On 6/3/2025 9:14 PM, Timothy Y. Chow" via "Bug reports for and general discussion about GNU
Backgammon. wrote:
> Vixra was founded by people who objected to the arXiv's policy.....
> in practice Vixra has become a junk heap of crackpot papers,
> interspersed with a few rare gems.
Who is crackpot may be subjective but impostors are even worse.
While reading about Arduino-Powered BotGammon, I came across
an incredible example at https://en.wikipedia.org/wiki/Arduino:
In April 2017, Wired reported that Musto had "fabricated his
academic record. On his company's website, personal LinkedIn
accounts, and even on Italian business documents, Musto was,
until recently, listed as holding a Ph.D. from the Massachusetts
Institute of Technology. In some cases, his biography also
claimed an MBA from New York University." Wired reported that
neither university had any record of Musto's attendance, and
Musto later admitted in an interview with Wired that he had
never earned those degrees.
> My first instinct, without studying the paper in detail, is that the
> paper makes some interesting and valid observations,
Do you want to expand on what you found as "interesting and valid
observations"?
> As for whether *any* backgammon paper has been published in
> a scholarly mathematical journal, the answer is yes, although I
> don't think there have been many.
I had in mind mathematical theories supported by empirical data.
Math is not the problem. How people try to apply it to backgammon
is the problem.
> Among the most famous is the 1975 paper by Keeler and Spencer.
> https://bkgm.com/articles/KeelerSpencer/OptimalDoublingInBackgammon/
Interesting early example of how these "skill theories" start out well
but then turn into bullshit with no proven practical use. In chapter-3
he starts talking about doubling at 0.6, 0.65, 0.67, 0.7, 0.8, etc. And
ends up saying:
More important is the fact that backgammon is not really continuous,
especially near the end of the game. ..... a player should double
whenever p > 0.5. At no time should a player decline a double if
p ≥ 0.25 ..... The farther from the end of the game, the less the
probability of winning changes with each move, and the closer to
each other the doubling point and the folding point become.
I kept talking about two subjects related to this in RGB for decades.
1- The extrapolated, estimated equities and winning chances aren't
accurate enough in early and middle games to pretend that one can
suggest such precise (not to ignore "subjective") doubling points.
2-If one insists that they are accurate, then the player who first gains
a 50+% advantage should double, since if the rest of the game is
played "perfectly" for billion times, that player will win 50+% of the time.
I have shown these and more with my "mutant experiments" but it's
hard to convince "flat-math-ers" with empirical data.
MK
