mwyoung wrote: ↑
Wed Oct 14, 2020 1:35 am
maac wrote: ↑
Wed Oct 14, 2020 1:15 am
mwyoung wrote: ↑
Wed Oct 14, 2020 12:12 am
I said "Really, then why does TCEC need bias openings? Lc0 was taught to play chess for this unbias position.
That we all know and love!"
Not at all.First, rule thumb, never say (anyone, not personal) "always" or "all",
second, "biased" positions are the bread and butter of chess any day, so, one of the
main functions of engines is the eval of such positions, case in point, as a stubborn
Scandi player i have studied certain positions where the score after opening is in
the vicinity of +1.0 for white, guess what, thanks to latest SF i have found the
refutations after many hours of (cyclic) analysis.
Overall, to all: why simply doesn't recognize the greatness of SF (and LC0 btw) ?
We are not talking about a weak player like you. Who is playing a weak opening to win against a weak player.
That is total B.S.
What are you talking about. Then yes when using garbage openings then use Stockfish. Then I highly recommend Stockfish. That is why I voted for Stockfish winning TCEC.
"as a stubborn Scandi player"-----EXACTLY!!!!
Use what ever engine that works, both engines are the best. But that is not the point of proper engine testing. And you clearly don't force inferior openings on a player, that does not play for example the Scandi, because they know it sucks.
And I did notice you did not give the position that Stockfish saved your position. What was the position, and what was Lc0 score. +1 after many hours of search, but you found a refutation with only Stockfish.
Why would you even bother to keep playing such line to find only a way to draw that you thought was a clear loss. That is what you get from the starting position. I know "a stubborn Scandi player".
Ok, first B01 doesnt suck! By incredible that it sounds, and i know i am against rhe stream,
but there are some myths in chess still, the lines that i found equalzed are
1) white plays 5. Ne5 Cbd7 6.f4 g6 6.7. Bc4 e6 8. a4 and
2) 5. Ne5 Be6 6. Af4 Qd8 7. Qd2 both in the 3. ...Qd6 scandinavian.
But it isn't the point and you would be know, the point is 'perfect" openings with about +- 0.20
are less common that those with say us +0.50 after the opening,in that sense TCEC is very well
And, bias or not bias, the conditions are the same (duh) for both engines, so, if engine
B is not enough trained in that openings (hypothesis..) that engine A, the fault of who is!?