I don't really know a lot about the area but I suspect that an undesirable example of nudging would have to satisfy two conditions:
1) The 'choice architect' (not a phrase I had heard before but I think I know what you mean), has made an incorrect choice of the 'default rule' by picking one which does not, in fact, maximise welfare.
2) The default rule is, contrary to what the choice architect believes, not really a 'nudge' at all because the choice architect has vastly underestimated the costs (whether financial, informational, or whatever) in switching from the default position to another option.
I'm not sure that I agree that a rolling contract is a 'nudge' type situation at all. I would have thought the idea in a 'nudge' was that the choice architect had as his goal maximising social welfare. The drafter of the rolling contract has as his goal maximising the profit of the company he is drafting for. He wants to make it as costly (economically) as he can for consumers to cancel the contract without falling foul of any statute law prohibiting such terms.
Have you read the Posner article responding to Jolls, Sunstein, and Thaler's article? If I recall correctly it is in the same edition of the Stanford Law Review as the Sunstein article. There are also (from what I vaguely remember) some other critical pieces in the same journal.