"You don't 'diversify away' from a good investment." -- that's exactly what you do. Given that there's nothing as a 100% sure good investment, you diversify away from investments that seem great to distribute your risk.
Let's say Company X has a 10% chance of being worth nothing, and 90% chance of being worth five times what it is today. Is that a good investment? Of course. Would you want to bet a billion dollars on it? If you had $20 billion, sure -- if you had $1 billion, probably not.
You can easily create a real-world analogy, a dice game where you pay me $X and roll a single die. I keep your money if you roll a six, but give you five times your money back if you roll anything else. You can only play once. How much do you bet? I hope you do bet something, because it's a very favorable game to play, but I also hope that you don't sell your car, house, everything in your refrigerator, and your dog to wager everything you have.
The IPO price is not really relevant since he couldn't sell at that price -- he sold nearly everything at the first chance he had, which was wise of him regardless of whether Facebook's stock price goes to $0 or $500.
The bottom line is that this has nothing to do with Facebook and is based solely on the mathematics of random variables.