Hmmm...seems obligatory to do a first post. I wanted to comment on a couple of articles, but it seems that discussion is limited to at least apprentice level (assuming I'm not missing an obvious "reply" button). Oh well. Here's my input, both based on the nuggets I picked up at the last IETF meeting I was at:

• TCP Splicing -- Seems like an interesting area of pursuit, although I'd say that many NAT box administrators consider the idea of inbound TCP connections not working a big plus. My rudimentary grasp of TCP negotiation isn't going to make my critique any easier, but this idea should probably be floated with the folks who set up the MIDCOM BOF at the last IETF (presumably with the goal of forming a working group).
• Simple Multiplexing Protocol -- This idea came up so much, the IETF actually got around to defining a couple protocols for this domain space. There's SCTP, which is on the same layer as TCP, and provides framing and reliability without enforcing order, and there's BEEP, which is currently under the name BXXP until the IETF actually advances it to Proposed Standard (which is RSN). BEEP tries to do for protocols what XML did for file formats: it defines structure and a ruleset. It would have been really handy to have had it been around when we were defining RTSP.