Pekka wrote:
Tahnan wrote:
I wonder how true that is; if my random numbers happen to fall out such that the first several rounds, the ones my opponent is throwing with a zero bid, are rounds where I'm betting only a point or two...
It will be very rare to make small or large bids with the kind of randomness I described.
Oh, I see! I admit I didn't read what you posted carefully enough--I assumed that "
playing randomly"
meant "
assigning a random number of points to each round"
as opposed to "
assign a random round to each point"
. The latter feels much less random to me, in fact, insofar as--as you noted--the more points you have, the more likely they are to distribute evenly. The strategic play of [0, 15, 15] is certainly likely to beat something that's going to look roughly like [10, 10, 10]; the strategic play of [0, 0, 33, 33, 34] is even more likely to beat something that's going to look roughly like [20, 20, 20, 20, 20].
That's why I read "
randomly assigning points"
as picking a random number of points (out of the points left) for each round, using everything that's left for the final round. That's a case in which something like [2, 20, 8], which beats [0, 15, 15], isn't nearly as unlikely as it is in the pick-a-round-for-each-point random approach. Granted, even this random strategy isn't terribly good: on the 30-point 3-round game, the [0, 15, 15] strategy wins something like 65% of the time, and on the 100/5 game, the [0, 0, 33, 33, 34] strategy wins around 84% of the time, which does indeed amount to "
quite probably winning the rest"
, though it's at least somewhat better than the other random strategy.
Just for the heck of it, I wondered what happened if you pitted these random strategies--let's call them the Pekkarandom strategy and the Tahnanrandom strategy--against each other. Ironically and perhaps paradoxically, Pekkarandom beats Tahnanrandom with about the same percentages that the "
strategic"
play does: 65% for the 30/3 game, 88% for the 100/5 game! I didn't expect that Tahnanrandom, which does notably better against the "
throw-the-first-half"
strategy than Pekkarandom, would lose so badly to it.
Python script attached, for anyone who wants to check or vary my work.