Your network blocks the Lichess assets!

lichess.org
Donate

AlphaZero's Mind-Blowing Chess Strategy: Double Pawn Sacrifice and Queen Dance

Thank you for revisiting those games. And yes, if we trusted engine top tournament ELOs to give us the ultimate chess analysis, before those NN came into play, what is the difference now? Should we always trust engine tournament top performer to be having a complete view of chess. If it had blind "sub-spaces" before and managed to patch that breach enough for the same tournaments formats where it is not clear what is being optimized by increasing ELOs in those pools, how do we know if such top engine does not have still such blind sub-spaces, or holes, holes that can be whole sub-trees of the complete tree, not only pockets, we just don't know, but we could try to know. Characterize what they are doing careful independently of ELO measures. Or put analytic constraints on some types.. while playing with other types.. aim being chess insight, not toy performance.

Also since you are looking at many such games. I am glad that you put them in studies.. So those us of that are slow, or don't have much time.. and have study plan laid out using those study and your blog for future work.

now 3 blogs with that approach.. And you are developping the theme of installations. across those chosen games. So that would be my target notion to consider in some future plan of study, based on games.

I think one take home from studying those games, is not necessarily about best chess discrepancies. but that 2 foreign design engines can create enough human visible and perceivable chess insight, than only one could.

This should transfer to having at least a 2 type analytical tools for positions analysis with engines.

Thank you for revisiting those games. And yes, if we trusted engine top tournament ELOs to give us the ultimate chess analysis, before those NN came into play, what is the difference now? Should we always trust engine tournament top performer to be having a complete view of chess. If it had blind "sub-spaces" before and managed to patch that breach enough for the same tournaments formats where it is not clear what is being optimized by increasing ELOs in those pools, how do we know if such top engine does not have still such blind sub-spaces, or holes, holes that can be whole sub-trees of the complete tree, not only pockets, we just don't know, but we could try to know. Characterize what they are doing careful independently of ELO measures. Or put analytic constraints on some types.. while playing with other types.. aim being chess insight, not toy performance. Also since you are looking at many such games. I am glad that you put them in studies.. So those us of that are slow, or don't have much time.. and have study plan laid out using those study and your blog for future work. now 3 blogs with that approach.. And you are developping the theme of installations. across those chosen games. So that would be my target notion to consider in some future plan of study, based on games. I think one take home from studying those games, is not necessarily about best chess discrepancies. but that 2 foreign design engines can create enough human visible and perceivable chess insight, than only one could. This should transfer to having at least a 2 type analytical tools for positions analysis with engines.