When you operate at Google scale, thinking about morality means trying to anticipate and aggregate the total effect of your actions. Requiring a unified publicly visible and "real name" discoverable identity associated with all (Google-owned) online activity can have huge chilling effects on significant demographics - generally not the demographics deciding these policies, building these systems, or even the most profitable advertising demographics - but large groups of people nonetheless.

Whether that qualifies as "evil" is semantics, but excusing it with a comparison to puppy kicking isn't valid. It's a fundamentally different kind of issue and requires a fundamentally different kind of reasoning.

I've been using Google+ since the first day it was available; I figure about half the people I have circled are not using their real names.

[(People harmed by G+ policy) * ~1/2] is still O(People harmed by G+ policy)

(Also, presumably "half the people you happen to know who joined G+ that you circled" has some selection bias to it.)