"Attacks" are bad press. With a HTTP site, an ill-intentioned attacker could alter data in transit, making his name or logo appear in lieu of the intended page. That's not critical in any way for a site which hosts only public information (e.g. Wikipedia), but it looks bad nonetheless. With HTTPS, there is some level of "visible protection": the attacker will not be able to put his signature in the genuine site; instead, he will have to mount an alternate version with a fake certificate, at which point the browser will display a warning. It is all about making such attacks "obvious" in the eye of the public at large.
Another reason, less rational but probably more common, is the following "logic" as it goes in the minds of many managers: "Security is GOOD, it protects against BAD PEOPLE and EVILDOERS and TERRORISTS. HTTPS is security, so let's go sprinkle HTTPS everywhere we can." This is flawed in may ways, but this does not prevent people from thinking that way.
And there is fashion, of course. If Google goes HTTPS-everywhere, then it would be marketingly suicidal, and possibly bad taste, not to do the same.