It's also full of inaccuracies. The "interesting details" are largely wrong. Unfortunately it's written in an authoratative style, so people assume it's true and upvote, demonstrating one of the major flaws in vote based systems.
The most common decisions cited are:
* Realizing that search was important.
* Focusing on speed.
* Maintaining corporate control.
* Focusing on the user experience and deferring revenue.
* Scaling out computation over a large number of cheap computers.
* Hiring very smart people.I believe that a good Q&A site should use "voting" to recursively find smart people who are similar to you and can provide you with good answers....as opposed to showing the most popular/well-written/pithiest answer. "PeopleRank", maybe.
The problem is preventing this leading a site dominated by a group of people who share a set of extreme opinions of one sort or another. There's no now site where you can't get karma for reinforcing some commonly held group opinion and if Karma became self-reinforcing, the problem would get harder.
It might not be impossible - maybe you could have secret algorithm for rewarding people only for making well-respected non-polarizing posts. But it would be quite tricky.
The Q&A PeopleRank idea will probably be my next start-up, if Vark doesn't beat me to it. Reading http://blog.vark.com/?p=352, it seems that they're on the right track. (Although their UX is a little clunky).
However, even if Vark builds a kick-ass product, you can still apply the PeopleRank idea to news-sites like HN.
Anyone interested in building a Q&A/news site feel free to message me.