In a recent talk, I see that you gave some rough estimates for how many there might be in the universe.
Yes—very rough estimates. It's pretty straightforward. It's basically saying that every sunlike star should have one Earth, or close to it. And there are roughly 1011, or 100 billion, stars in our galaxy. Most of them are one solar-mass or less, and a lot of them are probably only half a solar mass, but those things probably have Earths orbiting them as well. So I'd say that, within a factor of two or something, roughly every one of those stars should have an Earth-like planet, so you're up to 1011 planets within our galaxy.
And then, well, let's talk about the whole universe. With its deep-field pictures, the Hubble Space Telescope showed—again, very crudely speaking—that there are maybe 1011 galaxies in the observable universe, and presumably those galaxies have planetary systems just like ours does. There's no particular reason to think our galaxy is an oddball, although you might start worrying about elliptical galaxies versus spirals and take out a factor of two or three, but we're playing with rough orders of magnitude here.
So I'd say there are basically 1011 galaxies, and each of them has roughly 1011 stars, meaning 1011 planets, so you're up to 1022 planets (10 sextillion) overall, in very round numbers—and that's a lot of planets. It's mind-boggling but it's inescapable. Even if Earths are a thousand times less frequent than I claim they are, you're down to 1019 (10 quintillion). So is 1019 not big enough for you? I hate saying the word, but these are literally astronomical numbers—numbers that make complete sense in terms of an astronomer's view of the universe but that human beings can't think about except, perhaps, in terms of how much money we're spending on the economic stimulus package. Trillion-dollar bailouts are getting us up toward the right order of magnitude.
Assuming our assumptions of planet formation aren't way off, how long will it take before Kepler turns up some Earths?
We'll find the hot Jupiters quickly and maybe a few hot super-Earths within the first year, I would guess. But the real treasure is the Earth-like planets—and, by definition, you can't get them until you've run for at least three or four years. You're looking for one-year orbital periods, roughly, for solar-type stars, and the first time the transit occurs, you think, "That's interesting." You've got one blip. You need at least two blips to say, "Now I have an orbital period, let's see if there's a third blip at the right time." So if you get the third blip you say, "Hmmm, that looks good, I've got blips that are separated by roughly the same amount of time, maybe I'll just be really cautious and wait for a fourth blip." And if you do that then you're up to a few years of observing time. So my feeling is that by 2013 we'll have some Earths to announce.