Sunday Jellyfish marathon and MK wins again.
Murat Kalinyaprak
with JF again. First I played a bunch of games
taking back moves, experimenting with alternatives,
etc. just to see if I had really "figured out" this
little "fish". Then I decided to start from the
start (i.e. lowest seed/counter values allowed),
increment the seed each game, and play 5 matches
of 15 points.
So, seed=1 counter=0 for the first match, seed=2
counter=0 for the second match. But I forgot to
increment the seed after the second match and
kept playing with it set at 2 (ending counter=1640).
JF is rolling the dice. (all scores in MK-JF order
with the winner indicated in parantheses)
Match1: 1-0 3-0 3-2 4-2 4-4 6-4 8-4 8-6 40-6 (MK)
Match2: 2-0 3-0 4-0 5-0 5-2 6-2 6-6 6-7 8-7 8-11 8-15 (JF)
Match3: 2-0 2-8 2-10 3-10 3-14 3-16 (JF)
Match4: 0-2 6-2 8-2 8-4 8-6 8-10 10-10 12-10 28-10 (MK)
Match5: 0-2 0-3 2-3 2-5 3-5 4-5 6-5 8-5 9-5 13-5 45-5 (MK)
Total points: MK=124 JF=52
If you eliminate the cube and 3 point gammons (as
in traditional Middle-Eastern style):
Match1: 1-0 2-0 2-1 3-1 3-2 4-2 5-2 5-3 6-3 (MK)
Match2: 1-0 2-0 3-0 4-0 4-1 5-1 5-2 5-3 6-3 6-4 6-5 (MK)
Match3: 1-0 1-1 1-2 2-2 2-3 2-5 (JF)
Match4: 0-1 3-1 4-1 4-2 4-3 4-5 5-5 7-5 8-5 (MK)
Match5: 0-1 0-2 1-2 1-3 2-3 3-3 4-3 5-3 6-3 7-3 8-3 (MK)
Total points: MK=30 JF=21
If you also eliminate 2 point games (i.e. 1 point
per game):
Match1: 1-0 2-0 2-1 3-1 3-2 4-2 5-2 5-3 6-3 (MK)
Match2: 1-0 2-0 3-0 4-0 4-1 5-1 5-2 5-3 6-3 6-4 6-5 (MK)
Match3: 1-0 1-1 1-2 2-2 2-3 2-4 (JF)
Match4: 0-1 1-1 2-1 2-2 3-2 3-3 4-3 5-3 6-3 (MK)
Match5: 0-1 0-2 1-2 1-3 2-3 3-3 4-3 5-3 6-3 7-3 8-3 (MK)
Total points (games): MK=28 JF=18
At the expense of not sounding humble (and perhaps
as everyone may think of themselves) I always thought
I was a pretty darn good player. But folks, I sure
ain't no professional, let alone being a world-class
player. So, what's happening here...?
I had recently read here that JF had won against one
world-class player and lost against another by about
50 points in each case, in a series of 200 games. So
far, I won 47 out of 78 games I reported here. If
you extrapolate the ratio, that would be equivalent
of 120 games out of 200, against JF's 80 wins, with
me winning 2 out of 3 and leading by 40 games. Am I
being just lucky for the moment and things will change
drasticly to JF's favor by the time we reach 200
games...? What if it doesn't...?
MK
PS: I would be glad to post the match files if anyone
is interested in looking at how I played against JF.
Patti Beadles
-Patti
--
Patti Beadles |
pat...@netcom.com/pat...@gammon.com |
http://www.gammon.com/ | "I trust you. It's just
or just yell, "Hey, Patti!" | that I'm scared of you."
Kevin Dickover
Kalinyaprak) wrote:
>
>So, seed=1 counter=0 for the first match, seed=2
>counter=0 for the second match. But I forgot to
>increment the seed after the second match and
>kept playing with it set at 2 (ending counter=1640).
>
I am curious. What level was Jellyfish set to? What was the time
factor and how fast is the computer that you are using? All of these
things will affect the level that jellyfish plays at. (I still
remember switching to a 233 with 64 megs of ram from a 486 with 16
megs of ram a while back and seeing jellyfish suddenly get MUCH
stronger-back to the woodshed for more whippings :-) )
Solo Deo Favente,
Kevin Dickover
<)))><
Christopher D. Yep
> [...]
>little "fish". Then I decided to start from the
>start (i.e. lowest seed/counter values allowed),
>increment the seed each game, and play 5 matches
>of 15 points.
> [...]
>Total points: MK=124 JF=52
>Total points: MK=30 JF=21
After reviewing your posts in the last week, I see that you touched upon three
different goals in backgammon matches:
(1) Maximize the chance of winning each game
(2) Maximize the expected number of points/game (i.e. avg # of your
points/game - avg. # of opp's pts./game; actually you go even farther by
eliminating backgammons)
(3) Maximize the probability of winning the match
In summary, these goals are:
(1) Games matter
(2) Points matter
(3) Matches matter
(1) is equivalent to treating every game as a one-point match.
(2) is (almost) equivalent to money play (people who play for money allow
backgammons to count and often use the Jacoby rule and beavers; ignore this
minor distinction if you are not familiar with the terms).
(3) maximizes one's chance of winning the match.
In match play, Jellyfish only considers goal (3). The three goals above are
different. For example, someone who only considers goal 1 should never drop a
double (unless the game is a guaranteed loss).
I only briefly looked through your posted games, but from your comments I
infer that you are essentially using goal (2) [points matter], while JF is
using goal (3) [matches matter]. If you want JF to use goal (2) then play a
"single game" (start a new game with "File-New-Single Game"; even though it
is a "single game," it is really the "points goal")). If you want JF to use
goal (3) then play a "match." If you want JF to use goal (1) then play a
1-point match.
>I had recently read here that JF had won against one
>world-class player and lost against another by about
>50 points in each case, in a series of 200 games. So
>far, I won 47 out of 78 games I reported here. If
>you extrapolate the ratio, that would be equivalent
>of 120 games out of 200, against JF's 80 wins, with
>me winning 2 out of 3 and leading by 40 games. Am I
>being just lucky for the moment and things will change
>drasticly to JF's favor by the time we reach 200
>games...? What if it doesn't...?
The only results that matter to JF are matches. In your posts I see that
with a small sample size you have a slight edge in matches won. Nothing
really unusual. On the other hand you have a significant edge in games and
points. This could be attributed to luck, but I will make an even stronger
prediction: if you were to play a large number of 15 pt. matches with JF you
would be ahead in both points and games (assuming that your playing strength
is that of a strong intermediate/ expert or better and that you are only
considering goal 2 [points]). On the other hand, you would likely be very
close or slightly behind in matches.
The cumulative points won in your matches with JF are distorted because you
are sacrificing some of your chances ("equity") to win the match in order to
increase your expected number of pts. won per match (see below).
For example consider match 3 of your post on "Fri, 17 Jul 1998 14:00:08
-0600." In this 15 pt. match, after 3 games you are leading 13-0. In game 4
you double to 4 and then later to 16. From JF's view, dropping either of
these doubles results in a sure match loss. Therefore JF accepted both of
those doubles, regardless of its winning chances. From your point of view,
doubling to 4 and later to 16 increased your expected number of points won
(I didn't look at the positions, but will assume that you were the favorite
both times), while decreasing your chance of winning the match. Note that by
doubling to 16, the match is converted to "double match point" (this means
that whoever wins the current game wins the match). So if JF had gotten
lucky, then you would have lost the match. On the other hand, if you had left
the cube at 2, then if you had lost the game, you would still be a huge
favorite to win the match, leading 13-2 (assuming a single game loss). Note
that from JF's point of view, after you doubled to 4 JF lost nothing by
immediately doubling to 8. I.e. if JF loses the game, it loses the match
whether the cube is at 4 or 8, while if JF wins the game, it wins an increased
number of points, thereby increasing its chance of eventually winning the
match. In fact, if the match had been a 17 pt. match then after your double
to 16, JF would have immediately redoubled you to 32. In other words, these
"mysterious doubles" by JF are a standard part of match strategy.
In conclusion, JF only considers matches won in "match" mode, therefore the
only result that you should use to judge its/your relative strength is the
match result.
You are probably a good player, but on the other hand, don't read too much
into the games and points won when playing in match mode with JF.
IMHO, playing with JF is a great way to improve one's bg skill.
Good luck
Chris
Robert-Jan Veldhuizen
[various snips]
MK> increment the seed each game, and play 5 matches
MK> of 15 points.
MK> Match1: 1-0 3-0 3-2 4-2 4-4 6-4 8-4 8-6 40-6 (MK)
MK> Match2: 2-0 3-0 4-0 5-0 5-2 6-2 6-6 6-7 8-7 8-11 8-15 (JF)
MK> Match3: 2-0 2-8 2-10 3-10 3-14 3-16 (JF)
MK> Match4: 0-2 6-2 8-2 8-4 8-6 8-10 10-10 12-10 28-10 (MK)
MK> Match5: 0-2 0-3 2-3 2-5 3-5 4-5 6-5 8-5 9-5 13-5 45-5 (MK)
MK> Total points: MK=124 JF=52
Those figures mean nothing; if you play 15pt matches against JF it will
just try to win the match; it doesn't care what the actual score is.
The real score is only this: You won 3 15pt matches and lost 2, a good
achievement against JF but nothing spectacular.
MK> At the expense of not sounding humble (and perhaps
MK> as everyone may think of themselves) I always thought
MK> I was a pretty darn good player. But folks, I sure
MK> ain't no professional, let alone being a world-class
MK> player. So, what's happening here...?
MK> I had recently read here that JF had won against one
MK> world-class player and lost against another by about
MK> 50 points in each case, in a series of 200 games. So
MK> far, I won 47 out of 78 games I reported here. If
MK> you extrapolate the ratio, that would be equivalent
MK> of 120 games out of 200, against JF's 80 wins, with
MK> me winning 2 out of 3 and leading by 40 games. Am I
MK> being just lucky for the moment and things will change
MK> drasticly to JF's favor by the time we reach 200
MK> games...? What if it doesn't...?
You obviously are not familiar with the concept of *match* play. In a
match, the goal is to be the first to get to or pass the required number
of points. The final score is: you win or you lose.
If you'd like to play for points, that's called *money* play. In that
case, each game stands on its own, and losing two points is two times as
bad as losing one, winning four is for times as good as winning one etc.
I don't have JF but I believe there's an option to play money games
against it. You could also try to play 1pt matches, but then there's no
cube and (back)gammons.
--
Zorba/Robert-Jan