lichess.org
Donate

Why You Shouldn't Always Trust The Engine

QxQ is not only the safest, but also the fastest win, in terms of practical game duration. 90+% of the time, your opponent will resign in response.
@Periastron said in #21:
> QxQ is not only the safest, but also the fastest win, in terms of practical game duration. 90+% of the time, your opponent will resign in response.

If White has 5 seconds, Black 60 seconds, no increment,
then after QxQ White would probably loose on time.
(Look at White's rooks which are placed very clumsy for an attack. )

It is not that difficult to see that after a Queen check either the rook falls with check
(and the Queen will fall afterwards - pin or skewer)
or at least one White rook will enter the attack with check
which should enable White to take off both black pieces and ensure the draw - even in five seconds.

If you look at the five best variants with an engine, maybe you will find out this motif
even if you have missed it in the game.

I would conclude "Don't trust the engine but take its suggestions serious."
Queen check is not good because of Matt in seven but because a rook enters the attack.
@Chris_7591 said in #7:
> "... they hunt down the rest of the enemies’ pieces, as if the one with the most material would get an extra point."
> _________________________________________________________
> I sometimes hunt down my opponents pawns if they only have a minor piece + king and a few pawns; This has to do with time pressure; for example: I have a clearly winning position but I'm very low on time and do not see a forced mate.
> I do see a way to capture all the opponents pawns - even sacrificing a piece to do it -- which guarantees me 0.5 point.
> .
> Sometimes there is still time for the checkmate - if the opponent is trying to dirty-flag me.
> Thanks - good post!
@SurroundSoundDad said in #8:
> While I def agree with the overall idea of the post it is worth noting a few things. 1. if you see the mate in 7 (I know that isn't me) you should take it because M7 has much less room for error than +7. That said if you have this fuzzy feeling that you have a strong attack and nothing concrete planned out then the +7 is the better option for you specifically. I feel that this becomes more and more true the better the player you are. At the ultra grandmaster levels of chess it really comes down to who attains the real attack first without blundering material by the end of it and that means sometimes stepping into minefields based on a fuzzy feeling. Personally what has helped me best is not looking directly at the computer evaluation but taking a look at it and then following up until I understand fully what the computer wanted. sometimes that takes quite a while because I am not a great player so I will look at the computers best line and I will systematically try any other moves that are NOT the computer line and see how the computer challenges each one of those and in the process I feel that my understanding of the position in question is far superior to just looking at the evaluation. it takes a lot of honesty with yourself to not blindly agree.
>
> Again, overall I agree with the main point but I really wanted to bring to attention that you should really dig into the computer evaluation to get the most and try to learn what makes an attack work or not work. in a tournament you absolutely should always play for the highest win rate but practicing online or casual otb I think it is far better to focus on learning and exploring those quasi attacks that you think are good but are not sure.
I agree with this Article . But you said that you saw people having Rooks and Queens up like there is no Tomorrow , the reason that people (such as me) do this is that we like to see our opponent trying to find a move but actually there is no move for them , and honestly it is fun to get the opponents' pieces (at least this reason is for me).
@IbrahimOre13_STJ said in #2:
> some people can analyze their games and end up with the surprise that the machine makes such a move and you don't know why, but as always, even machines can make mistakes

I think a mistake by the machine is almost impossible, it is just if I can't understand why a "better move" is really better, playing it would be a mistake for me (even if the move is still better).
@Chris_7591 said in #7:
> "... they hunt down the rest of the enemies’ pieces, as if the one with the most material would get an extra point."
> _________________________________________________________
> I sometimes hunt down my opponents pawns if they only have a minor piece + king and a few pawns; This has to do with time pressure; for example: I have a clearly winning position but I'm very low on time and do not see a forced mate.
> I do see a way to capture all the opponents pawns - even sacrificing a piece to do it -- which guarantees me 0.5 point.
> .
> Sometimes there is still time for the checkmate - if the opponent is trying to dirty-flag me.
> Thanks - good post!

Or just play with an increment, and none of this will happen.