That's definitely a good assessment of the problem with the process. I'm not sure that the specific problem that needs to be solved is one of commitment level; as I've described in the past, we already have a problem with throwaway commitments - most people who commit, don't follow through, for whatever reasons, so I don't think this would actually help the definition phase.
Here's how I would frame the problem:
The definition phase defines the scope of the site - what is on or off-topic.
Almost anybody is allowed to vote examples up or down.
On a live site, the equivalent action is voting to close/reopen. However, on a live site, this is a privileged action available only to those who have established
themselves as having some level of subject-matter expertise through reputation.
Therefore, the Area 51 definition phase is outsourcing a critical and generally privileged action to those who have not established themselves as being qualified to act.
The whole process is actually too democratic. As another member pointed out in chat, it is a bit like allowing a Canadian or American citizen to vote on government policies in Europe. Of course we'll vote for higher taxes in order to subsidize hotels and touristy restaurants. It'll come in very handy when we go for our vacations there.
Of course we'll vote up the "How do I fight my speeding ticket?" example question in the Law proposal. No lawyer would ever want to touch that with a 10-foot pole, but it's the kind of question we'd want to be able to ask.
Part of this is definitely motivational - having no skin in the game, so to speak - but commitments are purely symbolic and I can think of no practical way to make them any more "material" - except to give them a real material cost, but I doubt that's ever going to happen (nor am I saying it should).
The larger problem is that, even if these people are all acting in good faith and genuinely want to create the best proposal possible, they aren't experts in the field. Every proposal should identify its experts at the outset - Stack Overflow was for Programmers, Cooking was for Cooks/Chefs, Physics is for Physicists, Law would be for Lawyers, and so on and so forth. You really don't have the experience necessary to vote on Area 51 questions unless you're a practicing member of that community.
Maybe you don't need to be an actual lawyer - maybe a law student is fine - but "programmers interested in law" is a lousy seed audience for a site that's supposed to eventually attract lawyers.
As I've said before - having a site defined by amateurs/dabblers only serves to guarantee that the site will be for amateurs/dabblers. It's right there in the Area 51 FAQ - "Ask Real, Expert Questions." But if you're not an expert, how can you even identify an expert question, let alone ask one?
How is somebody with no plumbing experience even going to think of a question like, "If you run 2.5 GPM through 50 feet of 1/2" galv pipe, how many psi will be lost to friction loss?" I don't even know what some of those abbreviations mean! No, that won't happen, participants are almost certainly going to ask and vote for the "How do I unclog a drain?" question. It's not that they're actively trying to undermine the proposal, they just don't know any better.
I've heard a number of suggestions that I think could work, but I honestly don't know enough to rationalize which ones would work, so I'm just going to list them in no particular order:
Ask members what level they're at - see Robert's answer there, but instead of doing this during Commitment, do it during definition, and don't allow votes from the beginner and curiosity crowds. (N.B. Anybody can still propose questions, the restriction only applies to voting). This relies on member honesty to no small degree but, I think most people are pretty honest.
Engage individual professionals/organizations to help define the sites. This might cost a lot of money, but it is very likely to produce good definitions.
Engage same professionals/organizations as curators, i.e. moderators of a sort, who could weed out the really poor questions popular with beginners and dabblers. This would probably have less of a capital cost but could still be reasonably effective at preventing crap proposals from going out the door.
I'm honestly not terribly crazy about any of the above suggestions, I'm just spelling out what I've heard. I do think that we need to somehow raise the bar for definition phase, and that forcing a commitment is not going to be enough unless the meaning of "commitment" is changed substantially.