Re: What it would take... People to address my points with reasoning instead of rhetoric

Op 13.mei.2025 om 23:49 schreef olcott:

On 5/13/2025 4:28 PM, Richard Heathfield wrote:

On 13/05/2025 22:16, olcott wrote:

On 5/13/2025 12:06 PM, Richard Heathfield wrote:

On 13/05/2025 17:21, dbush wrote:

On 5/13/2025 12:01 PM, olcott wrote:

>
<snip>
>

The actual reasoning why HHH is supposed to report
on the behavior of the direct execution of DD()
instead of the actual behavior that the finite
string of DD specifies:

>
Quite simply, it's the behavior of the direct execution that we want to know about.

>
Why?
>
DDD doesn't do anything interesting.
>
If it were a universal halt decider we'd have a reason to care, because its very existence would overturn pretty much the whole of computability theory and enable us to clean up many of the unsolved problems of mathematics.
>

>
Sure and we could achieve the same thing by
simply hard-coding the actual all-knowing
mind of God into a formal system.

>
No, we couldn't.
>

The question is not about any universal halt
decider

>
Yes, it is.
>

that must be literally all knowing.

>
That sounds like hyperbole, but it's actually not far off. It could, at least, be used as an oracle; you'd just need to find a way to express your question as a YNA program.
>

It has always actually only been about things
that could prevent consistently determining
the halt status of conventional programs.

>
No, it's all about demonstrating that some computational problems can't be solved. The whole halting thing is just a vehicle that can be used as an example of an undecidable computation.
>

 In cannot include things that humanity has no knowledge of
such as the Goldbach's conjecture and the meaning of life.
 The time is quickly coming when AI will be thousands-fold
smarter than the smartest human. Such as AI might figure
out the Goldbach's conjecture.
 

But it /isn't/ a universal halt decider, so who (apart from Mr Olcott) gives a damn whether it stops? About the only reason I can think of for caring is to set Mr Olcott straight, but he has made it abundantly clear that he's unsettable straightable.
>

>
There is no time that we are ever going to directly
encode omniscience into a computer program.

>
Right.
>

The
screwy idea of a universal halt decider that is
literally ALL KNOWING is just a screwy idea.

>
There's nothing screwy about proving that such a program can't be written.
>

 Requiring it to be ALL KNOWING was always a little nuts.
Requiring it to get fooled on fewer and fewer inputs
is the rational goal.
 HHH does compute the mapping from its input
finite string to the behavior that this finite
string specifies and this includes HHH emulating
itself emulating DDD.

No, HHH computes the wrong mapping by skipping the most important part of the input, the part that specifies the conditional abort. Due to a bug, HHH aborts before it sees that part of the specification of the input.