A lot of the Web 2.0 startups getting started these days are of the variety where their risk is completely unquantifiable a priori. Mostly, these are highly social applications which require a large group of people to change their behavior slightly or to adopt a new behavior to work. The list includes peer-to-peer lending, social networks for recommending things, new group communication systems, downloadable photo sharing clients, etc. While some similar schemes have worked in the past (notably, MySpace, Flickr, EBay, Skype, Craigslist), the set that have failed are much much larger.
The problem with starting one of these businesses as an entrepreneur is that you basically have to get up the adoption curve before any rational VC will think about investing in you. Sure, if it works out, these businesses can have phenomenal network effects, but in advance of user adoption, it is nearly impossible to quantify the adoption risk in any way, and so it becomes a very very hard sell to VCs. Even if one partner believes that service will be adopted, he / she cannot convince the rest of the partnership with any data. As such, the few of these kinds of companies that get funded tend to either get funded because of a celebrity entrepreneur, or because of bubble-investing mania.
Basically, if you don’t have a track record and are starting a Web 2.0 company which requires a critical mass of users to do something that there is little evidence of people on the web doing to date, then you’re either going to have to bootstrap it for a while, give VCs some other proxy to go on (celebrity entrepreneur, core technology development), or find an individual angel who believes (easier said than done).
The flip side of this is that entrepreneurs who are doing hard-core technology development with relatively deterministic value tend to look at the successful businesses in social networks / P2P with tremendous envy. The latter seem to have unbreakable monopolies, organic growth, and no complex development requirements. However, this is just survivor bias at work. The odds of complete failure in the social / P2P businesses is much higher, which is not obvious just by looking at the winners.