Stockfish Engines

[menniepals]

I have two questions. First, What Stockfish version would you consider the most aggressive? Second, What version you think would be more positional? Thanks.

[MOBMAT]

I would have to say that none of the official releases of Stockfish are specifically designed to be "aggressive", and that term is somewhat subjective.

Many have cloned Stockfish to solve technical positions, such as puzzles and mate finding more efficiently, but those clones aren't as well suited to play over the board type games.

So, if you want a well rounded (gee, Stockfish has been at the top of the lists of best engines for years) engine, just use the most recent version of Stockfish. If you want to solve puzzles or mates, look for a clone or another engine.

What is your intent?

V

[menniepals]

I am observing that the later versions of Stockfish are becoming more and more positional, and of course, all are tactically strong. I thought starting from SF 13 to the present, plays a more solid and positional style. Whereas, SF 12 and versions before it, were more tactically motivated. That is just my observation.

[menniepals]

I love going over games and using SF to find positional or tactical lines.

[Eelco de Groot]

I am observing that the later versions of Stockfish are becoming more and more positional, and of course, all are tactically strong. I thought starting from SF 13 to the present, plays a more solid and positional style. Whereas, SF 12 and versions before it, were more tactically motivated. That is just my observation.

You may very well be right! Just going by what others have said on the forum recently, I think it was Eduard Nemeth, Stockfish's playing style has become more positional because of using more Lco games to train its neural net. Like Alpha Zero before it, positional chess is its strength. Before the neural net (NNUE) introduction in Stockfish, the evaluation was all handtuned and material still played a huge role in that despite all efforts to bring more positional elements into it. Now maybe it is time to try bring back some more Paul Morphy into it, if not the romantic, but not always (or mostly not) totally sound chess, of Adolf Anderssen. Meaning, maybe more specuIative chess in the case of the romantics? I don't know much about those old chessgames of Morphy, it has been a long time that I played myself actively and I was really not any good at tactics.

[CornfedForever]

I am observing that the later versions of Stockfish are becoming more and more positional, and of course, all are tactically strong. I thought starting from SF 13 to the present, plays a more solid and positional style. Whereas, SF 12 and versions before it, were more tactically motivated. That is just my observation.

You may very well be right! Just going by what others have said on the forum recently, I think it was Eduard Nemeth, Stockfish's playing style has become more positional because of using more Lco games to train its neural net. Like Alpha Zero before it, positional chess is its strength. Before the neural net (NNUE) introduction in Stockfish, the evaluation was all handtuned and material still played a huge role in that despite all efforts to bring more positional elements into it. Now maybe it is time to try bring back some more Paul Morphy into it, if not the romantic, but not always (or mostly not) totally sound chess, of Adolf Anderssen. Meaning, maybe more specuIative chess in the case of the romantics? I don't know much about those old chessgames of Morphy, it has been a long time that I played myself actively and I was really not any good at tactics.

I think that 'tactical style' generally equates to 'less precise' and therefore more subject to loss.

If you could tune to 'tactical' as you wish and the engine STILL play at the level it is, you might have something...but 'tactics' don't normally occur on their own and are either the result of mistakes or weak play by the 'other guy'. If the other guy is a similarly strong engine...good luck, because I don't see it happening. The current versions do are generally perfectly good at finding tactics...where they really exist.

[menniepals]

SF 12 and below seems to press forward hard and unleash tactics when the opponent makes a mistake. SF 13 and above, strong tactically, but plays safer, solid, somewhat plays more human and positional.

[abgursu]

I am observing that the later versions of Stockfish are becoming more and more positional, and of course, all are tactically strong. I thought starting from SF 13 to the present, plays a more solid and positional style. Whereas, SF 12 and versions before it, were more tactically motivated. That is just my observation.

You may very well be right! Just going by what others have said on the forum recently, I think it was Eduard Nemeth, Stockfish's playing style has become more positional because of using more Lco games to train its neural net. Like Alpha Zero before it, positional chess is its strength. Before the neural net (NNUE) introduction in Stockfish, the evaluation was all handtuned and material still played a huge role in that despite all efforts to bring more positional elements into it. Now maybe it is time to try bring back some more Paul Morphy into it, if not the romantic, but not always (or mostly not) totally sound chess, of Adolf Anderssen. Meaning, maybe more specuIative chess in the case of the romantics? I don't know much about those old chessgames of Morphy, it has been a long time that I played myself actively and I was really not any good at tactics.

I think that 'tactical style' generally equates to 'less precise' and therefore more subject to loss.

If you could tune to 'tactical' as you wish and the engine STILL play at the level it is, you might have something...but 'tactics' don't normally occur on their own and are either the result of mistakes or weak play by the 'other guy'. If the other guy is a similarly strong engine...good luck, because I don't see it happening. The current versions do are generally perfectly good at finding tactics...where they really exist.

You're halfway right!
You're right about almost everything you said, but current Stockfishes miss sometimes, due to time control actually. There was an engine, Kayra 1.5 Attack. It was the SF clone Kayra with some changes from Eduard and it's style was indeed, attacking. I loved it and still use it for my opening preferations actually, but it was weak against Stockfishes. I started a 100-gamed 3'+2'' marathon and it lost something like +36 =60 -4 for Stockfish and I analyzed those 4 Kayra wins. One of them was a game with positional sacrifices that Sf sees but misevaluates and the other three were simple sacs whom SF actually sees in analysis but missed in that time control. Those games wasn't the thing that amazed me, it was actually the fact that how sf missed those tactics and attacks. So what I'm saying is, you can't focus on tactics and make an engine far better than Stockfish but even Stockfish sometimes misses tactics.

[M ANSARI]

Interesting observation but I think SF main progress has been because it has become more "positional" and thus plays more accurately. I think a tactical engine is much more of a nightmare against humans rather than against other stronger engines. When compared to engines, humans play much stronger positionally than they do tactically. I would guess that the gap between the strongest human vs engine positional play is maybe 500 to 800 ELO less than an engine ... while in complicated tactical position the gap is probably double that or more. I personally really like how SF plays now as it is much easier to understand. Of course strong positional play has to have tactical awareness as positional play without tactical awareness is useless.

[menniepals]

The overall strength of the later SF is very strong but the slow positional grind it plays allow humans to survive a few more moves than the previous ones. It seems like these versions play around with positional ideas before swooping down on its victims. Maybe this style of play is annoying for other engines to play against. Playing against the insanely strong tactical monsters sf 12 and previous versions did not allow humans that possibility. It was guaranteed crush within 20 moves. I thought that was very exciting.