First, approach choosing between startup schools much the way you approached choosing between universities (or colleges... I'm Canadian). When I was choosing a university to attend, I had a choice between the top computer science school in Canada, the University of Waterloo and the top rated overall school in Canada -- a misnomer as this was simply determined by a well-read publication in Canada, Maclean's -- the University of Toronto. The U of T's program was not bad; far from it. The curriculum, insofar as I could tell, mimicked Waterloo's. I was accepted into both co-op programs, though Waterloo had the edge here as it seemed to have the better reputation. Socially, I had a girlfriend and going to the U of T would mean I could continue a relationship with her (being a dumb teenager geek I gave much more importance to this factor than I should have). Financially, I pretty much would have to work my way through university. My parents are not well-off and while they had saved up some money to get me through 1st year, subsequent ones were financially up-in-the-air. When the U of T gave me a generous first-year scholarship and Waterloo didn't -- I had messed up Algebra and had earned an average score on Waterloo's math contest -- my decision was essentially made. Both schools, in my check-point analysis were neck-and-neck. Academically speaking they were both good, although Waterloo was probably better for my subject. In terms of co-op, Waterloo was better but U of T was not far behind either. So, knowing that I was essentially settling for second-place, I went to U of T.
That turned out to be the biggest mistake of my life. I don't regret a lot of mistakes I have made, but this decision is one. You see, the difference between first place and second place, while it might appear to be miniscule, is actually immense. Don't get me wrong. I was very much correct in my analysis. Academically, I was learning the same stuff that my friends over at Waterloo were. We were both kept busy working on projects, learning the same algorithms, and pulling our hair out over the tough math courses. For a while, I was quite happy with my decision. But there was an unsettling feeling. I didn't seem to have as many interesting peers for one. I was hard pressed to find somebody that was genuinely interested in programming or computing theory. I gave up asking people why they had chosen to study CS because I was tired of hearing "because that's where the jobs were" (we're talking just before the dot com bubble burst folks). I was further disappointed when it came time to do my co-op term. My co-op facilitator just didn't seem to get it. I had been courted by a startup (of 7 Waterloo grads ironically) and I wanted to work there. But she wanted me to go work at a "major company" like IBM and actually laughed at me when I said I'd already signed the papers in January with this startup to work for them in May. The big problem was that I had found this position on my own, not via the school. They had agreements with a bunch of companies, but mostly B players like IBM and ones that I just was not interested in. Waterloo had skimmed the better companies like Microsoft (back then guys!) and RIM for its students and left the U of T to settle for seconds.
So, my peers didn't seem to be as focussed on CS as the ones my friends told me about at Waterloo, and my co-op experience had turned out to be second-rate. The reality of my decision was beginning to set in. But what really convinced me I had made a catastrophic mistake was what happened one day after being really impressed with a 2nd year computer architecture (i.e. let's all learn assembler) lecture. While everybody dashed out of the lecture as soon as it was done having been bored silly, I rushed to the front and caught the prof while he was shutting down his laptop. "This is really interesting, the only problem is I can't actually program anything on my computer with it because, well, it's in imaginary ISA. I've heard x86 is ugly but can you recommend me some good x86 tutorials?". I had anticipated that he might say "Not off the top of my head, but I'll email you some good book titles/websites" or "uggh x86, why not learn PowerPC it is so much more elegant". I had NOT anticipated what he actually responded with, which was "Sorry, I don't know x86...actually the ISA in the textbook is the only one I'm fluent in". I was absolutely shocked. The only ISA he was familiar with was an imaginary one that served no practical purposes. Gutted, I left. You see, those are the sort of professors that can get along perfectly well at a 2nd place school.
Ok, so if you're still reading, you're probably wondering what the moral of the story is. Well, first off, ignore financials. Those ended up being the least of my problems. They'll be the least of your problems too. The most important things you can get out of startup school are:
-being introduced to the most interesting and intriguing peers
-having access to top-notch business opportunities
-and being mentored by passionate people
Being in my mid-20s I can call this one from the gut. You need to go with the original startup school, the #1. It's the only one that will get you all 3 of those. Sure, you will meet interesting people at other startup schools, but not THE most interesting ones, the cream of the crop. Similarly, you're just not going to have access to as good of a business opportunity in a tier 2 startup school (e.g., just pickup Founders at Work and look at all those RECOGNIZABLE leaders). And finally, you're just not going to have access to as passionate a set of mentors as you would have with YC. Again, proof of this is in the pudding: Founders at Work, PG's essays, this website (for crying out loud, another school used a freaking blogger hosted solution for their site and ripped off YC's application form). Just trust me guys. Don't let my mistake be in vain :-) There is nothing worse than wondering where you could have been if you had been able to maximize your potential by running with the lead pack. Miniscule differences are magnified over time, so don't settle.