There were guys in Columbia university who (supposedly) had a much better system than Mark Z, and are probably writing code in a cubicle at Google or equivalent.
I suppose Zuckerberg got his start because he knows how to code, but he is an excellent CEO, something neither of these 2 guys nor Friendster had. Do you really think facebook is winning because of a better architecture?
I despise facebook, and don't use it. But I recognize that Zuckerberg has been doing an excellent job in growing the company, seizing opportunities, thinking huge, and most importantly -- bringing the right people in at the right time.
I tend to agree with Fred Wilson about the three things a CEO does: develop vision, acquire talent, raise money. Zuckerberg has done all three very, very well.
Every single thing a user could do was based on FOAF-keyed permissions, and graph traversals were slow as hell.
In implementing Facebook, Zuckerberg redefined the problem space to give him an implementation advantage at every turn. Limiting it to manually-added colleges didn't just restrict the total userbase to be manageable -- it also introduced the concept of 'networks' that mostly eliminated the graph traversal problem. Photos were severely limited for a long time to constrain the storage problem. They also had a massive advantage in that they came of age in the post-bradfitz era of big consumer webapp architecture.
They had a chance to make it better when they re-wrote their app from Java to PHP no? (If... that is their main cause of failure. Which probably isn't.)
Friendster. With better architecture they are sitting on billions and Mark Z is writing code in a cubicle at Google...