NIP-10 formalizes a convention for the use of e-tags. The convention establishes a positional meaning for those e tags. The first e tag is the id of the message being replied to. The last is the root of the thread containing the reply. Any intervening e tags are simply citings of other relevant messages.
Unfortunately this leads to a nasty ambigutity.
An event that is the root of a thread, that also cites another message, looks like a reply according to NIP-10, even though it is not a reply.
Worse a root event that cites two other messages looks like a reply with a root according to NIP-10. And that could lead to some severe data corruption problems in clients that conform to NIP-10 since the two cited messages are not likely to be on the same thread, much less the actual reply-to and root.
I can see no way to resolve these issues other than to eliminate the positional meanings of "e" tags as described by NIP-10 and replace them with specific "reply-id" and "root-id" tags.
Thoughts?
