Sounds like survivorship bias to me: yes, every opinion now accepted was once eccentric, but most eccentric opinions of the past have been rejected so thoroughly that we're unlikely to have ever heard of them.
However, if we accept the premise of the quote that every opinion now accepted was once eccentric (and the implicit inference that every opinion that will be accepted in the long run is now eccentric), and your goal is to have opinions that are accepted in the long run, then it is true that you have to pick opinions that are currently eccentric (even if it is also true that most eccentric opinions, both in present and past, are crazy ones that never become accepted).
The main problem is that the premise is false. Most conventional opinions of the past are still accepted today (e.g. "clouds lead to rain", "fire is hot", et cetera).
Betrand Russell's Ten Commandments for teachers.
I find this to be of use not just for teachers but for rationalists in general. #8, especially, is an especially eloquent formulation of Aumann's Agreement Theorem.