Rebel endgame blunder

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

Colin-G
Posts: 191
Joined: Mon Oct 31, 2016 6:30 pm
Location: England

Rebel endgame blunder

Post by Colin-G »

I was playing a game today between Rebel 12 (DOS version from 2004 running in Windows 98 in VirtualBox on a linux computer) and Fire 7.1 running in Windows 10 on another computer. I was making the moves manually on each computer.
Although Rebel 12 has an auto232 function in its drop down menus, which would have allowed the use of an RS232 serial cable to transfer moves, I could not get it to work. I remember reading a long time ago that Ed had disabled this function on his programs for some reason.
Rebel was playing at 10 sec per move and Fire at 40 moves per 4 mins. This made it a fairly equal playing field allowing for VirtualBox efficiency and the different speeds of the 2 computers. Permanent brain was off for both.
I like to play engines against as many different opponents as I can. I believe it gives a more accurate elo table. I sometimes play newer engines against very old engines to try and equalise out the opposition elo values in the table.
Anyway, this interesting endgame position was reached.

[d]2K5/3P4/8/8/p3k1Pb/8/8/2B5 b - -
Rebel had done very well to hold out so far and I believe the position is drawn.
Rebel then blundered by playing a3?? Black took the pawn and soon won.
At the time I saw it was a blunder and thought White should have gone for the g pawn, leaving White's bishop looking after the a pawn. On analysing later the only move not losing for Black is Be7.
I then tried the position with the UCI version of Rebel Century 3 in Arena. That also plays a3??
Even ProDeo 2.2 (also by Ed) running in Arena on a fast computer wants to play a3?? It takes a few minutes before settling down to Be7 and not reverting to a3 again.
This is surprising since all the other engines I tried, including several weak ones find Be7 almost straight away.
SSTATHIS
Posts: 35
Joined: Tue Jun 22, 2010 7:55 pm

Re: Rebel endgame blunder

Post by SSTATHIS »

Yes! I confirm that!
Prodeo 2.92 (same as 2.2) and Benjamin 1.0 makes the same mistake!
Very strange.....
User avatar
Rebel
Posts: 6995
Joined: Thu Aug 18, 2011 12:04 pm

Re: Rebel endgame blunder

Post by Rebel »

Oh la la.... that's a real bad one.

A quick look reveals that after 1..a3? 2.Bxa3 Kf4 the (at first glance) silly 3.g5! the engine prunes too much and overlooks 3..Bxg5 4.Bc1+ Kg4 5.Bxg5 Kxg5 6. d8=Q

Very strange indeed.
90% of coding is debugging, the other 10% is writing bugs.