• The correct foundation of the theory of computation

    From olcott@polcott333@gmail.com to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Sat Dec 13 10:44:06 2025
    From Newsgroup: comp.ai.philosophy

    At the most fundamental level:
    All Turing machines only compute the mapping
    from input finite strings to some value.

    Turing machine Deciders are a subset of this
    where the value indicates accept or reject a
    finite string by some criterion measure.

    A further elaboration of this comes from the
    notion of computable functions and Rice's
    Theorem:

    Turing machine deciders compute functions from
    *finite string inputs* to {accept, reject} values
    according to whether the input has a syntactic
    property or specifies a semantic property.

    Turing machine deciders only report on the behavior
    of Turing machines indirectly through the proxy of
    finite string inputs. *This key detail has been ignored*
    --
    Copyright 2025 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 3.21a-Linux NewsLink 1.2
  • From Richard Damon@Richard@Damon-Family.org to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Sat Dec 13 14:14:06 2025
    From Newsgroup: comp.ai.philosophy

    On 12/13/25 11:44 AM, olcott wrote:
    At the most fundamental level:
    All Turing machines only compute the mapping
    from input finite strings to some value.

    Turing machine Deciders are a subset of this
    where the value indicates accept or reject a
    finite string by some criterion measure.

    A further elaboration of this comes from the
    notion of computable functions and Rice's
    Theorem:

    Turing machine deciders compute functions from
    *finite string inputs* to {accept, reject} values
    according to whether the input has a syntactic
    property or specifies a semantic property.

    Turing machine deciders only report on the behavior
    of Turing machines indirectly through the proxy of
    finite string inputs. *This key detail has been ignored*


    So, if the function isn't "Computable", it is still a Function, but no
    Turing Machine (or equivalent) can compute it.

    Nothing says that the Halting Function is computable, and in fact, the
    Halting Problem proof says it is not.

    This doesn't make it "invalid", just not computable.

    Thus, there can not exist an actual Halt Decider, as Halting as defined
    just isn't a computable function.

    This also doesn't mean that the string can't have the semantic meaning
    of the behavior of the Machine it describes, we just find that Semantic Properties tend not to be Computable, as Turing Machines are powerful
    enough to break any Decider trying to decide on an actual Semantic Property.

    This is just like an any logic system with sufficient power (roughly
    able to express the poperties of the Natural Numbers) will have
    statements that are true in them, but are not provable.

    There is a isomophism between Computability and Provability.
    --- Synchronet 3.21a-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 Sat Dec 13 19:33:40 2025
    From Newsgroup: comp.ai.philosophy

    On 13/12/2025 16:44, olcott wrote:
    At the most fundamental level:
    All Turing machines only compute the mapping
    from input finite strings to some value.

    Turing machine Deciders are a subset of this
    where the value indicates accept or reject a
    finite string by some criterion measure.

    I continue to reject the use of "accept" and "reject" here. And I also
    reject the use of "indicates" wrt to them.

    There's no good reason to suppose this is about grammars and strings
    rather than propositions about programs. To the extent that they might
    be isomorphic you need to introduce the notion of grammars and string acceptance by them to give "accept" and "reject" and "indicates" the
    correct meaning.
    --
    Tristan Wibberley

    The message body is Copyright (C) 2025 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 3.21a-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Sat Dec 13 13:50:57 2025
    From Newsgroup: comp.ai.philosophy

    On 12/13/2025 1:33 PM, Tristan Wibberley wrote:
    On 13/12/2025 16:44, olcott wrote:
    At the most fundamental level:
    All Turing machines only compute the mapping
    from input finite strings to some value.

    Turing machine Deciders are a subset of this
    where the value indicates accept or reject a
    finite string by some criterion measure.

    I continue to reject the use of "accept" and "reject" here. And I also
    reject the use of "indicates" wrt to them.


    My goal is to have accepted definitions as my only basis.

    There's no good reason to suppose this is about grammars and strings
    rather than propositions about programs. To the extent that they might
    be isomorphic you need to introduce the notion of grammars and string acceptance by them to give "accept" and "reject" and "indicates" the
    correct meaning.



    This is a grammatically correct English sentence:
    "flpm erf09-25k (*&j^*&NJ*&jkNef", reject.
    --
    Copyright 2025 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 3.21a-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Sat Dec 13 14:13:40 2025
    From Newsgroup: comp.ai.philosophy

    On 12/13/2025 1:14 PM, Richard Damon wrote:
    On 12/13/25 11:44 AM, olcott wrote:
    At the most fundamental level:
    All Turing machines only compute the mapping
    from input finite strings to some value.

    Turing machine Deciders are a subset of this
    where the value indicates accept or reject a
    finite string by some criterion measure.

    A further elaboration of this comes from the
    notion of computable functions and Rice's
    Theorem:

    Turing machine deciders compute functions from
    *finite string inputs* to {accept, reject} values
    according to whether the input has a syntactic
    property or specifies a semantic property.

    Turing machine deciders only report on the behavior
    of Turing machines indirectly through the proxy of
    finite string inputs. *This key detail has been ignored*


    So, if the function isn't "Computable", it is still a Function, but no Turing Machine (or equivalent) can compute it.


    There seems to be no such thing as uncomputable
    functions when Turing machines are construed
    entirely in terms of finite string rewrite rules.
    --
    Copyright 2025 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 3.21a-Linux NewsLink 1.2
  • From Richard Damon@Richard@Damon-Family.org to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Sat Dec 13 15:26:11 2025
    From Newsgroup: comp.ai.philosophy

    On 12/13/25 3:13 PM, olcott wrote:
    On 12/13/2025 1:14 PM, Richard Damon wrote:
    On 12/13/25 11:44 AM, olcott wrote:
    At the most fundamental level:
    All Turing machines only compute the mapping
    from input finite strings to some value.

    Turing machine Deciders are a subset of this
    where the value indicates accept or reject a
    finite string by some criterion measure.

    A further elaboration of this comes from the
    notion of computable functions and Rice's
    Theorem:

    Turing machine deciders compute functions from
    *finite string inputs* to {accept, reject} values
    according to whether the input has a syntactic
    property or specifies a semantic property.

    Turing machine deciders only report on the behavior
    of Turing machines indirectly through the proxy of
    finite string inputs. *This key detail has been ignored*


    So, if the function isn't "Computable", it is still a Function, but no
    Turing Machine (or equivalent) can compute it.


    There seems to be no such thing as uncomputable
    functions when Turing machines are construed
    entirely in terms of finite string rewrite rules.


    Sure there are, Your problem is you have the ideas reversed.

    Not all functions need to be generatable by Turing Machine.

    Those that can not be, are uncomputable.

    Since not all Functions, that is generic mappings of the complete set of
    one domain, to anther, can be implemented in Turing Machines, not all
    are computable.

    If you think there doesn't exist an uncomputable Function, show me how
    you compute the Halting Function, being the mapping of a Turing Machine
    (with its input) to whether it halts or not.

    Since you have tried to claim that it is impossible to even ask this
    question of a Decider, it must be non-computable.

    But of course, you get a lot of things backwards since you chose to work
    with zero-principle knowledge.
    --- Synchronet 3.21a-Linux NewsLink 1.2
  • From Mikko@mikko.levanto@iki.fi to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Sun Dec 14 12:43:47 2025
    From Newsgroup: comp.ai.philosophy

    On 13/12/2025 18:44, olcott wrote:
    At the most fundamental level:
    All Turing machines only compute the mapping
    from input finite strings to some value.

    Turing machine Deciders are a subset of this
    where the value indicates accept or reject a
    finite string by some criterion measure.

    A further elaboration of this comes from the
    notion of computable functions and Rice's
    Theorem:

    Turing machine deciders compute functions from
    *finite string inputs* to {accept, reject} values
    according to whether the input has a syntactic
    property or specifies a semantic property.

    Every property a Turing machine can compute is a syntactic property.
    A Turing machine sees only its input. It does not see any neaning of
    the input. Therefore it has no access to semantic properties. A
    semantic property can be computed only if it is equivalent to a
    syntactic property.
    --
    Mikko
    --- Synchronet 3.21a-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 Sun Dec 14 12:51:33 2025
    From Newsgroup: comp.ai.philosophy

    On 13/12/2025 19:50, olcott wrote:
    On 12/13/2025 1:33 PM, Tristan Wibberley wrote:
    On 13/12/2025 16:44, olcott wrote:

    Turing machine Deciders are a subset of this
    where the value indicates accept or reject a
    finite string by some criterion measure.

    I continue to reject the use of "accept" and "reject" here. And I also
    reject the use of "indicates" wrt to them.


    My goal is to have accepted definitions as my only basis.

    Oh! I just noticed it's a new statement with "by some criterion measure"
    which makes it excellent. I retract my rejection.


    [snip]
    This is a grammatically correct English sentence:
    "flpm erf09-25k (*&j^*&NJ*&jkNef", reject.


    That's impertinent.
    --
    Tristan Wibberley

    The message body is Copyright (C) 2025 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 3.21a-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to comp.theory,comp.ai.philosophy on Sun Dec 14 15:28:07 2025
    From Newsgroup: comp.ai.philosophy

    On 12/14/2025 3:00 PM, Tristan Wibberley wrote:
    On 14/12/2025 16:16, polcott wrote:

    This is my first principle
    All Turing machines only compute the mapping
    from input finite strings to some value.

    I maintain that's wrong. it should be "compute only".


    All Turing machines compute only the mapping
    from input finite strings to some value.

    Computable functions are the basic objects of study
    in computability theory. Informally, a function is
    computable if there is an algorithm that computes
    the value of the function for every value of its argument. https://en.wikipedia.org/wiki/Computable_function

    Now we tie that to a model of computation:
    Turing machines computable functions only compute
    the mapping from input finite strings to some value.

    *I might just use this as the new first principle*
    Turing machine Deciders compute the mapping from
    input finite strings to an accept or reject value
    by some criterion measure.

    You previously said it means the same either way around but that's not
    true for "only" even if it's true for many adverbs because of a
    syntactic ambiguity due to the many parts of speech "only" can fulfil.

    [only compute] [the <something>]
    [compute] [only the <something>]

    the first can actually be wrong if nonterminating programs (ie, keep operating without making progress producing a result) are not said to compute, which I'm told they are not. Those turing machines do not
    compute so it can't be that compute is the only thing they do.


    Your correction is correct.

    That the only thing they compute is <something> could be true depending
    on the <something>, which is another matter.


    --
    Copyright 2025 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 3.21a-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Sun Dec 14 17:46:09 2025
    From Newsgroup: comp.ai.philosophy

    On 12/14/2025 4:43 AM, Mikko wrote:
    On 13/12/2025 18:44, olcott wrote:
    At the most fundamental level:
    All Turing machines only compute the mapping
    from input finite strings to some value.

    Turing machine Deciders are a subset of this
    where the value indicates accept or reject a
    finite string by some criterion measure.

    A further elaboration of this comes from the
    notion of computable functions and Rice's
    Theorem:

    Turing machine deciders compute functions from
    *finite string inputs* to {accept, reject} values
    according to whether the input has a syntactic
    property or specifies a semantic property.

    Every property a Turing machine can compute is a syntactic property.
    A Turing machine sees only its input. It does not see any neaning of
    the input. Therefore it has no access to semantic properties. A
    semantic property can be computed only if it is equivalent to a
    syntactic property.


    If that was true then Rice's theorem would be
    wrong before it even got started.

    In computability theory, Rice's theorem states
    that all non-trivial semantic properties of
    programs are undecidable. A semantic property
    is one about the program's behavior https://en.wikipedia.org/wiki/Rice%27s_theorem
    --
    Copyright 2025 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 3.21a-Linux NewsLink 1.2
  • From Mikko@mikko.levanto@iki.fi to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Mon Dec 15 12:15:31 2025
    From Newsgroup: comp.ai.philosophy

    On 15/12/2025 01:46, olcott wrote:
    On 12/14/2025 4:43 AM, Mikko wrote:
    On 13/12/2025 18:44, olcott wrote:
    At the most fundamental level:
    All Turing machines only compute the mapping
    from input finite strings to some value.

    Turing machine Deciders are a subset of this
    where the value indicates accept or reject a
    finite string by some criterion measure.

    A further elaboration of this comes from the
    notion of computable functions and Rice's
    Theorem:

    Turing machine deciders compute functions from
    *finite string inputs* to {accept, reject} values
    according to whether the input has a syntactic
    property or specifies a semantic property.

    Every property a Turing machine can compute is a syntactic property.
    A Turing machine sees only its input. It does not see any neaning of
    the input. Therefore it has no access to semantic properties. A
    semantic property can be computed only if it is equivalent to a
    syntactic property.

    If that was true then Rice's theorem would be
    wrong before it even got started.

    No, the Rice's theorem does not imply that the Rice's theorem be wrong.

    In computability theory, Rice's theorem states
    that all non-trivial semantic properties of
    programs are undecidable. A semantic property
    is one about the program's behavior https://en.wikipedia.org/wiki/Rice%27s_theorem

    Where non-trivial means 'not equivalent to a syntactic property'.
    --
    Mikko
    --- Synchronet 3.21a-Linux NewsLink 1.2
  • From olcott@polcott333@gmail.com to comp.theory,sci.logic,sci.math,comp.ai.philosophy on Mon Dec 15 10:13:58 2025
    From Newsgroup: comp.ai.philosophy

    On 12/15/2025 2:09 AM, Tristan Wibberley wrote:
    The message body is Copyright (C) 2025 Tristan Wibberley except
    citations and quotations noted. All Rights Reserved except as noted in
    the sig.

    On 14/12/2025 16:16, polcott wrote:
    On 12/14/2025 6:51 AM, Tristan Wibberley wrote:
    On 13/12/2025 19:50, olcott wrote:
    On 12/13/2025 1:33 PM, Tristan Wibberley wrote:
    On 13/12/2025 16:44, olcott wrote:

    Turing machine Deciders are a subset of this
    where the value indicates accept or reject a
    finite string by some criterion measure.

    I continue to reject the use of "accept" and "reject" here. And I also >>>>> reject the use of "indicates" wrt to them.


    My goal is to have accepted definitions as my only basis.

    Oh! I just noticed it's a new statement with "by some criterion measure" >>> which makes it excellent. I retract my rejection.


    Good.

    This is my first post on the halting Problem
    https://groups.google.com/g/sci.logic/c/V7wzVvx8IMw/m/ggPE6a-60cUJ

    I worked for 15 years mostly on the basis of intuition.
    Then 2 more years creating fully operational code. Then
    3 years of discussing this code.

    Now I am finally getting around to anchoring these
    intuitions and my working code in standard definitions.

    My current working code.
    https://github.com/plolcott/x86utm/blob/master/Halt7.c

    I had to refrain from learning the standard definitions
    before now or they would have boxed me into the standard
    views.

    My insights are entirely from slight nuances of meaning
    that are abstracted away in the standard definitions.

    I had to carefully reverse-engineer the exact details
    of what was actually happening before I could see what
    nuances of meaning were being left out. Initially I
    had to use my own non-standard terminology to do this.

    This is my first principle
    All Turing machines only compute the mapping
    from input finite strings to some value.

    My second correction of three that are needed (I will return to the
    third later once I've thought more):

    "value" is not defined, a first principle must be elementary (introduces terms of art but does not depend on them).

    Here are two alternatives to illuminate the matter, but I think they're
    not good enough:

    1st alternative: You need a prior principle defining "value". That's not entirely intuitive though we pretend it is, like so much that you
    correctly find to be a problem.

    2nd alternative: All Automatic Turing Machines compute only an output
    finite string from an input finite string.

    They're wrong because a turing machine has a state (m-configuration),
    whose change--and given a constant initial state then also itself--is,
    in effect, computed.

    The 3rd alternative is contingent as noted, a contingency whose premise
    I think you suppose:

    3rd alternative, admissible when considering an ATM to be a physical
    machine rather than a mere formal system: An Automatic Turing Machine is
    one of those things restricted--at least in part--such that

    it computes
    nothing more than an output finite string along with its own halting
    state from an input finite string along with its own pre-nominated
    initial state.



    A TM halt decider computes the halt status specified
    by an input finite string on its tape. It begins in its
    own start state and ends in one of its its own final
    halt states.

    To say that a TM halt decider determines whether or
    not machine M halts on input w is less than precisely
    accurate.
    --
    Copyright 2025 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 3.21a-Linux NewsLink 1.2