Can You Even the Odds?
This Week's Fiddler
There are effectively two states at any point: the current player is about to go once, or the current player is about to go twice. Let be the chance the current player wins in the former case, and let be the chance in the latter case.
For , the chance is of winning outright. Upon the chance of not winning outright, the current player switches and gets two rolls, so the chance of winning given that it gets to that point is . In other words:
For , the chance is of winning outright. Upon the chance of not winning outright, the current player stays the same with one roll left, so the chance of winning is . In other words:
Hence:
Which gives us our answer, .
An alternative solution is to split the chance of A winning ABBA into the sum of two geometric sequences: every fourth term in the first position, and every fourth term in the fourth position. The first term is , and the fourth term is . The ratio between every fourth term would be . This means the sum is:
Which gives us the same answer!
Answer: 31/61 (about 50.82%)
Extra Credit
Interesting problem! The Thue-Morse sequence is so beautiful, and I expect the answer to be so as well. Simulations say it's around 50.16% for A to win; about even with an ever-so-slight edge to A.
Answer: ¯\_(ツ)_/¯