First of all, wanting to get into social is accepting to play second fiddle to Facebook - google is not a leader but a follower in this game. They're known for innovating and this seems to be at odds with that.
Second, as the author points out, Google is not very good when it comes to people and how they work play and interact. This is what social is all about, and it's very hard to put into an algorithm which is the Google way of doing things.
Third, I agree that making all bonuses dependant on the success of their social endeavours is a huge mistake. What happens to the really good algorithmic search people at Google? Do they bust their ass of doing what they do best, or do they try to implement some kind of social search that they know nothing about because that's what their bonus depends on. What will happen to my search results because of this? Yes, it will get worse, and yes it will undermine Googles core product. What Larry has done is basically told everyone that they should start working on something Google is historically not very good at and not worry about their core business. Huge mistake.
Fourth, this shows a poor understanding of the dynamics of incentives, which is something that a CEO of a company as large as Google should be very good at. Paying everyone a bonus based on how one division performs is at best disillusioning to the people who don't work in that division and have absolutely no control over their bonus. It doesn't matter how hard a division works if it doesn't have anything to do with social. What will this do to their morale? At best they'll just hum along, knowing there's not much they can do, at worst they'll start cannibalising their product to try and make it social.