as true then your "Proof Theoretic Semantics"
    is incomplete.


    My long‑term goal is to make ‘true on the basis of meaning’
    computable.

    As meaning is not computable, how can "true on the balsis of meaning"
    be commputable?

    Under *proof‑theoretic semantics*
    "true on the basis of meaning expressed in language"
    has always been entirely computable.

    Have you already put the algorithm to some web page?


    I am still working on refining the presentation.


    Which, based on your previousl work, is apt to take you 30 years or
    more, as you keep on needing to change to work around the flaws that
    people point out.

    But you don't actually fix the flaws, you just try to weasel word around
    them, as you don't actually know what you are talking about.
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Richard Damon@Richard@Damon-Family.org to sci.logic,comp.theory,sci.math,comp.ai.philosophy,comp.lang.prolog on Fri Jan 16 11:53:35 2026
    From Newsgroup: comp.ai.philosophy

    On 1/16/26 10:12 AM, olcott wrote:
    On 1/16/2026 3:17 AM, Mikko wrote:
    On 16/01/2026 01:38, olcott wrote:
    On 1/15/2026 3:48 AM, Mikko wrote:
    On 14/01/2026 19:28, olcott wrote:
    On 1/14/2026 1:40 AM, Mikko wrote:
    On 13/01/2026 16:27, olcott wrote:
    On 1/13/2026 3:11 AM, Mikko wrote:
    On 12/01/2026 16:29, olcott wrote:
    On 1/12/2026 4:44 AM, Mikko wrote:
    On 11/01/2026 16:18, olcott wrote:
    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string >>>>>>>>>>>>>>>>>>>>> inputs by finite string transformation rules into >>>>>>>>>>>>>>>>>>>>> {Accept, Reject} values.

    The counter-example input to requires more than >>>>>>>>>>>>>>>>>>>>> can be derived from finite string transformation >>>>>>>>>>>>>>>>>>>>> rules applied to this specific input thus the >>>>>>>>>>>>>>>>>>>>> Halting Problem requires too much.

    In a sense the halting problem asks too much: the >>>>>>>>>>>>>>>>>>>> problem is proven to
    be unsolvable. In another sense it asks too little: >>>>>>>>>>>>>>>>>>>> usually we want to
    know whether a method halts on every input, not just >>>>>>>>>>>>>>>>>>>> one.

    Although the halting problem is unsolvable, there >>>>>>>>>>>>>>>>>>>> are partial solutions
    to the halting problem. In particular, every >>>>>>>>>>>>>>>>>>>> counter- example to the
    full solution is correctly solved by some partial >>>>>>>>>>>>>>>>>>>> deciders.

    *if undecidability is correct then truth itself is >>>>>>>>>>>>>>>>>>> broken*

    Depends on whether the word "truth" is interpeted in >>>>>>>>>>>>>>>>>> the standard
    sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory >>>>>>>>>>>>>>>>> expressions are correctly rejected as semantically >>>>>>>>>>>>>>>>> incoherent thus form no undecidability or incompleteness. >>>>>>>>>>>>>>>>
    The misconception is yours. No expression in the >>>>>>>>>>>>>>>> language of the first
    order group theory is self-contradictory. But the first >>>>>>>>>>>>>>>> order goupr
    theory is incomplete: it is impossible to prove that AB >>>>>>>>>>>>>>>> = BA is true
    for every A and every B but it is also impossible to >>>>>>>>>>>>>>>> prove that AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string >>>>>>>>>>>>>>> inputs by finite string transformation rules into >>>>>>>>>>>>>>> {Accept, Reject} values.

    When a required result cannot be derived by applying >>>>>>>>>>>>>>> finite string transformation rules to actual finite >>>>>>>>>>>>>>> string inputs, then the required result exceeds the >>>>>>>>>>>>>>> scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be >>>>>>>>>>>>>> derived by
    appying a finite string transformation then the it it is >>>>>>>>>>>>>> uncomputable.

    Right. Outside the scope of computation. Requiring anything >>>>>>>>>>>>> outside the scope of computation is an incorrect requirement. >>>>>>>>>>>>
    You can't determine whether the required result is
    computable before
    you have the requirement.

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    Which only nmakes sense when the requirement that H must
    determine
    whether the computation presented by its input halts has already >>>>>>>>>> been presented.

    *ChatGPT explains how and why I am correct*

       *Reinterpretation of undecidability*
       The example of P and H demonstrates that what is
       often called “undecidable” is better understood as >>>>>>>>>>>    ill-posed with respect to computable semantics.
       When the specification is constrained to properties >>>>>>>>>>>    detectable via finite simulation and finite pattern >>>>>>>>>>>    recognition, computation proceeds normally and
       correctly. Undecidability only appears when the
       specification overreaches that boundary.

    It tries to explain but it does not prove.

    Its the same thing that I have been saying for years.
    It is not that a universal halt decider cannot exist.

    It is proven that an universal halt decider does not exist.

    “The system adopts Proof-Theoretic Semantics: meaning is
    determined by inferential role, and truth is internal to the
    theory. A theory T is defined by a finite set of stipulated
    atomic statements together with all expressions derivable from
    them under the inference rules. The statements belonging to T
    constitute its theorems, and these are exactly the statements
    that are true-in-T.”

    Under a system like the above rough draft all inputs
    having pathological self reference such as the halting
    problem counter-example input are simply rejected as
    non-well-founded. Tarski Undefinability, Gödel's
    incompleteness and the halting problem cease to exist.

    A Turing
    machine cannot determine the halting of all Turing machines and is >>>>>>>> therefore not an universla halt decider.

    This is not true in Proof Theoretic Semantics. I
    still have to refine my words. I may not have said
    that exactly correctly. The result is that in Proof
    Theoretic Semantics the counter-example is rejected
    as non-well-founded.

    That no Turing machine is a halt decider is a proven theorem and a >>>>>> truth about Turing machines. If your "Proof Thoeretic Semnatics"
    does not regard it as true then your "Proof Theoretic Semantics"
    is incomplete.


    My long‑term goal is to make ‘true on the basis of meaning’
    computable.

    As meaning is not computable, how can "true on the balsis of meaning"
    be commputable?

    Under *proof‑theoretic semantics*
    "true on the basis of meaning expressed in language"
    has always been entirely computable.

    Have you already put the algorithm to some web page?


    Proof Theoretic Semantics Blocks Pathological Self-Reference https://philpapers.org/rec/OLCPTS


    Which basically confuses Truth with Known.

    After all, by your definitions something starts out not being "Not-well-founded" if we haven't yet found a proof or refutation for it.
    But that status CHANGES if we discover one.

    This can only keep Truth Values consistant in a system with a finite
    fully enumerated set of possible proofs in it so we can know we have
    looked at all of them before calling something "Not-Well-Founded".

    I guess that is the only systems you are going to consider, ones that
    are that much of a TOY.

    That or you consider it acceptable that Truth changes.
    --- Synchronet 3.21b-Linux NewsLink 1.2
  • From Richard Damon@Richard@Damon-Family.org to sci.logic,comp.theory,sci.math,comp.ai.philosophy,comp.lang.prolog on Fri Jan 16 12:08:26 2026
    From Newsgroup: comp.ai.philosophy

    On 1/16/26 10:12 AM, olcott wrote:
    On 1/16/2026 3:17 AM, Mikko wrote:
    On 16/01/2026 01:38, olcott wrote:
    On 1/15/2026 3:48 AM, Mikko wrote:
    On 14/01/2026 19:28, olcott wrote:
    On 1/14/2026 1:40 AM, Mikko wrote:
    On 13/01/2026 16:27, olcott wrote:
    On 1/13/2026 3:11 AM, Mikko wrote:
    On 12/01/2026 16:29, olcott wrote:
    On 1/12/2026 4:44 AM, Mikko wrote:
    On 11/01/2026 16:18, olcott wrote:
    On 1/11/2026 4:13 AM, Mikko wrote:
    On 10/01/2026 17:47, olcott wrote:
    On 1/10/2026 2:23 AM, Mikko wrote:
    On 09/01/2026 17:52, olcott wrote:
    On 1/9/2026 3:59 AM, Mikko wrote:
    On 08/01/2026 16:22, olcott wrote:
    On 1/8/2026 4:22 AM, Mikko wrote:
    On 07/01/2026 13:54, olcott wrote:
    On 1/7/2026 5:49 AM, Mikko wrote:
    On 07/01/2026 06:44, olcott wrote:
    All deciders essentially: Transform finite string >>>>>>>>>>>>>>>>>>>>> inputs by finite string transformation rules into >>>>>>>>>>>>>>>>>>>>> {Accept, Reject} values.

    The counter-example input to requires more than >>>>>>>>>>>>>>>>>>>>> can be derived from finite string transformation >>>>>>>>>>>>>>>>>>>>> rules applied to this specific input thus the >>>>>>>>>>>>>>>>>>>>> Halting Problem requires too much.

    In a sense the halting problem asks too much: the >>>>>>>>>>>>>>>>>>>> problem is proven to
    be unsolvable. In another sense it asks too little: >>>>>>>>>>>>>>>>>>>> usually we want to
    know whether a method halts on every input, not just >>>>>>>>>>>>>>>>>>>> one.

    Although the halting problem is unsolvable, there >>>>>>>>>>>>>>>>>>>> are partial solutions
    to the halting problem. In particular, every >>>>>>>>>>>>>>>>>>>> counter- example to the
    full solution is correctly solved by some partial >>>>>>>>>>>>>>>>>>>> deciders.

    *if undecidability is correct then truth itself is >>>>>>>>>>>>>>>>>>> broken*

    Depends on whether the word "truth" is interpeted in >>>>>>>>>>>>>>>>>> the standard
    sense or in Olcott's sense.

    Undecidability is misconception. Self-contradictory >>>>>>>>>>>>>>>>> expressions are correctly rejected as semantically >>>>>>>>>>>>>>>>> incoherent thus form no undecidability or incompleteness. >>>>>>>>>>>>>>>>
    The misconception is yours. No expression in the >>>>>>>>>>>>>>>> language of the first
    order group theory is self-contradictory. But the first >>>>>>>>>>>>>>>> order goupr
    theory is incomplete: it is impossible to prove that AB >>>>>>>>>>>>>>>> = BA is true
    for every A and every B but it is also impossible to >>>>>>>>>>>>>>>> prove that AB = BA
    is false for some A and some B.


    All deciders essentially: Transform finite string >>>>>>>>>>>>>>> inputs by finite string transformation rules into >>>>>>>>>>>>>>> {Accept, Reject} values.

    When a required result cannot be derived by applying >>>>>>>>>>>>>>> finite string transformation rules to actual finite >>>>>>>>>>>>>>> string inputs, then the required result exceeds the >>>>>>>>>>>>>>> scope of computation and must be rejected as an
    incorrect requirement.

    No, that does not follow. If a required result cannot be >>>>>>>>>>>>>> derived by
    appying a finite string transformation then the it it is >>>>>>>>>>>>>> uncomputable.

    Right. Outside the scope of computation. Requiring anything >>>>>>>>>>>>> outside the scope of computation is an incorrect requirement. >>>>>>>>>>>>
    You can't determine whether the required result is
    computable before
    you have the requirement.

    *Computation and Undecidability*
    https://philpapers.org/go.pl?aid=OLCCAU

    We know that there does not exist any finite
    string transformations that H can apply to its
    input P to derive the halt status of any P
    that does the opposite of whatever H returns.

    Which only nmakes sense when the requirement that H must
    determine
    whether the computation presented by its input halts has already >>>>>>>>>> been presented.

    *ChatGPT explains how and why I am correct*

       *Reinterpretation of undecidability*
       The example of P and H demonstrates that what is
       often called “undecidable” is better understood as >>>>>>>>>>>    ill-posed with respect to computable semantics.
       When the specification is constrained to properties >>>>>>>>>>>    detectable via finite simulation and finite pattern >>>>>>>>>>>    recognition, computation proceeds normally and
       correctly. Undecidability only appears when the
       specification overreaches that boundary.

    It tries to explain but it does not prove.

    Its the same thing that I have been saying for years.
    It is not that a universal halt decider cannot exist.

    It is proven that an universal halt decider does not exist.

    “The system adopts Proof-Theoretic Semantics: meaning is
    determined by inferential role, and truth is internal to the
    theory. A theory T is defined by a finite set of stipulated
    atomic statements together with all expressions derivable from
    them under the inference rules. The statements belonging to T
    constitute its theorems, and these are exactly the statements
    that are true-in-T.”

    Under a system like the above rough draft all inputs
    having pathological self reference such as the halting
    problem counter-example input are simply rejected as
    non-well-founded. Tarski Undefinability, Gödel's
    incompleteness and the halting problem cease to exist.

    A Turing
    machine cannot determine the halting of all Turing machines and is >>>>>>>> therefore not an universla halt decider.

    This is not true in Proof Theoretic Semantics. I
    still have to refine my words. I may not have said
    that exactly correctly. The result is that in Proof
    Theoretic Semantics the counter-example is rejected
    as non-well-founded.

    That no Turing machine is a halt decider is a proven theorem and a >>>>>> truth about Turing machines. If your "Proof Thoeretic Semnatics"
    does not regard it as true then your "Proof Theoretic Semantics"
    is incomplete.


    My long‑term goal is to make ‘true on the basis of meaning’
    computable.

    As meaning is not computable, how can "true on the balsis of meaning"
    be commputable?

    Under *proof‑theoretic semantics*
    "true on the basis of meaning expressed in language"
    has always been entirely computable.

    Have you already put the algorithm to some web page?


    Proof Theoretic Semantics Blocks Pathological Self-Reference https://philpapers.org/rec/OLCPTS


    One more issue with this paper. You state:

    Some statements are neither true nor false in T. These are the non- well-founded statements: statements whose inferential justification
    cannot be grounded in a finite, well-founded proof structure.

    The existance of an inferential justification would be a fact that
    requires full examination of possible cases, so is effectively truth-condtional. Trying to use a proof-theoretic meaning requires to
    first do the exhaustive search before being able to apply that meaning,
    which for most system is an uncomputable task.

    This "breaks" your system in that truth can't actually be defined in the system, as the truth of a statement isn't an invariant but can change
    based on knowledge derived in the system.
    --- Synchronet 3.21b-Linux NewsLink 1.2