New draft replaces the old Jumbo options draft that I did: draft-eddy-tcp-loo-04.html

I've posted some of the new patches to netdev - the kernel mailing list for such things:

• TCP options clean up
• Remove options logic from SACK calculations in tcp_input.c

More motivation:

• ArsTechnica: NebuAd ads inserted via "man-in-the-middle attack"
• Sweden approves wiretapping law
  

I wrote up the bits of the talk that I gave at allmydata.com that seemed to work the best:

http://code.google.com/p/obstcp/wiki/Reasoning

Last night's presentation at the allmydata.com Hackfest went well I think. I'm sure the audience were too polite and friendly to point out what didn't work but, none the less, I think I got a good feeling for it.

I'll be writing up my note from that presentation (which was slideless) over the weekend, hopefully.

Reddited

Well, Obfuscated TCP got Reddited (which is like a more intelligent version of a Slashdotting). I'm still processing all the comments, but it was very helpful to see from the point of view of people who haven't been thinking about this for a couple of months.

I've also been asked to talk about it tonight at a small gathering. If that goes well I'll write a better justification for ObsTCP from my notes for that talk.

Additionally, someone from the IETF pointed out http://tools.ietf.org/html/draft-eddy-tcp-loo-03, which is very close to my Jumbo TCP options spec. I'll be switching to using that draft as soon as I get a chance.