I understand the idea behind the commit phase, but when a site can easily spend weeks in the commit phase, isn't there a big danger that it's going to lose momentum? That all those who signed up for it forget about the site's existence? That some of them move on?
It doesn't really matter that you had 3000 people commit to using a site, if only 50 actually committed within the last week, and most of the others can't even remember committing.
I'm not sure what the solution is, but at the moment, it seems like you're not just setting high requirements for the first wave of sites (which is understandable and sensible), but actively strangling them by setting the wrong requirements.
The first wave of sites absolutely should have to run a tough gauntlet to prove their worth. But the gauntlet should be to prove the site's support from its actual community. And the commit phase doesn't do that.
The commit phase, as it is now, appeals to very few people. It appeals to the people who hang around here on Meta. The rules lawyers, the rep addicts, the people who are more interested in gaming the system, in "being in charge" (or at least, in having an impact on the SE platform). But it does not appeal to the actual community it was meant to gauge support from.
To succeed, the site needs support from the community it targets. And yet it gets created if it can demonstrate support from an entirely different group.
A site might have tens of thousands of people itching to use the site, but they won't commit because they're not interested in all the power games and politicking of shaping a new system. They just want to share information and ask questions. There is nothing for them in the commit phase.
Yes, every site needs a few expert users basically to ensure that the nontrivial questions get answered. But it doesn't need 500 of them. Even StackOverflow probably doesn't have 500 of them. Something like 10-20 would be reasonable for a new site. Again, most questions are not answered by the type of user who would "commit" to using the site. They are answered by people who come here to get a question answered.
So a much more reasonable requirement would be to run the "commit" and "beta" phases in parallel, and to revise the "commit" phase to specifically ask for a small number of expert users, willing to commit to answering questions, while the "beta" phase is used to gauge support from the broader user community.
If the site can get 20 such expert users to commit, and, say, a few thousand ordinary users to actually use the beta site for a certain length of time, it has a much better chance of surviving than one which passes the current test of requiring hundreds of the kind of user who's dedicated "playing the game", rather than merely answering questions about the subject matter.
But time has to factor into it. The question should not be "can we get X users to sign this virtual petition to please create the site", but rather "if the site is there, does it sustain its momentum, or do people just check it out, and never visit again?"
I committed to using the LaTeX site. I could easily drum up 500 casual users. All it'd take is a single email to my university's main mailing list. All the CS students struggling to write their reports and assignments in LaTeX would see it, and use the site the next time they're trying to write a report. If I sent it out to the Physics and Math departments as well, we could probably call it 1500-2000 potential users, give or take. Just from me.
But asking them to commit to a site they can't use would be a waste of their time. I'm not even going to bother asking. If I did, a handful would perhaps commit. The rest would see that it doesn't actually help them, delete the email, and forget about it. And then how much time would they spend looking at the next email I send, the one which says "the site is now live"?
The commit phase is hurting the upcoming sites by turning away their actual users. It tries to accumulate a number of "signatures", but it's bleeding users because actual users have absolutely no motivation for committing to use something later when what they're trying to do is get questions answered now.