[takeda]

The reason is only one. Having a website use Google service requested on random pages gives Google ability to get insight what pages users are visiting.

This goes together with other things Google provides for "free" such as jQuery and other interesting libraries.

The author don't seem to understand it, based on him comparing it to Google Reader and quickly dismissing it is not the data. They don't care about tracking page owner, they want to track the page users.

By providing a service that majority of websites use and that is essential for the page to work makes much harder to block through adblock, ghostery and friends.

This basically gives them an edge over any other competitors, whose only option to gather similar data is to make users install their toolbars.

[hackuser]

I wonder what percentage of pages, and of page views, ping Google? Does anyone have data?

[takeda]

I would imagine it is huge, there are even templates and frameworks that utilize their services.