wrote:
[...] given the determination of organizers to use sudden-death, it
was the best anyone could come up with.

Well, you want there to be a fixed time when the next round starts,
right? The only options are sudden-death time controls or
adjournments.


Dave.

--
David Richerby Erotic Laser (TM): it's like an
www.chiark.greenend.org.uk/~davidr/ intense beam of light but it's
genuinely erotic!

On Mar 22, 7:27 pm, David Richerby
wrote:
wrote:
[...] given the determination of organizers to use sudden-death, it
was the best anyone could come up with.

Well, you want there to be a fixed time when the next round starts,
right? The only options are sudden-death time controls or
adjournments.


Wrong. There is another possibility which
still yields a "fixed time" for the start of the
next round; it's called adjudication.

-------------------------------------------------------------

Back in the old days, there were serious
problems with adjudication, including bias
on the part of the adjudicator which might
have him granting, say, Jose Capablanca a
draw where he has given Joe Patzer a loss.
But today, we have the answer to personal
bias: computers. It also just so happens
that the /strongest/ chess players in the
world are reasonably inexpensive programs,
and many tournament directors already
have a notebook computer handy, for their
pairings program. In addition, there are
sites on the internet which give easy
access to some basic endgame table-
bases (although of limited use for now).


-- help bot

David Richerby wrote:
wrote:
[...] given the determination of organizers to use sudden-death, it
was the best anyone could come up with.

Well, you want there to be a fixed time when the next round starts,
right? The only options are sudden-death time controls or
adjournments.


Dave.

--
David Richerby


That's correct. The question is whether the benefits of sudden death
exceed the price paid for it. The answer is far from obvious.

help bot wrote:
On Mar 22, 7:27 pm, David Richerby
wrote:
wrote:
[...] given the determination of organizers to use sudden-death, it
was the best anyone could come up with.

Well, you want there to be a fixed time when the next round starts,
right? The only options are sudden-death time controls or
adjournments.


Wrong. There is another possibility which
still yields a "fixed time" for the start of the
next round; it's called adjudication.

-------------------------------------------------------------

Back in the old days, there were serious
problems with adjudication, including bias
on the part of the adjudicator which might
have him granting, say, Jose Capablanca a
draw where he has given Joe Patzer a loss.
But today, we have the answer to personal
bias: computers. It also just so happens
that the /strongest/ chess players in the
world are reasonably inexpensive programs,
and many tournament directors already
have a notebook computer handy, for their
pairings program. In addition, there are
sites on the internet which give easy
access to some basic endgame table-
bases (although of limited use for now).


-- help bot


Nonsense. Adjudication is the worst possible solution. Computer
evaluations are simplistic, useful evaluation might take hours or
days, and deciding the result of a game on the basis of a computer
program's valuation of .1 of a pawn plus or minus is moronic.
Fortunately, there is zero chance of such a suggestion being taken
seriously.

help bot wrote:

David Richerby wrote:

wrote:

[...] given the determination of organizers to
use sudden-death, it was the best anyone could
come up with.

Well, you want there to be a fixed time when the next
round starts, right? The only options are sudden-death
time controls or adjournments.

Wrong. There is another possibility which
still yields a "fixed time" for the start of the
next round; it's called adjudication.

-------------------------------------------------------------

Back in the old days, there were serious
problems with adjudication, including bias
on the part of the adjudicator which might
have him granting, say, Jose Capablanca a
draw where he has given Joe Patzer a loss.
But today, we have the answer to personal
bias: computers. It also just so happens
that the /strongest/ chess players in the
world are reasonably inexpensive programs,
and many tournament directors already
have a notebook computer handy, for their
pairings program. In addition, there are
sites on the internet which give easy
access to some basic endgame table-
bases (although of limited use for now).

You would have to factor in ratings.
If a game between two 1500-rated players
is adjudicated as a win using a line that
only a 2500-rated chess computer can find,
that wouldn't be fair. But if you factor
in ratings, what of the player who is
playing his first rated game?

I wonder if some variation on the backgammon
doubling method of forcing a losing player
to resign might work...

Guy Macon http://www.guymacon.com/ wrote:
I wonder if some variation on the backgammon doubling method of
forcing a losing player to resign might work...

Doubling only works in matches. In a tournament, the correct
procedure would be for each player to double every move so that
whoever wins gets the highest possible score.


Dave.

--
David Richerby Broken Mentholated Painting (TM): it's
www.chiark.greenend.org.uk/~davidr/ like a Renaissance masterpiece but
it's invigorating and it doesn't work!

David Richerby wrote:
Guy Macon http://www.guymacon.com/ wrote:
I wonder if some variation on the backgammon doubling method of
forcing a losing player to resign might work...

Doubling only works in matches. In a tournament, the correct
procedure would be for each player to double every move so that
whoever wins gets the highest possible score.


Dave.


Not if you change the basic scoring to [-1, 0, +1] and you double BOTH
the positive gain for a win and the negative score for a loss.

Note that without the doubling, [-1,0,1] is the same as [0, 0.5, 1.0].

I suspect the actual effect would be to increase the number of draws.
After a few doubles, it becomes too expensive to take a risk and play
for the win.

Imagine how backgammon would change if you allowed players to agree to a
draw.

--
Kenneth Sloan
Computer and Information Sciences +1-205-932-2213
University of Alabama at Birmingham FAX +1-205-934-5473
Birmingham, AL 35294-1170 http://KennethRSloan.com/

Kenneth Sloan wrote:
David Richerby wrote:
Guy Macon http://www.guymacon.com/ wrote:
I wonder if some variation on the backgammon doubling method of
forcing a losing player to resign might work...

Doubling only works in matches. In a tournament, the correct
procedure would be for each player to double every move so that
whoever wins gets the highest possible score.

Not if you change the basic scoring to [-1, 0, +1] and you double
BOTH the positive gain for a win and the negative score for a loss.

That still doesn't quite work. In the last round, two middle-of-the-
field players may as well double to infinity because they've nothing
to lose and whoever wins the game will win the tournament. Perhaps
one could invent some sort of Crawford's rule analogue to deal with
this? (Crawford's rule says that a player cannot double in the first
game where his opponent needs just one point to win the match, though
he can redouble if his opponent is foolish enough to double. This is
to prevent the person who's trailing in a match using the `nothing to
lose' argument and doubling as soon as his opponent needs a single
point to take the match.)

Even without this sort of repeated doubling, suppose player A believes
he's in a won position and doubles. B, his stronger opponent redoubles
immediately because he can see that A is actually lost. Even assuming
that A takes the hint, the game is still worth four points and more
because of A's incompetence than B's skill.

I suspect the actual effect would be to increase the number of
draws. After a few doubles, it becomes too expensive to take a risk
and play for the win.

For really good players, yes. But it's hard to imagine really good
players doubling and redoubling -- games of chess don't tend to have
the to-ing and fro-ing that backgammon does.

Imagine how backgammon would change if you allowed players to agree
to a draw.

I've played a bit of backgammon but mostly against the computer. I
can't see any circumstances under which one might agree a draw -- if
the game looks level, you'd just wait and not double, since it'll
surely unbalance itself after a while. Since the men can only move
forwards in Backgammon, progress of a sort must be made all the time
so I don't see a situation comparable to the sort of honest draw in
chess where neither player can make progress by any means other than a
gross blunder from the opponent. I've not played backgammon except
socially and against the computer -- am I missing something?


Dave.

--
David Richerby Beefy Spoon (TM): it's like a piece
www.chiark.greenend.org.uk/~davidr/ of cutlery that's made from a cow!

On Mar 23, 9:25 am, Guy Macon http://www.guymacon.com/ wrote:
help bot wrote:

David Richerby wrote:

wrote:

[...] given the determination of organizers to
use sudden-death, it was the best anyone could
come up with.

Well, you want there to be a fixed time when the next
round starts, right? The only options are sudden-death
time controls or adjournments.

Wrong. There is another possibility which
still yields a "fixed time" for the start of the
next round; it's called adjudication.

-------------------------------------------------------------

Back in the old days, there were serious
problems with adjudication, including bias
on the part of the adjudicator which might
have him granting, say, Jose Capablanca a
draw where he has given Joe Patzer a loss.
But today, we have the answer to personal
bias: computers. It also just so happens
that the /strongest/ chess players in the
world are reasonably inexpensive programs,
and many tournament directors already
have a notebook computer handy, for their
pairings program. In addition, there are
sites on the internet which give easy
access to some basic endgame table-
bases (although of limited use for now).

You would have to factor in ratings.
If a game between two 1500-rated players
is adjudicated as a win using a line that
only a 2500-rated chess computer can find,
that wouldn't be fair. But if you factor
in ratings, what of the player who is
playing his first rated game?

I wonder if some variation on the backgammon
doubling method of forcing a losing player
to resign might work...

I have toyed around with the concept, looking at a lot of different
approaches to the issue, and saw some ways it can work. It always run
into issues, and lack of acceptance. I hear people use a doubling
cube for chess in some European countries as a way to wager.

I believe the issue here is mostly handling draws, and also having a
way to pressure one's opponent to fold. I believe using an offshoot
of a a Bronstein clock could be an answer here. Instead of just a
time delay, before you have your clock eaten into, you do it
differently. Each time a delay period (say a minute or 30 seconds)
passes, your opponent scores a point. If game is drawn, whomever has
the most points, wins the points for the draw. You can put pressure
on your opponent by reducing the amount of time each player has
between moves. You an also have it so that, if a player is going to
drag too long between moves, they can do a time out, to get extra
time. The end game can be reduced, by this time pressure method, to
10 seconds a move, causing the player worse off collapse. Of course,
you can also offer this version of preventing stalling. If a certain
number of these time periods passes without a move, a player can lose
the game.

Is this a heretical idea? Yes it is. Might it be worth considering?
I hope so. Part of it came out of the need to figure how you could do
time control for live chess that would make it more suitable for
spectators to watch.

Another way to do this, would be players periodically during the match
could reduce the amount of time left in the game. You look at how
much the person who has less time on their clock is left, and whomever
has control of the decision to reduce time, can reduce the time left
on the clocks by half what the player with less time on their clock
has. Think you have the game wrapped up? Well halve the time each
player has. That player then manages to get to decide if they want to
reduce time later. You could also set this time reduction up so that
players end up having to decide whether to quit or not. Maybe you
double the points the match is worth, like the doubling cube.

- Rich

David Richerby wrote:

I've played a bit of backgammon but mostly against the computer. I
can't see any circumstances under which one might agree a draw -- if
the game looks level, you'd just wait and not double, since it'll
surely unbalance itself after a while. Since the men can only move
forwards in Backgammon, progress of a sort must be made all the time
so I don't see a situation comparable to the sort of honest draw in
chess where neither player can make progress by any means other than a
gross blunder from the opponent. I've not played backgammon except
socially and against the computer -- am I missing something?


Yes. To address your last point first - OF COURSE backgammon checkers
can move backwards! But, the real point is - you often find yourself in
a position where the entire game will hinge on the next roll. This is
not unlike an "unclear" chess position where neither player can see
enough to be sure who is winning. Another analogy might be a "race" in
poker where the odds are about 50-50 with a large amount of money on the
line.

In chess, players may accept a draw rather than continue in an unclear
position. Note that this is DIFFERENT than agreeing to a draw in a
clearly drawn position, and even different than agreeing to a draw in a
LEVEL position.

In poker, players may "run it twice" (if allowed), or even "chop". In
both cases, the point is to reduce variance when the odds are even but
the result will be a win or a loss (and very rarely a draw).

In backgammon, I've *often* been in positions where the cube has
(legitimately) gone back and forth and the position on the board is now
dead even....except that one player or the other will win based on the
next roll of the dice. In that position, I'd much rather agree to a draw.

In summary - these are all cases where the EV is 0.5, the stakes are
(now) very high, and the result CANNOT be 0.5 but instead must be either
0.0 or 1.0. Both players may well rather accept the (well earned) 0.5
than gamble.




--
Kenneth Sloan
Computer and Information Sciences +1-205-932-2213
University of Alabama at Birmingham FAX +1-205-934-5473
Birmingham, AL 35294-1170 http://KennethRSloan.com/