tygxc why arent you addressing the fact that we've proven you wrong again and again?
do you get some sort of sick kick from spreading misinformation?
tygxc's 'strongest' position so far ... as in 'least weak' is something that he just mentioned but has been saying for years.
Like this:
"1 e4 e5 2 Ba6? loses for white, without needing any game tree."
Now I would say that's reasonable.
It's a very reasonable guess. But if you think it is a strictly CERTAIN conclusion, you are being unreasonable (in a way many people are).
White simply 'drops his bishop' for a compensation that is just too much less.
Right so, we are using our positional evaluations as human players to conclude there is no need to analyse that line further (let's face it, none of us have bothered to look further).(black then has doubled isolated a-pawns)
An example of a crude positional factor used as part of our positional evaluation.So I would say that the computer not bothering to analyze further from there - is reasonable 'approximate solving'.
Reasonable from the point of view of a human player (of what strength).
A term unfortunately coined is 'weakly solving'.
Nothing unfortunate about it. It's a label, perfectly understood by everyone who seriously uses it (or who can read a wikipedia article).After bxa6 there ... black is not 'weak'. He's strong. A bishop up.
Irrelevant conflation. And it's as if you don't know that when you are constructing a weak strategy as black for chess that reaches the position 1. e4 e5, you need to demonstrate a forced draw (or win) for black from the position 1. e4 e5 2. Ba6. Not just guess one. You need first to pick a move for black (easy enough, capture the bishop the way you want - you can use the knight if you don't like doubled pawns) then deal with the 25 possible moves (if I have counted correctly in my head) for white on move 3.--------------------------
but tygxc is using the logic of that one argument - to build illogic.
For the computer to keep dismissing 'further game tree' on any line because its materially up -
isn't legitimate because 'compensation' varies and it isn't easy to assign definitions of what is 'enough compensation' or 'more than enough compensation' ...
but tygxc would then be likely to use 'computer evaluation' ... like have the computer 'dismiss' if its evaluation is for more than two point advantage for one side.
Yes. This is junk. There is no doubt that you could find a million bishop sacrifices it would get wrong. Probably the easiest way would be to have them reaching a tablebase position that it gets wrong.But he doesn't get it that that's a circular argument that fails to take into account that the computer is fallible and hasn't 'solved' chess in the first place and can't do so.
Exactly.He's arguing that the computer could validate itself and is right because it says so.
Does tygxc believe in his own false argument or is he being deceitful?
Has tygxc fallen into a pitfall because in tactics puzzles - the puzzle doesn't continue beyond the solution moves? Has tygxc 'fallen' because players Resign in lost positions?
-------------------------
Experience and observation over decades tells me that such things aren't A or B.
Whoever - if invested enough - stops caring much whether they think they believe their own nonsense or not.
So they're not even aware either way.
Is there another specific instance of this in tygxc's spiels?
Yes.
It becomes more apparent when he asserts 'We don't Care about the number of operations per second that supercomputers can do - we only Care about Nodes per second!'
That is basically revealing about tygxc's internal illogic.
He doesn't care because he doesn't know.
He doesn't know because he doesn't want to know.
---------------------------------
You'll find this in many crass denial of logic or denial of science or denial of reality situations around the world.
Do flat-earthers know they're pushing nonsense or they think they aren't?
They stopped caring long before - as to which - and they stopped being internally aware of same. And don't want to be.
Result - nobody can talk to them - about them.
And the result is projection. Which always involves falsehood by the person so doing.
'No! Its You! And MSM! Main Stream Media. The world is Flat!'
Let me make an observation. Stockfish is the strongest chess player in the world. It is so based on just three essential things:
- its positional evaluation routine
- its ability to prioritise search of variations guided by this
- doing this very fast
Up to quite recently, many years were spent to handcraft a positional evaluation routine, with many contributors improving it, testing the improvements and iterating this process.
Then all this was replaced by the training of one neural network from scratch. This single change - ignoring all the human input and previous testing - increased Stockfish rating by a massive amount, around 100 points, by far the biggest step since it had become world class.
Using this neutral network and its algorithm for searching, it does not ignore positions like 1. e4 e5 2.Ba6. Rather it gives them a low priority. When they are near the start of analysis it will definitely analyse a lot of other more sensible looking stuff first, but when a lot of good lines have been quite well analysed, it will use a little CPU time to look a bit deeper at 2. Ba6. Just in case. It leaves no stone unturned (it just doesn't turn really bad-looking stones until it has had a good look at all the better looking stones).
The "reasonable" choice to ignore 2.Ba6 completely either shows that Stockfish is getting it wrong and needs some advice from us humans or that it is only approximately reasonable and is actually pedantically misguided. If you want to know which is true, it's probably best to think of the comparison between many years of collaborated development of a positional evaluation routine and leaving the neutral network do it itself.
When somebody lies - but they internally choose to believe their own lie and then maintain in that way - are they still then lying?The short answer is yes.
Is anybody exempt from this?
i'd say that nobody is exempt from this, but our intellectual worth is determined by our effort to be true in our knowledge.If you tell yourself that something is true that isn't - are you lying?
Some would say 'if whoever was given false data to work with - then that could be an instance about being honest in telling a falsehood'.
That's the medium answer.
---------------------------------
Could a computer be assigned to simply analyze all lines of chess and stop analyzing any lines where the advantage for one side has reached +5 ?
Could chess be 'approximately solved' that way?
Or to have a chance - would the 'stopping advantage' have to be as little as 0.5 ?
That would be ridiculous.
Stockfish could 'solve' chess right now if the stopping advantage only had to be 0.1
It would Solve chess in seconds. Already has.
those advantages can only really be useful in pruning one side for a "solve" so im not entirely sure.