How much goes in the first release?

As little as possible, but no less. The feedback you get from being in production is absolutely priceless. You learn qualitatively different things from production than you ever do in GreenfieldDevelopment [thanks to JoelSpiegel for this insight). This implies that DevelopmentSupportsProduction, which is also a little extreme, but necessary.

Okay, that's the case for making the first release smaller. How small is too small? Too small is if you don't make any money for Business. (There are lots of other possible criteria, but this one works nicely for me. I've seen too many "first releases" that were annoyingly good but didn't actually help. These projects are generally cancelled. I generally try to avoid having my projects cancelled.)

So, you've got a pile of story cards on the table that the team has estimated will take 19 years, 4 months, and 6 days to implement. Business is a little pissed. Listen carefully for the phrase, "Well, if you could just do this [accompanied by grabbing a handful of cards, mostly from the high priority pile], then we'd have something." Bingo, you win. That's the first release.