[seqfan] Problem reproducing A019989, A019990, and A019991

[Sean A. Irvine]

The sequences A019989, A019990, and A019991 partition the integers into
three disjoint sets.

 I tried to implement these sequences as described, but all the functions
a, b, c, A, B, C quickly become fixed on true.  From the actual values, it
would appear that exactly one of a, b, c, should be true at each even
iteration, so I think there must be either something missing from or
incorrect in the description.

I tried using XOR on place of OR, and various other possibilities, but was
unable to reproduce the values of the sequences.

Ideas?

Sean.