Sujet : Re: Bad faith and dishonesty
De : polcott333 (at) *nospam* gmail.com (olcott)
Groupes : comp.theoryDate : 30. May 2025, 17:06:36
Autres entêtes
Organisation : A noiseless patient Spider
Message-ID : <101cl2d$hfof$11@dont-email.me>
References : 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20
User-Agent : Mozilla Thunderbird
On 5/30/2025 10:41 AM, Richard Heathfield wrote:
On 30/05/2025 16:29, Mike Terry wrote:
@Richard: so you cannot make HHH decide non-halting simply by looping for a long long time, hoping HHH will get fed up! That would just result in HHH simulating for a corresponding long long time. You need to feed it a program that halts, but matches one of his non-halting behaviour tests. For example DDD.
What if I don't know whether it halts?
I followed up to vallor with a pseudocode sketch of such a program.
If HHH could deliver a reliably correct report for that program within a year or so, that would probably be enough to earn Mr Olcott a place in the history books.
int DD()
{
int Halt_Status = HHH(DD);
if (Halt_Status)
HERE: goto HERE;
return Halt_Status;
}
_DD()
[00002162] 55 push ebp
[00002163] 8bec mov ebp,esp
[00002165] 51 push ecx
[00002166] 6862210000 push 00002162
[0000216b] e862f4ffff call 000015d2
[00002170] 83c404 add esp,+04
[00002173] 8945fc mov [ebp-04],eax
[00002176] 837dfc00 cmp dword [ebp-04],+00
[0000217a] 7402 jz 0000217e
[0000217c] ebfe jmp 0000217c
[0000217e] 8b45fc mov eax,[ebp-04]
[00002181] 8be5 mov esp,ebp
[00002183] 5d pop ebp
[00002184] c3 ret
Size in bytes:(0035) [00002184]
The fact that DD correctly emulated by HHH cannot
possibly stop running unless aborted and cannot
possibly reach its own "ret" instruction final halt
state conclusively proves that the input to HHH(DD)
specifies a non-halting sequence of configurations.
But at some point we have to place a ceiling on "long long time". A reporting program that keeps saying "maybe next year" isn't much of a reporting program.
As soon as HHH sees that DD is calling HHH(DD) twice
in sequence without any conditional branch instructions
inbetween the beginning of DD and its call to HHH(DD)
Then HHH has proof that its input specifies a non-halting
sequence of configurations.
As long as an input finite string cannot possibly reach
its own "ret" instruction final halt state then we know
that it can be rejected as non-halting.
-- Copyright 2025 Olcott "Talent hits a target no one else can hit; Geniushits a target no one else can see." Arthur Schopenhauer