MS Game Zone backgammon & statistics

[Prof. Harmon]

Just talked to my unnamed source at the Zone - let's use Deep Throat (DT)
as his/her sobriquet. DT says that the Zone does indeed keep statistical
reports on the frequency of all numbers being spit out by their random
number generator. According to DT, a true math whiz of a professor at a
truly superior Boston university, doubles will occur 1 in 36 rolls (1/6 *
1/6). This is not to say that if one rolls the dice 36 times one will roll
1 double. Statistics cannot operate at such a low occurance of chances.

What these reports do show is that at times when "certain" employees are on
duty in the server room in IT, the ferquency of doubles is higher than
statistically probable. Following this revelation the Zone developed
software to determine if individual players are being given an unfair
advantage. My source (DT) hasn't seen these reports yet. When he does I'll
pass along a synopsis of the results.
--
Your diligent reporter

[Jim Fouch]

I am glad I didn't study math under that whiz from a truly superior Boston
university - actually, I never studied much math at all but would be very
surprised if doubles did not have a probability of 6 in 36 rolls or 16.7%.

Cheers, Jim.

"Prof. Harmon" <r_h_h...@Zmail.com> wrote in message
news:Xns966BE9B66CBB6r...@207.217.125.201...

[happyjuggler0]

Dear "Professor Harmon",

I have no love for the Zone but if you are going to try to scam us
I suggest you get your math right first. Doubles occur 1/6 of the time,
not 1/36 of the time. Furthermore, I strongly suspect that if they were
going to cheat on behalf of someone they would use other rolls as well
as doubles. Doubles can truly suck a lot of the time. Finally, an AWFUL
LOT of bg games (and dice rolls) are happening at the same time and
cheating for a friend each "shift" will still likely produce dice
trials still well within expected variance.

Yours truly,
Professor happyjuggler0

[3dfibster]

This is a fascinating report indeed. No doubt the news media will jump
all over it 30 yrs from now when this DT crawls out of a nursing home
and reveals the true identity. ummm. I can't hardly stand the suspense
:-)

[namecr...@gmail.com]

On Monday, June 6, 2005 9:23:47 AM UTC-5, happyjuggler0 wrote:
> Dear "Professor Harmon",
>
> I have no love for the Zone but if you are going to try to scam us
> I suggest you get your math right first. Doubles occur 1/6 of the time,
> not 1/36 of the time.

Just a note: programmers who try to write random dice use this to defend that they truly are. "The chance of rolling a double is 1 in every 6 rolls or 16.7% of the time"

That is a false statement. The chance of rolling a double on your next roll is 16.7%. However, it is not true that the dice will double 16.7% over the course of 50, 100, 1000 or even a million rolls.

There is a distinction there. If you roll real dice, they aren't going to match that percentage over time.

Programmed random dice are more like slot machines that pay off 16.7% of the time. Programmed dice will make sure that doubles occur 16.7% of the time. real dice would not, no more than flipping a coin would also always be 50% heads and 50% tails over time.

You cannot expand the calculations of the odds of one dice roll and assume it should apply over any number of rolls. That calculation is for one roll.

[Michael Petch]

On 2012-11-01 16:44, namecr...@gmail.com wrote:
> Programmed random dice are more like slot machines that pay off 16.7% of the time. Programmed dice will make sure that doubles occur 16.7% of the time. real dice would not, no more than flipping a coin would also always be 50% heads and 50% tails over time.

Dice that are programmed to evenly distribute doubles to guarantee
payoff 1/6th of the time (It sounds like that is what you are
suggesting) would likely be terrible dice, and could be discovered with
simple statistical analysis. The probability of rolling a double with a
proper PRNG or real dice is 1 in 6, but there is no guarantee that
doubles will occur 1/6th of the time over a small sample set.

Properly generated dice using a well sudied Pseudo RNG (like Mersenne
Twister or Blum Blum Shub) would be indistinguishable from real world
sources of randomness. These PRNGs don't generate dice values
specifically, they produce a sequence of 1's and 0's that happen to
appear random. Just like real world randomness you can have droughts of
0's (or 1's) or have long runs of 1's and 0's. These bits are converted
into values between 1 and 6 to be used as dice. They aren't specifically
programmed to make sure there is a double 1/6th of the time. It just so
happens that like with real life dice, over the longer term the trend
would be that doubles would approach 1/6th of the time, but the doubles
are not necessarily evenly distributed (as you would expect).

I had to read the rest of this thread in its entirety as the post your
responded to is 7+ years old. Professor Harmon's original post is funny
indeed, and I'd question the value of a Boston University statistics
degree. Of course the original post by Professor Harmon stated

"According to DT, a true math whiz of a professor at a truly superior

Boston university, doubles will occur 1 in 36 rolls (1/6 * 1/6). " . 1
in 36? LOL

I have a hard time believing that Professor Harmon was being serious,
and if he was, he was living in a different universe than the one I am
accustomed to, or his universe exists through the use of some really
good drugs.

[rebc...@gmail.com]

it is clear that no one has written a reasonable RGN. Or the alternative is true, incompetence is rampant or worse, backgammon sites are total scams. Take your pick but probably a combination of both. For example, fibs is as bad as it gets. I doubt it is possible to write a worse RGN which the BOTs (total scam) take full advantage of. I imagine every other site is similar.

I at least wish these shit heads would come clean and admit this is the case and that they want to do better but just don't know how.