Yvain has also written on this (though I can't find the post quickly): that bad ideas will tend to have better arguments for them than good ideas, because the bad ideas need good arguments more. Though I think that is more in the form you put it: that unaccepted ideas will generally have better arguments than accepted ideas.

Yvain's post was about popular ideas, not necessarily good ideas. In particular this rephrasing violates the law of conservation of expected evidence.

Yvain also fails to note that his argument also implies that over time the mainstream position will itself drift further and further away from truth towards whatever is most convenient for signaling.