Nits

[bifurcation]

Comment by @paulwouters

I personally don't like the underscored words in the html rendering of the document

It "blank" the new American spelling of "blanc" ?

has a corresponding _hash_ that summarizes the contents

I find "summarizes" a strange word here. It is not the hash contains a summary one can read.

[SECG] has no IETF equivalent? Should it?

proposaland -> proposal and

           Figure 13: Cleaning up after removing the third member

I would say "after removing member C" (avoid ambiguity about "every third")

   This section defines _tree
   hashes_, and _parent hashes_ are defined in [Section 8.9](https://www.ietf.org/archive/id/draft-ietf-mls-protocol-15.html#section-8.9)

I read this wrong the first way, the "," vanishes due to the unscores. Maybe: This section defines tree hashes, the next section defines parent hashes. Or alternatively, leave onlthe the first sentence in 8.8, then put the remainder and next section in two subsections (so 8.8.1 and 8.8.2)

For example, a malicious group member could send

Maybe: For example, a malicious group member could otherwise send

to previously compromised public keys.

Maybe: to public keys whose private key was previously compromised.

Post-compromise security is also provided for new groups

This reads a little odd as "new groups" would have no history and thus no previous compromise.

[bifurcation]

Filed #820

I personally don't like the underscored words in the html rendering of the document

I think it's important to emphasize the terms being defined, and this seems more readable than a definition list.

has a corresponding hash that summarizes the contents

I find "summarizes" a strange word here. It is not the hash contains a summary one can read.

Note that a hash function is often called a "message digest", in the same sense as the "Weekly github digest" that goes to the mailing list.

[SECG] has no IETF equivalent? Should it?

I think referring to SECG is still pretty state-of-the-art. But apparently RFC 8446 moved on from that, so I updated to use that representation.