As a serious answer; Lots of web-based revenue sources are volume based. The more they grow the more they can make. In advertising this is somewhat exponential - the more popular your service, the larger CPM (cost per thousand page views) advertisers are willing to pay (all other things being equal & to a point).
As such, if you don't need a revenue source when you start, it makes sense to build your business first and add revenue later. You could make more in your first month of revenue than all other previous months combined. Twitter could easily add some combination of advertising, premium accounts or premium customisations at any point.
However, whether that's good in practice or not is another matter...
How would they make money in the long run? Sell pro tools and statistics... similar to the way Feedburner makes money. Just study Feedburner and you'll see plenty of parallels.
Until today it's funded by Evan Williams who bought Twitter and Odeo from it's previous investor last year.
What's different about Evan's story that compelled him to repay them?
There are two models I can imagine for Twitter: 1) working in advertising, or 2) become a feature of some larger business.
You might also speculate that there would be some permutation of Twitter that would be interesting enough to become Twitter Pro or somesuch. But those kind of things are not often very exciting.