One Move of the Algorithm

• Start at a standard knowledge state over (x,y,z).

• Read a request r ≠ z. (The case r = z is trivial.)

• Update the estimator. No payment either way.

• Move the distribution. We pay.

• Adjust the estimator. The adversary pays us.

• Optionally forgive; lower the estimator selectively. No payment.

• Las Vegas. Randomly pick a subsequent according to
  an appropriate distribution. No payment: "fair" wager.

• We are at a standard knowledge state over
  (x,y,r), (y,x,r), (x,z,r), (z,x,r), (y,z,r), or (z,y,r).