On 5/16/2025 4:40 AM, Øystein Schønning-Johansen wrote:
What would it take to convince you, for example that
the doubling cube magnifies luck more than skill in
gamblegammon?
This is actually an interesting question. Intuitively
the cube reduces the luck-factor, but my (and others)
intuition can be wrong. Can you suggest an experiment
that can answer such questions?
I had already suggested and also actually ran several
experiments on this question, which weren't discussed
here but were debated at length in "daily gammon". I
no longer have access to those archives but Ian may
and contribute some quotations from them if he wants
to participate in this thread.
Before running my experiments, I asked readers whether
a weak player would do better against a strong player,
in long enough cubeless or cubeful sessions. They all
predicted that he would do better cubeless, (based on
the "intuitive" fallacy that cube promotes skill), but
the results were to the contrary.
Those experiments are available on my backgammon site,
with Python scripts, results, comments, etc.
I had posted on this in "bgonline.org" a few months ago:
https://www.bgonline.org/forums/webbbs_config.pl?read=215345
For the ones who may not want to follow the link, I'll
copy it below. Towards the end of it, I explain in full
details how you all can run your own experiments using
GnuBG or XG without even any need for fancy scripts. (For
this kinds of experiments GnuBG is far superior to XG, in
how detailed player configurations can be, etc.)
I'm curious to see what kind of comments may come from
the theoreticians, mathematicians and "scientists" here.
MK
========================================================
cube magnifies luck
Posted By: MK <[email protected]>
Date: Saturday, 11 January 2025, at 9:39 a.m.
In Response To: Cancelgammon (ah_clem)
ah_clem said: a) ..... it seems prudent to wait until you opponent rolls
something very lucky .....
Obviously, this is another sketchy variant with the purpose of reducing the element of luck in
gamblegammon, while one of the best ways of achieving that to a certain degree, without a set of
complicated rules, would be to simply eliminate the use of doubling cube.
I had started suggesting that the cube magnifies luck more than skill barely two years after I was
exposed to Jellyfish, FIBS and cubeful gamblegammon, (after having played 20+ years of backgammon).
If you search the SCT archives with keywords "mk magnifies luck", you will find 18 threads with the
first one that I had started on the subject. You don't need to read them all. This is just to show
you that it had been a long going subject of discussion.
https://groups.google.com/g/rec.games.backgammon/search?q=mk%20magnifies%20luck
I had first used the words "The cube magnifies the luck factor" in 1998, in
this post:
https://groups.google.com/g/rec.games.backgammon/c/42gg-RUTobI/m/czpeTxGD33gJ
In response, Michael Zehr had offered his "theory for why the cube will magnify skill instead of
luck" in this post:
https://groups.google.com/g/rec.games.backgammon/c/2fCYjYSo9Ts/m/u0UjSR3EPusJ
Even as I kept arguing for it for decades, I didn't have any ways to demonstrate it other than my
own playing against bots, which were honest but easily dismissable subjective experiments.
But recently, oh well how time flies eh? :(, about six months ago or so, I had finished my last set
of experiments that demonstrated that the cube did indeed magnify luck. But I kinda lost interest in
trying to prove to a god-fearing church flock that god doesn't exist and didn't update my web site
nor post about it in BG forums. But this cancelgammon thread sparked my interest to write about it
again.
The beautiful thing about this is that you don't even need to trust my experiments, run complicated
scripts, etc. to prove it to yourselves. All you need to do is make a bot play against itself as two
different checker skilled opponents; first cubless, then cubeful; and compare the results.
Set one player to, let's say, one ply and the other to three ply. Let them play, let's say, 10,000
cubeless money/point games and write down the ratio of points won by each side.
Then turn on the cube and set the same cube skill level, let's say two ply, for both players; let
them play 10,000 cubeful money/point games and write down the ratio of points won by each side.
Now compare the results. You will see very clearly that the weaker checker player wins much more
with the injection of the cube. This is undeniable evidence that cube magnifies luck more than it
magnifies (if any at all) skill.
If you have the time and need to convince yourselves even more, run several combinations of checker
and cube skills. The difference is more visible with "mutant bot vs bot" experiments if you can do
those but if not, even though the skill differences between the lowest and highest bot settings is
small, you will see it given that you run long enough sessions.
I have done my experiments using GnuBG but I assume you can do the same
experiments using XG.
So there you have it. Maybe I'm too soft-hearted to feel pity for you guys but I suspect that deep
down inside, you all silently pity yourselves also...
========================================================
