Dave Winer correctly points out some serious flaws in twitter, and while he might not get brownie points or whuffie for it, he's right. He also shows where he thinks things could go. It's all about the metadata, and this subtle point seems to fly right past most people who read him.
I think it's time to explode twitter into its components making each of them public, implementable as a service, or as a hosted app, or something run on the end user's hardware.
The success of Twitter is because it allows for the rapid spread of messages from a controlled user base... there is a central control that can (but doesn't always) ban a user, etc. This means that every message is authenticated a bit... and all messages are tied to an identity. This makes filtering possible.
There is NOT any really good rating metadata. The messages are too short. This sets expectations, but really does cripple it for important stuff.
Blogging is seen as too slow, but RSS is a slow version of twitter. I think that metadata richness is the fix to this whole thing. Trade a bit of speed (twitter is too fast anyway) for expressive power.
We need to be able to aggregate our own stuff, which is one of the strengths of RSS. The ability to follow (and unfollow) an authenticated message channel is a great plus.
Collective ratings are ok, but the scales of services such as digg (up or down), and slashdot (funny is the same as insightful), are very limiting. It would be more useful for longer term (slower) conversations to be able to add more expressive metadata. Funny, Insightful, Biased, SelfPromoting, etc... could all be but a few of the plethora of possible bits of critical review that could be added.
It would be nice to be able to actually markup hypertext... but that still appears to be outside the range of feasibility.
Maybe it could be called TLM - Text Language Markup?