Sometimes I wonder if people take classes in game theory just so they can 'invent' new web services.<p>Snarky comment "Let's end double blind scientific testing so you can know which of your test subjects really have the cure and which have the placebo!"<p>Perhaps less snarky comment. When you are looking to 'improve' a system, ask yourself what knowledge you 'know' and what you 'derive' and then ask yourself if more 'knowing' would affect the quality of what you 'derive.'<p>It would be great to know that some folks in your social graph are more influential than others. However knowing that, and steering more information through them, will change their position in your social graph (sometimes it will instantly convert them from very influential to not forwarding anything you send to them). Its a classic dilemma where more knowledge changes the behavior of the actors in the system which then changes the derived knowledge etc etc. You get a 'game' (at least for the folks who are inclined that way) and like moths to the flame they will start tweaking to optimize. And the optimizations may spiral out of control leading to bigger and more overt tweaks which then can cause the whole system to implode.<p>Sometimes not knowing is better.