XPost: comp.theory, sci.logic, sci.math
From: 643-408-1753@kylheku.com
On 2025-10-21, olcott wrote:
> On 10/20/2025 9:11 PM, Kaz Kylheku wrote:
>> On 2025-10-21, olcott wrote:
>>> On 10/20/2025 8:27 PM, Kaz Kylheku wrote:
>>>> On 2025-10-20, olcott wrote:
>>>>> On 10/20/2025 4:03 PM, Kaz Kylheku wrote:
>>>>>> On 2025-10-20, olcott wrote:
>>>>>>> On 10/20/2025 1:29 PM, Kaz Kylheku wrote:
>>>>>>>> On 2025-10-20, olcott wrote:
>>>>>>>>> On 10/19/2025 2:39 PM, Kaz Kylheku wrote:
>>>>>>>>>> On 2025-10-19, dart200 wrote:
>>>>>>>>>>> i don't get y polcott keep hanging onto ai for dear life. anyone
with
>>>>>>>>>>
>>>>>>>>>> Throngs of dumb boomers are falling for AI generated videos,
believing
>>>>>>>>>> them to be real. This is much the same thing.
>>>>>>>>>>
>>>>>>>>>> AI is just another thing Olcott has no understanding of. He's not
>>>>>>>>>> researched the fundamentals of what it means to train a language
>>>>>>>>>> network, and how it is ultimately just token prediction.
>>>>>>>>>>
>>>>>>>>>> It excels at generating good syntax. The reason for that is that the
>>>>>>>>>> vast amount of training data exhibits good syntax. (Where it has bad
>>>>>>>>>> syntax, it is idiosyncratic; whereas good syntax is broadly shared.)
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>> I provide a basis to it and it does perform valid
>>>>>>>>> semantic logical entailment on this basis and shows
>>>>>>>>
>>>>>>>> But you're incapable of recognizing valid entailment from invalid.
>>>>>>>>
>>>>>>>
>>>>>>> Any freaking idiot can spew out baseless rhetoric
>>>>>>> such as this. I could do the same sort of thing
>>>>>>> and say you are wrong and stupidly wrong.
>>>>>>
>>>>>> But you don't?
>>>>>>
>>>>>>> It is a whole other ballgame when one attempts
>>>>>>> to point out actual errors that are not anchored
>>>>>>> in one's own lack of comprehension.
>>>>>>
>>>>>> You don't comprehend the pointing-out.
>>>>>>
>>>>>
>>>>> You need to have a sound reasoning basis to prove
>>>>> that an error is an actual error.
>>>>
>>>> No; /YOU/ need to have sound reasonings to prove /YOUR/
>>>> extraordinary claims. The burden is on you.
>>>>
>>>> We already have the solid reasoning which says things are other than as
>>>> you say, and you don't have the faintest idea how to put a dent in it.
>>>>
>>>
>>> In other words you assume that I must be wrong
>>> entirely on the basis that what I say does not
>>> conform to conventional wisdom.
>>
>> Yes; you are wrong entirely on the basis that what you say does not
>> follow a valid mode of inference for refuting an argument.
>>
>> If you are trying to refute something which is not only a widely
>> accepted result, but whose reasoning anyone can follow to see it
>> for themselves, you are automatically assumed wrong.
>>
>> The established result is presumed correct, pending your
>> presentation of a convincing argument.
>>
>> That's not just wanton arbitrariness: your claims are being
>> directly refuted by elements of the established result which
>> we can refer to.
>>
>> I cannot identify any flaw in the halting theorem. It's not simply
>> that I believe it because of the Big Names attached to it.
>>
>
> And when I identify a flaw yo simply ignore
> whatever I say.
Nope; all the ways you say claim you've identified a flaw have been
dissected by multiple poeple to a much greater detail than they deserve.
It is disingenuous to say that you've simply had your details ignored.
>> I'm convinced by the argumentation; and that conviction has
>> the side effect of convincing me of the falsehood of your
>> ineffective, contrary argumentation.
>>
>
> Not really it actually gives you the bias to refuse
> to pay attention.
LOL! the world at large is incredibly biased against giving a crank
like you any attention.
Those of us present are overcoming the world's /immense/ bias
and actually indulging the details of your claims.
>
>>> That is not any actual rebuttal of the specific points that I make.
>>
>> No, indeed /that/ isn't; but plenty of those have also been made not
>> only by me but various others, over a considerable time span.
>>
>
> Never any actual rebuttal ever since Professor
> Sipser agreed with my words.
You're forgetting (of course, isnce it was more than 48-72 hours
ago) that I (almost) also agree with those words.
> Those exact same
> words still form the basis of my whole proof.
They don't do that, though.
> *Here is an accurate paraphrase of those words*
>
>
>
> Perform the complete simulation specified in main() and provide a
> full execution trace before doing anything else.
>
> Simulating Termination Analyzer HHH correctly simulates its input until:
Yes; it correctly steps each x86 instructions with Debug_Step until:
> (a) It correctly detects that its simulated input cannot
> possibly reach its own simulated final halt state then:
It correctly detects the situation that: if it doesn't abort,
the simulation will not end.
What this means is that if, hypothetically, HHH //were differently
defined// as a non-aborting decider, then DD //would also be
differntly defined// as a non-terminating case.
This hypothesis doesn't mean fuck all because it's not reality. HHH is
not differently defined other than as it is, and likewise DD is not
differently defined. HHH is required to report on the current definition
of DD, which is built on the current definition of HHH.
In any case, yes; the abort is necessary to avoid non-termination.
> abort simulation and return 0 rejecting its input as non-halting.
And yes, I agree that it aborts the simulation, returns 0
which indicates that it's rejecting the input as non-halting.
(Unfortunately, that is wrong).
But mostly the words can be rationally agreed to with the caveat that
HHH's result may not be interpreted to be about a hypothetical different
version of itself acting on a different input.
HHH must be reporting about the actual instruction string DD
that it is actually given. (As you like to repeat.) Not some fantasy
other versions of these.
--
TXR Programming Language: http://nongnu.org/txr
Cygnal: Cygwin Native Application Library: http://kylheku.com/cygnal
Mastodon: @Kazinator@mstdn.ca
--- SoupGate-Win32 v1.05
* Origin: you cannot sedate... all the things you hate (1:229/2)
