Most startups fail. Bogus plans and an inability to close deals are probably just the signs that failure is coming for that company. As someone going to work for a startup, you're in much the same position as an investor. It won't be fun or lucrative to work for one that fails, so you're trying to predict which ones will succeed. In fact, it's even more important for you than for investors, because your portfolio con…
I'd go slightly further than that: look for founders who are working on a problem that grew organically out of their own experiences in an industry in which they have real domain expertise.
A lot of today's startups are doomed to fail because the founders are trying to "disrupt" industries that they really don't know anything about and have no relationships in. Ironically, however, the people who have deep knowledge about an industry and the connections to navigate it often aren't what many prospective startup employees would consider "technical" (i.e. they don't "code"), so they're dismissed, particularly in places like Silicon Valley.
Finally, it's worth pointing out that successful companies often fail to deliver attractive financial returns for employees beyond salary, so it's not enough to try to identify companies you believe have higher probabilities of success. You have to be able to negotiate a quality equity package with them. The average prospective employee is not going to be able to do that, especially in Silicon Valley, where:
1. Exorbitant valuations help justify miniscule equity grants for most employees.
2. A reliance on capital from professional investors (angels, VCs) results in dilution and the granting of preferences to certain classes of stock, leaving employee equity extremely vulnerable.
The best advice I could give to someone wanting to join a startup who is driven, in part, by the potential for financial return, is to instead seek opportunities to partner with folks who are deep in a particular industry, preferably unsexy, who wouldn't get the time of day from an engineer in Palo Alto because they can't write code and are probably older than 35. These people are not hard to find if you leave Bubbleville.