Lists of size Integer.MAX_VALUE

[Roedy Green]

Right, but these cases aren't applicable to Bloch's updated
binary search code. The original (lo + hi) / 2 code was
definitely a flaw, but I don't think it's fair to say that
his updated ones are flawed since they work exactly as claimed.

It is flawed in that the algorithms do not do what is advertised "set
m to the average of l and u, truncated down to the nearest integer".
It is not flawed because in the context of a binary search, the bugs
don't hurt it.