The cutoff point

Having asked the question on the weekend, I'll provide my back-of-the-envelope answer as to how many votes I'd expect a candidate to need based on what we know now about the NDP leadership race (which of course figures to take a few more turns before it's done).

Let's start with what I'd consider to be reasonably conservative assumptions about where the party's membership numbers should be by the end of the race.

The most obvious variable facing candidates is the question of how many members there will be from Quebec by voting day. As a starting point, let's figure that in the course of the leadership race, a well-motivated set of candidates should at least be able to push Quebec to the same per-capita membership as the next-lowest region in the country (Atlantic Canada, with roughly 4,000 members for 2.3 million people). Prorated to Quebec's population level, that would set the party's membership in Quebec at just over 13,000.

Across the rest of the country, let's then assume a 10% bump in membership based on both the NDP's improved popular support and the attention of the leadership campaign itself - with the exception of B.C., where the current membership numbers are already based on an active leadership campaign.

The approximate provincial/regional levels would then be as follows (guesstimated to within 500):
B.C./North 30,000
Alberta 10,000
Saskatchewan 10,000
Manitoba 11,500
Ontario 24,000
Quebec 13,000
Atlantic Canada 4,500

That would result in a national membership total of 103,000. But we shouldn't stop there, since we can't assume that all members will wind up voting. After all, in the B.C. leadership election, about 19,000 of those 30,000 current members cast ballots; and to my recollection two-thirds seems to be about the right number based on other leadership contests as well.

Which means that at the low end, we might expect about 70,000 voters to cast ballots by the time all is said and done. And in turn, a candidate who can reasonably hope to identify 35,000 possible supporters for the final ballot should have at least some chance of winning - while one without some plausible path to that total is likely doomed from the outset.

Of course, the above membership estimates could well be off in either direction (though I'd consider it far more likely that they'll be on the low side). And candidates will need to adjust their membership targets accordingly. But at the very least, the 35,000 target should make for a reasonable cutoff point for candidates deciding whether or not to take the plunge.

[Edit: fixed typo.]