Privacy constraints impair Google innovation
Google was becoming too big to manage, with far too many bits and pieces which could in theory help the broader company but which in practice, like Reader, just sat there using up resources and contributing very little in return. So Larry Page decided that he would start killing them off, and making Google more focused; I'm sure that decision was made easier by the fact that if Google now needs to control the amount of information it collects about people, it can't have engineers freewheelingly making unilateral decisions to start collecting exactly that kind of information. Dick Costolo's ideas were probably great in 2005; in 2013, they would be politically suicidal.
The result is that Google is going to be less of a utility, less of a public service, and more of a company with a constrained set of products. The problem with the death of Reader is that it was the architecture underpinning lots of other services -- the connective tissue of just about all RSS readers and services, from Summify to Reeder to Flipboard. You didn't even need to use Google Reader; it was just the master central repository of your master OPML list, all the different feeds that you were subscribed to. Google spent real money to provide that public service, and it's going to be sorely missed. As Marco Arment says, "every major iOS RSS client is still dependent on Google Reader for feed crawling and sync."