Re: bustabit.com -- The Social Gambling Game (formerly moneypot.com)

Here's a better example: I flip a coin. It lands tails three times in a row. I get ready to flip it again. Do you think it's more likely to land heads on this flip than tails? Why? The coin has no memory.

I really like the coin example, because people can look at a coin and see it has no memory of the past. What tends to really confuse people is more abstract things like "low game median". E.g. over the last 1000 games, let's say the median is bust was 1.9. We know the median is going to trend back towards 1.97x -- so surely it's a good time to bet?

If they could place an infinite number of bets, knowing that it will converge to 1.97x, it seems like that would make it +EV. But going back to the coin example, if it lands on tails 100 times in a row, we know it'll be 50% heads on an infinite number of flips, but unfortunately we can't bet an infinite number of times to catch that profit :-P

I don't see the illusion.. I believe the past events should be entirely discarded. So whether it was 1.9 or 2.1 or a 100, it shouldn't matter.

If there are two 1% house edge sites with same bankroll at the point, one with -1% (past) actual return on wagered and another with 2% actual return, and you are to invest in one of them, I wouldn't see any difference between the two.
But depending on how we think, we are like to prefer one over the other.

Edit: which turns out to be exactly the same as what RHaver said just in a different way. Grin

Eventually the number of past events will be so small when compared to infinity that the past events doesn't matter.