Jupp53 wrote on 11/16/13 at 13:30:45:

Yes, English has two similar words for it; effective and efficient.

I once saw the difference explained as "effective is doing the right thing; efficient is doing the thing right"! 
In other words, "effective" is about whether something works at all; "efficient" is about whether it works anywhere near optimally, especially time-wise.

@whitecraw: I agree. The real point is to identify strengths and weaknesses and use that to plan training work and maybe opening choices.

But the rating part of this is a fun extra; it can be motivating to put a number on one's different skills just like the Khmelnitsky books do. You could argue the entire multiple choice/rating structure is a trick to fool oneself into actually working on chess! After all, solving positions against a clock to simulate tournament play is universally agreed to be a great training method. Now I did learn something about my strengths and weaknesses in line with Khmelnitsky's categories, but also some things that didn't fit neatly. For example, my calculations were generally too chaotic and inefficient, which had an effect on many different categories.

@Jupp53:
Very nice, thank you! I can see myself getting by with just these deviation scores. For example I could categorize each position into one or more decision types (with hindsight) and position types, and create average deviation statistics for each type.

@Stigma
I think the real benefit I got from Khmelnitsky book was to show the "relative" strengths and weaknesses in my play, and to point to where I should try and lead play (to my strengths) and train (weaknesses). Even Khmelnitsky himself says that at the back of the book.

Maybe that is really the answer, ie look for relative differences in how you handle different types of positions rather than an absolute rating (which will never be truly accurate anyway). This could be achieved using Chessbase or SCID to help you determine performance ratings for different types of positions, pawn structures etc. Chessbase has tactical, strategy and endgame keys, as well as opening keys that could be used to filter positions and determine performance ratings. I have thought about doing this to get an objective measure of how well I intuitively play certain middle-game positions using data from the 3000 or so ICC blitz games that I have played.

whitecraw wrote on 11/14/13 at 20:45:03:


Thinking about this for some time I came to the personal conclusion it's not worth the work.

Regan has more interest in optimizing the mathematical background of his work.

For normal club players it's possible to set a standard engine (I took stockfish 2.2 for that) and check the tactical blunders via setting a delta between the engine move an the players move in SCID. Then you come to the expectable ( an for many disappointing) result, that your rating is not so  wrong if you compare this with data from persons with a similiar rating.

Then you can either start discussing standard errors of a rating group and individuals or you start the chess work about your data. Why did you blunder at that point? What was the reason for the lack of understanding of that specific position? This will lead you out of your comfort zone. Compared to working through a book of tactical positions I don't know which method leads to better results.

But maybe it's to far from the topic as it's asking for having such a program. 

You can build your own data very easy. Take a big open tournament. Autoanalyze all games. Build a spreadsheet with average deviation per move and rating of the players. Compare this with your last ten otb-games. Then you will get a good estimate.

For me the biggest obstacle is the coding to sort the data. Doing it manually is inacceptable these days.

@whitecraw Simple yes, but also time-consuming. Who wants to spend their precious chess time playing against a computer?

This "assessment" has even less to do with practical human vs human strength if you start from a different starting position. Suddenly knowledge of opening theory and early middlegame themes become irrelevant, while in real games of course they are not.

I can't add much to this discussion. I obviously fell in love with Smyslov's games and analysis early on. And I agree completely that Taimanov's games and analysis are excellent resources for improving players!

I just picked up the 2-volume Smyslov collection from Moravian Chess and 10 games in I'm really enjoying it. His annotations are right in my sweet spot, fairly light but occasionally making me stretch my visualizing and calculating ability, and somehow his style and openings are very simpatico with the way I like to approach chess (as opposed to Botvinnik, say, where I often think "Nice play, but I would be unlikely to get into a position like this").

Stigma wrote on 11/13/13 at 05:32:31:


That would be akin to "garbage time" in football, which different metrics define differently.  One way of defining garbage time is to say OK, what if one team had scored a touchdown and a 2-point conversion on every one of his final possessions?  Would he have won?  If not, the game was in garbage time (out of reach and hence "technically won") at that point.  There are other ways of defining garbage time as well.  

Still, I think it's harder with chess.  My point was sometimes humans know that a position is a win or draw far ahead of a computer.  So yes, don't penalize him for suboptimal moves then, but you still need a human to come along and make that determination!  It's perfectly feasible on a small scale; I'm talking about for large-scale stuff.

Stigma wrote on 11/13/13 at 05:32:31:


I really liked Smyslov's game collection, but it was the first I ever read and others find it dry, so maybe I just have fond memories.  Alekhine's not so much; the games are incredible, but when you look at modern analysis (Dvoretsky, Kasparov), it seems clear that Alekhine is lieing to you about what he saw in the game, trying to brag and make you think he saw everything.  Quite often in his games, it wasn't necessary to see to the "end" of a combination to play it, and yet Alekhine claimed to have seen it all; I remember Dvoretsky doubting that claim.  I always felt like Alekhine was lieing to us a bit.  I felt the same way about Nimzovich!  I never had that feeling from Smyslov or Taimanov or Tal, for instance.

Taimanov's game collection was the most enjoyable to me, for whatever reason.  I found it really instructive, especially the Nimzo and Sicilian games.