Main
Date: 29 Jan 2007 14:39:17
From: Jules
Subject: 50-move rule
I have a somewhat odd question. Is it possible to construct a
position (assume that a capture or a pawn move has just occured) that
is drawn, but in which one player has a win if we dispense with the 50-
move rule? I have have tried some positions, but have not found one
that takes more than 30-something moves for the win.





 
Date: 30 Jan 2007 23:19:23
From: Amarande
Subject: Re: 50-move rule
Jules wrote:
> I have a somewhat odd question. Is it possible to construct a
> position (assume that a capture or a pawn move has just occured) that
> is drawn, but in which one player has a win if we dispense with the 50-
> move rule? I have have tried some positions, but have not found one
> that takes more than 30-something moves for the win.

There are in fact TONS of such positions.

In fact, tablebases have shown that such "legally drawn, technically
won" positions are in fact the NORM for actually a fair number of piece
combinations.

* K+Q vs. K+B+B
There has been shown to be only one "true" drawn position here (barring
Black being able to trade B for Q), however, with a favorable position
for the K and B's this cannot be won in time, a lot of positions taking
70+ moves. (I think the max may be about 76 or 78 but I am not sure)

* K+B+B vs. K+N
"Attempts have been made to show that two B's win vs. a Kt in all cases,
but the results are inconclusive." -- Fine. Tablebases have shown that
the normal case here is in fact a win. However, with a favorable
position for the defender, even with an also favorable position for the
owner of the Bishops, this can take over 60 moves as well. What's worse,
even in the positions that can be won in time, the winning method is
almost like love - there has not really been found a consistent pattern
to it (that I know of) and it basically takes a computer with tablebases
to be able to find the thread.

Most infamously:

* K+R+B vs. K+N+N

Is a win for the R+B, but consider a position like this!

8/8/8/3n4/4kn2/8/8/RKB5 w - - 0 1

White to play wins in 202! I have heard of positions that take 250+ in
this situation as well ... in any case, it is the norm here, it seems,
for positions to take 170 or more moves (longer than all but the longest
GAMES, even).

From a technical standpoint one could suggest the reinstatement of the
old "extra moves allowed" rule (back in the day, it was either 50 moves,
or 2x the number of required moves if there was a forced win that needed
more than that), especially as now we can definitely establish the
required number of moves in a given position, however, from a practical
standpoint this would be quite a nighte (additionally, as I mentioned
above, many of the wins do not actually have really evident winning
plans, simply lines that somehow happen to win - do computers really
need more advantages? They already took the world crown last year ...),
hence things remain as they are.


 
Date: 30 Jan 2007 00:04:22
From: Bjoern
Subject: Re: 50-move rule
Jules wrote:
> I have a somewhat odd question. Is it possible to construct a
> position (assume that a capture or a pawn move has just occured) that
> is drawn, but in which one player has a win if we dispense with the 50-
> move rule? I have have tried some positions, but have not found one
> that takes more than 30-something moves for the win.
>
Yes.

http://en.wikipedia.org/wiki/50_move_rule