Tim May:
In any case, the Shor work on a quantum factorer is interesting, but is at least several decades away, in my opinion. And even then it is likely to be "workable" out to some number of digits (roughly, number of digits = precision needed), by which time the conventional advances in computer power will mean we're all using 10,000-bit moduli (especially if we have just heard that NSA has just spend $32 billion to build a Shor machine able to factor 3000-bit moduli :-} ).
I won't argue whether Shor's work will be implemented or not within any given time period, but I thought that one of the most important properties of it is that once (and if) achieved, the resources required to factor increasingly large moduli lengths go up only polynomially, not exponentially. Doug Cutrell