That said since a (very) long time I'm using separate Linux user accounts to: check my professional email + G+, surf my personal email + G+ + FB (my FB is using a fake but plausible name) and a third one to surf the Web.
The one surfing the Web is linked to a fake online identity: entirely made up, with fake friends / fake G+ circles, fake StackOverflow / OpenID and basically fake everything.
I then only ever surf using a transparent proxy for anything "work related": the IP can't be linked to my fake IP.
It's not difficult to set up: I did set up the transparent company Web proxy (VPN would to too) myself and basically Linux user accounts take care of the rest.
Now I'll start using different browsers too and, why not, maybe Tor in one of the account.
I take it I could take all this a step further and whitelist websites that my "personal" account is allowed to connect to (using iptables' owner-uid mod).