I have come to the conclusion that the MSN Messenger protocol is terrible. This because there is not a single implementation of it that works nearly as well as any implementation of the competing protocols - even Windows Live Messenger drops the connection or fails to sign in sometimes, and that comes straight from the source. (Incidentally, meebo.com (which is pretty much text-only) is fairly good - its been a while since I tried Trillian, but I don't remember any problems with it either (though it is also limited in feature set)). I've tried aMSN - it's got the same problems, but worse. Kopete, my primary messaging program, logs into my GTalk and AIM accounts almost instantly, but usually requires two or three pushes to log into MSN (and even then, takes a few seconds). Now, considering these experiences, I put the blame for flaky MSN performance on the actual protocol - admittedly, AIM and Jabber (GTalk's base) are more open, so external developers will have an easier time implementing them, but the fact that even the official Microsoft client is flaky lends support to my theory that the fault is in the protocol.
Why does this matter, you ask? I live in Canada - the most common messaging protocol in Canada? MSN. The solution: Microsoft does one of their major updates, that forces everyone to move to the new messenger, or not be able to talk to anyone (it has happened) - this update will include a backend update to a completely new protocol. Base it off something like Jabber (which is open-source, and stable), and open-source the extensions they add to it to support all their features. This should give them a stable base to work from. Also, work with Google - make it interoperate with GTalk (also Jabber based) (IM should be like email or the phone system - any vendor's product can talk to any others - this would be a step in the right direction). (Incidentally, I haven't used it for a while, but I loved the UI on the Google Talk desktop client - if they released a Linux version that could talk to MSN, I'd ditch Kopete in a second).
Anyway, that's what I want to see in IM tech - anyone else have anything they wish Microsoft (or AOL, Yahoo, or Google) would do?