Chess Tempo

Username:
Password:
/ Register

User Details

Username:
Blitz Rating:
Standard Rating:
Logout
August 21, 2008, 08:27:22 pm *
Welcome, Guest. Please login or register.
News: SMF - Just Installed!
 
Pages: [1]
Print
Author Topic: Problem 34479  (Read 340 times)
szaszkin
Newbie
*
Posts: 3


View Profile
« on: June 28, 2008, 12:55:52 pm »

In the problem 34479 after 1.Ba3 Kd8 I was looking for forcing moves first, as I was taught, and I quickly found 2.Rxd7+ Kxd7 3.Qxe7+ Kc6 4.e5+ Kb5 5.Qb4+ Ka6 6.Qa4# but to my surprise 2.Rxd7+ is a mistake! Instead the computer proposes quiet move 2.Bd6 which leads to the mate in 3, but OTB I would definiitely go for the forcing continuation. I propose to give 2.Rxd7+ status: good move, keep looking.
Logged
richard
Administrator
Hero Member
*****
Posts: 727


View Profile
« Reply #1 on: June 28, 2008, 01:57:54 pm »

Thanks szaszkin,

That particular move SHOULD have resulted in an alternative, however I just checked the code and unfortunately there is a bug that prevented the alternative from being recorded in this particular case. It is a fairly rare set of conditions that triggers this so it shouldn't impact that many problems. I will fix this bug in the generator for the next problem set release.

Richard.
Logged
revenant
Newbie
*
Posts: 44


View Profile
« Reply #2 on: July 21, 2008, 09:30:17 am »

2. Rxd7+ still produced a fail when I tried it just now.  Not sure what's going on.

Must say I admire 2. Bd6 though.  Pretty move.
Logged
richard
Administrator
Hero Member
*****
Posts: 727


View Profile
« Reply #3 on: July 21, 2008, 09:53:10 am »

Hi Revenant,

Generator runs take several weeks to go through the entire problem set and I still had a few other things to fix at the time I wrote that, so I've only just got to starting the verification run with the new generator. I started a few days ago, I'm guessing there is aprox 10-15 days to go.

The changes to the generator are:
- Fix this and one other bug that prevented some alternatives in mating situations from being accepted.
- Reduce the "winning" threshold on alternatives. Previously "winning" was quite close to a full piece, now the threshold is just under +2. The old value was too high, especially in endgame positions where having say 2 extra pawns is often clearly winning.
- Enforce a reasonable gap between the 'best move' and any alternatives, this avoids the "well I got 'try again' on a move which seemed just as good as the best move" problem.  Often the gap was so small that running the position on a different engine (or the same engine for longer) that it preferred the second best over the best.
- Collect the computer analysis of the 4 best moves at each players move (doesn't for now collect opponent moves other than the 'best').  This will allow presenting the analysis of the other lines and allowing (probably premium) users to navigate through the variations as well as the main line when viewing the problem after they have finished their attempt.

Regards,
Richard.
Logged
Pages: [1]
Print
Jump to: