You might even want to contrast >> with >>=, since this is what is setting up the expectation in the first place. Note that unlike [>>=], [>>] is not an OCaml value..., or something.

@aantron They are recorded, but it's going to be so mangled it might as well not be the case. I added a slight precision. I also added more pointers to the manual.
Your formulation for >> is better, thanks!

Strange, why aren't there links to comments on outdated diffs?

Looks good, apart from the semicolon in the message for 533ef6f :)