Sunday, September 24, 2017 15:56

Posts Tagged ‘Analysis Theory’

The Limits of Computer Analysis… or Human Analysis?

Monday, January 4th, 2010

On Saturday I posted about an adventure I had in the Graz Defense. Chris Morgan and I played our game. When we reached move 9. Be3, the position looked like this:

After 9. Be3

After 9. Be3.

I briefly looked at my alternatives, but had pretty much committed to 9. …Bxe3 when I saw the Bishop would be coming at me with the d3 pawn move.

I ended up losing that game, but got the privilege of running over the game afterwards with Chris and Joe Alford. (Analysis with Joe is quite a treat – he’s about the strongest player in Bloomington, if not the strongest.)

The conclusion of Joe and Chris was that this was pretty much the penultimate strategic moment. All of us agreed that the Bishop trade didn’t look like the best move. Among other factors with the recapture White gets to have the f file opened for him, and the Black Queen’s post on f6 is even less secure than it was.

There are viable alternatives for Black (some of which I didn’t really explore as fully as I should have.) It’s a simple enough question: Take or defend. Defensive moves involving my Bishop included 9. …Bb6, 9. …Ba7, 9. …Bd6, 9. …d6. We pretty much decided that was the priority, as well.

So I come home, upload the game to my database and start looking at it. Although I believe in the soundness of our analysis, I called up my analysis engine on the position. And it eventually chose as it’s first line…. Bxe3!?

I went ahead and plugged in all of the candidate moves, and used all three chess engines I’ve installed to this computer to analyze the position to between 12 and 14 ply. Almost all the engines agreed on 9. …Bxe3. Here’s a table showing the various candidate moves and their positional evaluations:

Human or Computer: Which is right? Responses to 9. Be3
Computer responses are postition evaluation/search ply depth. Closer to zero in position evaluation is better for Black.
Move Crafty 20.14Fritz 6Rybka 2.2n2Human RankdB Game Notes
Bxe3.12/13.28/13.14/13Last/Worst
Bd6.71/13.34/12.39/13Second Choice
Bb6.46/13.50/14.25/14First ChoiceNote 1
Ba7.41/15.25/13.20/13Alternate/Equal to Bb6Note 2
d6.50/13.34/11.19/12Near prime, tied second maybe.Note 3
Note 1: ½-½ Pena Gomez,M-Espineira Gonzalez,P/Poio 2001
Note 2: 1-0 Judd,M-Harding,H/Cleveland 1871
Note 3: ½-½ Makarova,O-Kozhamberdin,B/Orsk 2000

Note: The full game, along with Rybka 2.2’s analysis of that position and some of my other notes can be found here.

Now, to be sure, Fritz 6 in the table above finds 9. …Ba7 just micro-fractionally the best choice. Yet the one game my database yielded to me shows a win for White after 9. …Ba7, too.

There are other limits here as well: Note that we are talking about ultra-fractional points of a pawn in advantage. (.14 = 14 hundrethds of a Pawn in equivalent material advantage. OK, in all fairness the numerical evaluation doesn’t have to equal the same points we would use in calculating material advantages. But it is close, and Fritz’s difference was .03, and Rybka’s difference between first and worst is .25 – that’s small!) If the programs were forced to make the move at the point I stopped their analysis, taking the Bishop would have been their route.

What to make of this?

On the one hand, this could be the stuff that a Grandmaster might take seriously. (GM wants a draw in a particular round in a tournament, so finds this line where a human playing White would presume Black will not exchange. Yet by making the Bishop trade Black manages to pull out a draw – though it would take much more analysis than I’ve given to one move to reach that conclusion.)

But on the other hand, it is equally one of those times when you recognize your own limits as a player. Which means it’s time to switch the computer off and switch on your brain. I still agree with my fellow humans: I should have played one of the alternate moves there.

(And, for potential nit-pickers, yes the advice might be: Don’t play the Graz Defense! ;) )

When you play OTB, it won’t be Fritz, Rybka, or Crafty playing your opponent. It will be you. If you ever reach this position as Black, do you take or do you defend? This is one of those times when you choose the line that will best help you….

….Enjoy your Chess!

Thinking about Computer Analysis

Saturday, December 26th, 2009

I got a number of wonderful gifts for Christmas. Among them eNotate and I just ordered Fritz 12 from the USCF Sales outlet on Amazon! Yay!

Update late morning 12/26: Tomorrow’s blog post will detail some of my troubles – it turns out I may or may not be able to use the Fritz 12 I ordered. :(

The rest of this post may not be quite oriented to the beginning player – we’ll have to see! I am discussing theoretical options in regards to computer postmortem analysis for weaker players below the cut.

(more…)