I found Mysha's post quite to the point. Her or his first assertion gives one more solution to the OP's question, and the other one points to some sloppiness in DMcG's writing.

Hypothesis: (1 - 0.9*) is smaller than any arbitrary epsilon.
Mysha's "disproof": (1 - 0.9*)/2 for epsilon is even smaller.

Pedants, however temporary, should take particular care to be exact and correct themselves.