+* Eric has a great Maildir+inotify-based Bayes training setup
+ going back many years. Document, integrate and publicize it for
+ public-inbox usage, encouraging other admins to use it (it works
+ as long as admins read their public-inbox).
+
+* Custom, difficult-for-Bayes requires custom anti-spam rules.
+ We may steal rules from the Debian listmasters:
+ svn://anonscm.debian.org/pkg-listmaster
+
+* Full archives are easily distributable with git, so somebody else
+ can take over the list if we give up. Anybody may also run an SMTP
+ notifier/delivery service based on the archives.
+
+* Avoids bikeshedding about web UI decisions, GUI-lovers can write their
+ own GUI-friendly interfaces (HTML or native) based on public archives.
+
+Web notes
+---------
+
+* Getting users to install/run any new tool is difficult.
+ The web views must be easily read/cache/mirror-able.
+
+* There may also be a significant number of webmail users without
+ an MUA or feed reader; so a web view is necessary.
+
+* Expose Message-ID in web views to encourage replies from drive-by
+ contributors.
+
+* Raw text endpoint allows users to write client-side endpoints
+ without hosting the data themselves (or on a different server).
+
+What sucks about public-inbox
+-----------------------------
+
+* Lack of push notification. On the other hand, feeds seem popular.
+
+* some (mostly GUI) mail clients cannot set In-Reply-To headers
+ properly without the original message.
+
+Scalability notes
+-----------------
+
+Even with shallow clone, storing the history of large/busy mailing lists
+may place much burden on subscribers and servers. However, having a
+single (or few) refs representing the entire history of a list is good
+for small lists since it's easier to look up a message by Message-ID, so
+we want to avoid splitting refs with independent histories.
+
+ssoma will likely grow its own built-in ref rotation system based on
+message count (not rotating at fixed time intervals). This would
+split the histories and require O(n) lookup time based on Message-ID,
+where `n' is the number of history splits.