My perspective on things largely comes from my getting into some things unusually early and other unusually late. For example, I got into Unix when 16mhz Unix workstations cost $10,000 and there was no such thing as a 386 on which to run Unix. And as example of the second part, I'm still dabbling with startups.
Paul Graham, who I tend to strongly agree with and have thoughts provoked by, paints startups in a highly positive light. Recently he wrote something approximately that the worst thing that can happen going into a startup is that you'll come out a little wiser and a lot more experienced. Perhaps in his carefully self-selected finalists.
On the other hand, there's Pud's fuckedcompany.com, which (I haven't looked at it in years) made fun of the stupidity at startups. Net Slaves (fantastic reading, both of them) painted pictures of marriges destroyed and youth wasted and financial ruin.
Where's the truth?
Here are some startups I've worked on... igottago.com (defunct), debtreductionstore.com (briefly, as a consultant), Contact Designs (who seem to be doing fine now but started in the owner's house)... I think there may have been one or two others but I forget. Oh yeah, I'm working for one right now.
1. Owning a startup is completley different from working at one
2. Most startups I've been involved in (which might not be a good sample but I'm generalizing from anyway) have been funded by the owners at first with hopes of investment
3. Being paid by someone from their own pocket is awkward for you and for them
4. Being the first employee, or second employee sucks -- you're asked to do everything and do huge amounts of stuff, and the awkwardness of number 3 overshadows the awkwardness of trying to do work you're not prepared to do, which makes it hard to admit you're doing a poor job or describe why, as nothing can be done about it
5. Being the fifth or sixth or so employee rocks, as people start to move into their specialties at that point and there are generally enough people there that people aren't just floundering
6. Usually the owner, who has the money, is non-technical, contrary to the picture Paul Graham paints of two geeks working in their basement, and so the work isn't being done by extremely highly motivated technical jacks of all trade but by good but pretty normal programmers that got hired on -- ref number 3 and 4 again
7. Investment seldom comes, even if the product is great
8. The model of buying an office and staffing it with technical people and racing the clock to get investment before money runs out works piss poorly -- if your only expenses are rent and ramen and hosting, you might be able to do it, but office space and staff is too much burn for a product or site to become profitable on its own before money runs out in 99.9% of cases, and in most other cases, eventually there will be a market or product lull that runs the company out of money even after they've established their market
9. If things were awkward before, they get really awkward when the money starts to run out -- the owner has to start firing people, but it's seldom "I'm sorry, the money is out, I have to let you go", usually by that point his hopes have been so badly damaged and he's been so badly frustrated and stressed out, that the firing is personal, and it's vicious
10. People who start these things seldom learn from their mistakes -- they just hype things more, sell harder, and yell louder the second time around
The vast majority of "startups" out there are not Paul Gramahian startups.
I *love* the idea of a bunch of geeks crashing in an apartment, eating poorly, and churning out code, trying to make something that people want to use -- but of the many geeks I've met or known, approximately zero of them are interested in that. When the Atari Jaguar first came out, I wanted with every bone of my body (I was an Atari freak) to get the dev kit and play with making games for it -- I had learned my way around the internals of a few other systems, I thought the Jag was impressive (the PSX would later smear it using a far more traditional architecture), and I thought it was the perfect way for my creativity to shine, much like a thespian craves the stage in a specific role for a specific show. I talked to a lot of people who had kits, who were hanging out on USENET or IRC at the time, and no one would let me near their kits. Or really even talk to me. I couldn't get in the club. I've never personally known another geek who wanted something geeky so badly as I did -- well, except one, and he's now the CTO of a major software company. The point is, people with the kind of drive or even the interest in creating a Paul Graham-ian startup are exceedingly rare. The business guy who hires technical people scenario is more common by orders of magnitude.
So, if you come to me, and ask me if I join your startup, I'll have a lot of questions and apparent hesitation. Ask me if I want to join your small company that has a product and is profitable, I'll be interested. Ask me if I want to start a company with you and I'll be your friend.