veggivet wrote: ↑24 April 2022, 12:18
First game played today: 10 doubles in 30 moves. Please calculate the odds of essentially doubling the expected 16.7% frequency of throwing any double given a pair of fair dice. Table #2611705031 if you want to take a look. I realize that statistics is all about larger numbers, but having played nearly 1000 games, my sense is that doubles occur much more frequently than they would if the dice were truly random. If you experience a 1 in 500 year flood every 5-10 years, it's probably time to rethink the '1 in 500' terminology.
Your own analogy is the problem here - what you are describing is not even remotely like a "1 in 500 year flood". The probability of getting exactly 10 doubles in 30 moves is about 1.3%. As someone else posted earlier, the probability of getting 10
or more doubles in 30 rolls is higher than that - around 1.97%. If you played one game of Backgammon every day with fair/unbiased dice, you would expect to see it happening around 7 times a year. Your (incorrect) sense about how rare such an event is is leading you to draw the wrong conclusions about the random number generator. You'd have to be getting up to around 17 or 18 rolls out of 30 being doubles to get to the level of a "1 in 500 year event" (and even then, 1 in 500 years is rare, not impossible).
Not long ago, I lost a game that I was ahead on because the other player rolled 6-6, 6-6, 5-5 in a row, which was the minimum combination that would have let them win from that position. The probability of any one specific game finishing like that is low - but it's not "1 in 500 years" low, I've played enough backgammon in my life that the probability of that happening to me
some time is likely somewhere around 10-20%. It would only have been suspicious if my opponent had predicted it beforehand. No point complaining about it, they didn't cheat, rare events are just that - rare, not impossible. I'm sure there have been times where I've unexpectedly won from behind with a lucky combination of rolls too.
I'd be far more suspicious if the BGA (or any other site's) random number generator
never produced these sort of results, because that would suggest that they were manipulating the results to pander to certain peoples' misconceptions about probability and hence avoid complaints about the random number generator.