Chess will never be solved, here's why - Chess Forums - Page 553

TumoKonnin

Jul 19, 2024

0

#11041

Optimissed hat geschrieben:

DiogenesDue wrote:

Lol. When have I ever hidden behind anyone?

As for who can discuss topics in a clear and forthright manner, I'll let less delusional posters be the judge of that. You are a trainwreck. Look at the sheer amount of contortion and rationalization you do to make one simple reply...and still it's bollocks in the end.

Always. You like to know that you'll be backed up. You prefer odds of at least three to one.

Just observation. If you think my analysis of psychopathy is wrong (it was straight off the top of my head so it might be) then find a psychoanalyst and if s/he has any objections to it, ask 'em to be clear about them, instead of doing what you do, so that their objection may be questioned and a learning process may take place.

bro we're talking about chess what does this have to do with it

TumoKonnin

Jul 19, 2024

0

#11042

Optimissed hat geschrieben:

DiogenesDue wrote:

Lol. When have I ever hidden behind anyone?

As for who can discuss topics in a clear and forthright manner, I'll let less delusional posters be the judge of that. You are a trainwreck. Look at the sheer amount of contortion and rationalization you do to make one simple reply...and still it's bollocks in the end.

Always. You like to know that you'll be backed up. You prefer odds of at least three to one.

Just observation. If you think my analysis of psychopathy is wrong (it was straight off the top of my head so it might be) then find a psychoanalyst and if s/he has any objections to it, ask 'em to be clear about them, instead of doing what you do, so that their objection may be questioned and a learning process may take place.

quit yapping and actually answer the question

TumoKonnin

Jul 19, 2024

0

#11043

Optimissed hat geschrieben:

DiogenesDue wrote:

TumoKonnin wrote:

sure buddy keep yapping. why don't we see you make a chess engine that has "good programming"?

Careful, you are surely destined for imaginary cartel membership if you continue.

He's just a pillock, not a cartel member, He's too polite for your cartel. If his aeroplane stopped working mid Atlantic, instead of getting annoyed at the engineer/designer, of course he'd realise that he should have made the aircraft himself. Then he could accept his fate with equanimity.

....

bro what

go to an asylum

DiogenesDue

Jul 19, 2024

0

#11044

Optimissed wrote:

DiogenesDue wrote:

TumoKonnin wrote:

sure buddy keep yapping. why don't we see you make a chess engine that has "good programming"?

Careful, you are surely destined for imaginary cartel membership if you continue.

He's just a pillock, not a cartel member, He's too polite for your cartel. If his aeroplane stopped working mid Atlantic, instead of getting annoyed at the engineer/designer, of course he'd realise that he should have made the aircraft himself. Then he could accept his fate with equanimity.

Why do you always have to over reach the mark? It's cringeworthy.

TumoKonnin

Jul 19, 2024

0

#11045

Optimissed hat geschrieben:

Actually I think you have a new friend there. Same level of intellect as you and you might have even more in common than that.

more intellect then you

TumoKonnin

Jul 19, 2024

0

#11046

Optimissed hat geschrieben:

DiogenesDue wrote:

Optimissed wrote:

DiogenesDue wrote:

TumoKonnin wrote:

sure buddy keep yapping. why don't we see you make a chess engine that has "good programming"?

Careful, you are surely destined for imaginary cartel membership if you continue.

He's just a pillock, not a cartel member, He's too polite for your cartel. If his aeroplane stopped working mid Atlantic, instead of getting annoyed at the engineer/designer, of course he'd realise that he should have made the aircraft himself. Then he could accept his fate with equanimity.

Why do you always have to over reach the mark? It's cringeworthy.

I won the point so what more do you want? A bunch of flowers?

how did you win the point exactly? that's what i'm saying, you're delusional.

you say that he thinks he's right all the time. but look at your message.

"i won the point"

what a hypocrite.

Elroch

Jul 19, 2024

0

#11047

Optimissed wrote:

A good engine would know it. It would need to be programmed with a long, linear search which occasionally kicked in if there were no alteration to position assessment.

What's a long linear search meant to be? Just playing the top choice over and over again? If this finds it gets back to a known position, obviously the engine needs to look at alternative moves rather than giving up.

Modern engines already do very deep narrow searches, with a branching factor typically below 2, because this is found to provide the highest performance in practical play. This compares for example with Deeper Blue, which was much weaker than necessary in practical play because it searched too broadly. In a blocked position, engines will effectively gradually broaden the search as early tries get nowhere. But unless there are very few pieces, some positions will be quite deep.

Just bad programming really. Even the normal search should tell it if it were programmed right because it would see itself repeating moves and that the position evaluation, assessed correctly, doesn't alter.

It is normal for precise play to see the positional evaluation varying little. The alternative is the exception. But even if you are comparing blocked positions with minor rearrangements of the pieces, the evals are likely to vary slightly.

Note that I couldn't find the quoted post anywhere in the last day or so - perhaps it has been deleted? (I used @DiogenesDue's quote of the post).

TumoKonnin

Jul 19, 2024

0

#11048

Elroch hat geschrieben:

Optimissed wrote:

A good engine would know it. It would need to be programmed with a long, linear search which occasionally kicked in if there were no alteration to position assessment.

What's a long linear search meant to be? Just playing the top choice over and over again? If this finds it gets back to a known position, obviously the engine needs to look at alternative moves rather than giving uo.

Modern engines already do very deep narrow searches, with a branching factor typically below 2, because this is found to provide the highest performance in practical play. This compares for example with Deeper Blue, which was much weaker than necessary in practical play because it searched too broadly. In a blocked position, engines will effectively gradually broaden the search as early tries get nowhere. But unless there are very few pieces, some positions will be quite deep.

Just bad programming really. Even the normal search should tell it if it were programmed right because it would see itself repeating moves and that the position evaluation, assessed correctly, doesn't alter.

It is normal for precise play to see the positional evaluation varying little. The alternative is the exception. But even if you are comparing blocked positions with minor rearrangements of the pieces, the evals are likely to vary slightly.

Note that I couldn't find the quoted post anywhere in the last day or so - perhaps it has been deleted? (I used @DiogenesDue's quote of the post).

it is deleted

Thee_Ghostess_Lola

Jul 19, 2024

0

#11049

in the most objective of s/o's leftside ?...is still flawed design. and trust me its on purpose. i ended up reaching existential dread a long time ago. so i told myself to embrace it. and i got there. i learned to peaceably tolerate stuff unknown (start w/ infinity, a lover, real chinese food...).

accept that its gonna be awhile before this game gets wholly described. i have.

summa u needta go where i went. i can tell. once there ?...ull sleep w/a smile on your face . until then ?...ur getting sucked out to sea with a lifevest on.

TumoKonnin

Jul 19, 2024

0

#11050

Optimissed hat geschrieben:

TumoKonnin wrote:

Optimissed hat geschrieben:

Actually I think you have a new friend there. Same level of intellect as you and you might have even more in common than that.

more intellect then you

I defer to your perception.

Hi there.

hello there.

Elroch

Jul 19, 2024

0

#11051

Optimissed wrote:

The glitch is that it's been programmed to see its own assessments as correct .... you would say "objectively correct" and therefore it sees the minor variations as an indication that it ISN'T a draw. Bad programming.

It neither sees constant variations as a draw nor minor variations as not a draw. It sees both as a equal position, like the initial one.

It's worth observing that Leela evaluates positions in a different way that distinguishes more between equal positions and drawn positions.

Doves-cove

Jul 19, 2024

0

#11052

fr bro

Elroch

Jul 19, 2024

0

#11053

Would be nice if Leela could figure out blocked positions, but the problem is that the neural network is trained on competitive chess and just doesn't understand them!

Here (click to enlarge) you will see it concluding white has little chance of winning and less of drawing! The percentage at the start of each line is the expected score, the WDL at the end is what it thinks the chance of different outcomes is. Don't ask me why!

Doves-cove

Jul 19, 2024

0

#11054

😂

Elroch

Jul 19, 2024

0

#11055

So the two strongest engines in the world haven't got a clue in this position!

A bit more exploration without allowing much analysis time finds that it tries to sacrifice a rook (harmless enough) with black, white (unwisely) accepts it, and the evaluation gets nearer 50%

Doves-cove

Jul 19, 2024

0

#11056

Elroch wrote:

So the two strongest engines in the world haven't got a clue in this position!

bro of course not they're just bots. lol

DiogenesDue

Jul 19, 2024

0

#11057

Optimissed wrote:

The glitch is that it's been programmed to see its own assessments as correct .... you would say "objectively correct" and therefore it sees the minor variations as an indication that it ISN'T a draw. Bad programming.

Engines are doing what they are programmed to do. A glitch would imply some kind of bug. Would you say it was a "glitch" that fossil fuel cars don't uniformly achieve 100mpg or better?

Of course an engine sees its evaluation as correct enough to make a move. If it were not programmed when to pull the trigger and consider its eval the optimal achievable for the time control, it would think forever, like a human that runs their clock out by second guessing themselves.

One interesting question is whether machine learning engines were originally left to their own devices in terms of deciding optimal "think" time for a given move, or if they were given a framework to start with. Hopefully the former, humans being as fallible as they are.

TumoKonnin

Jul 19, 2024

0

#11058

Optimissed hat geschrieben:

TumoKonnin wrote:

Optimissed hat geschrieben:

TumoKonnin wrote:

Optimissed hat geschrieben:

Actually I think you have a new friend there. Same level of intellect as you and you might have even more in common than that.

more intellect then you

I defer to your perception.

Hi there.

hello there.

This is a bit of a funny thread because there are various underlying rivalries/tensions. Please join in if you think it's worthwhile. No-one really owns this thread. The moderators occasionally get involved but so long as there's no religion or politics they seem to let stuff go. I have to do some work. Pleased to have met you.

fr.

pleased to meet you too

Prixaxelator

Jul 19, 2024

0

#11059

hmm

DiogenesDue

Jul 19, 2024

0

#11060

Optimissed wrote:

Ah so you're agreeing with me that it's due to bad programming. I'm not sure but I think that think times were flexible. I've played chess on some really ancient playing code from the late 70s up to the late 80s and you could use the Stonewall Attack to beat the engine every time since the horizon was inflexible ... about nine moves .... and all you had to do was make a sacrifice which achieved a winning position in ten moves. You could repeat that ad infinitum to impress your friends. No machine learning then.

It's still a glitch, imo. Think back to the early 1980s, when perfectly good code could cause a glitch. That is, you'd done it properly but it didn't work, so you altered it a little and it worked. Conversely, you found a short cut that wasn't "officially" part of the code which, if you used it, speeded things up and made the code more compact. It's just that if you went to work somewhere else, the person coming in to take over from you wouldn't have understood your programming, since it was non-standard.

As to your question about cars, my old 2 litre Vauxhall that's been parked in our back yard for three years now could do 70 mpg up an incline in the right conditions, at maybe around 63 mph.

It's not bad programming, nor a glitch. You really should stay far away from computer-related topics.