I wonder how many websites this will break. A few times in the past I have tried seeing if websites support https and been served a default Apache page or something instead. Still, the more encrypted traffic the better.
That's anecdata, but to get good data one would have to try to fetch & compare results from both protocols for many websites - perhaps Google or the Internet Archive has done this?
The biggest annoyance for me in using https preferentially is that I often end up with multiple bookmarks for the "same" page, which differ only in their protocol - it would be nice if there were an auto-magic way to upgrade the old http bookmark to the https protocol.
You can see all the rulesets here: https://gitweb.torproject.org/https-everywhere.git/tree/src/...