I attended to pitch in on JSON and OAuth work, and because it was here in Vancouver. But this meeting was really about defending the Internet from those attacking it. Which is worth everyone’s attention and deserves more explanation than I’ve seen in the mainstream media.

Having said that about the mainstream, Besieged (in The Economist) is not terrible.

The Flavor of the IETF · If you read The Tao of IETF you’ll know most of the things that matter, and if you care about the Internet you likely should. Let me add flavor:

  • You see a lot of grey beards, from which you can deduce two things: First, some of the people here have been designing the Internet for a long time, and second, the representation of women is, as in most information-technology contexts, dismal.

  • The term “IETF greybeard” is sometimes used disparagingly. I’ve heard this from women lamenting the syndrome, common in so many milieux, of the unofficial cabal of old white guys who run everything. Second, from people with interesting or surprising new ideas which are getting nitpicked to death at endless length by said greybeads.
    [Disclosure: I have a grey beard.]

  • In every session, a high proportion of people have laptops or tablets lit up. On a high proportion of those screens, one or more windows contain the colorful monospaced visual rhythms of code.

  • A small but significant proportion of the people doing the work and attending the meetings have received funding from national intelligence agencies. An even smaller but nonzero proportion are employees of those agencies.
    [Disclosure: A company I co-founded sold software to the NSA while I was there; I have been inside Fort Meade and spoken at a large intelligence-community conference.]

  • The IETF moves slowly. Maddeningly so. On the other hand, given the importance of the Internet to human civilization, you don’t want to fuck with it incautiously. So the speed may be appropriate.

  • The IETF famously operates on the basis of “rough consensus and running code”. That consensus is very rough, sometimes. Given any particular piece of work the IETF is pursuing, it’s not hard to find a participant — a greybeard, even — who will explain at length why that work is misguided and dangerous. As for the code... well, you’re using it right now to read this.

IETF 88 HTTP-encryption session
· · ·
IETF 88 Technical plenary audience

Shared Beliefs · Remember about the consensus being rough. Having said that, I feel pretty comfortable saying that the people who build and maintain the Internet in general, and a whole lot of IETF participants in particular, feel that:

  • Pervasive surveillance, of the kind revealed in the Snowden-sourced documents, constitutes a misguided and damaging attack on civic society in general and the Internet in particular.

    Targeted surveillance of individuals, which in civilized societies requires approval from a judge, is not what we’re grouchy about.

  • To quote Snowden: The crypto is good but the end-points are weak. So the endpoints are where work to increase the security and privacy of the people who use the Net will likely have the best payoff.

  • Whenever we increase the proportion of Internet traffic that’s encrypted, we increase the cost and decrease the utility of pervasive surveillance. So let’s do that.

What Bruce said · I mean Bruce Schneier, probably the world’s single most visible cryptographer and security geek. Back in September, he called, in Take Back the Internet, for the IETF to “dedicate its next meeting” to these problems; that essay is worth reading.

So he got an invitation and appeared at the “Technical Plenary”, our all-hands meeting which is often less-than-gripping but on this occasion pretty well filled the largest room the hotel had.

IETF 88 Technical Plenary

At the technical plenary.

I captured some soundbites from that session; most but not all are from Bruce, but unfortunately I neglected to note the speaker:

“We’ll probably never know which products have been subverted.”

“The loss of ephemeral conversation”

“Amazing as it seems, the NSA had no contingency plan for the leakage.”

“Corporations have a cost/benefit analysis. Pre-Snowden there was no cost to co-operation. Now it’s different.”

“The goal is to make eavesdropping expensive.”

“We probably won’t win the stop-doing-this argument, but might win on tell-us-about-it.”

“More people should start running Tor exit nodes. Nobody’s ever been prosecuted.”

“We can not have a free society under self-censorship.”

“Pervasive surveillance is an attack.”

TLS · Transport Layer Security I mean, the technology that’s in play when you see the little lock in your browser bar. It means two things: First, that the communication is encrypted and hence private; Second, that you can be pretty sure that the address in the browser bar is really who you’re talking to.

I discussed this in Private By Default, and argued that that’s how everything on the Internet should be.

I think that a lot of IETF participants would agree; and almost everyone who understands the issue would agree that it would blow a hole in the whole pervasive-surveillance thing.

But there’s a lot of pushback against making TLS compulsory: I hear arguments that it’s too expensive, too complicated, screws up load balancing, may exclude poorer parts of the world, and so on. I think those arguments are largely wrong, but I acknowledge that they exist and that the chances of switching the whole Net over to private-by-default, at least in the short term, are pretty remote.

Opportunistic Encryption · This was probably the biggest idea floating around at IETF 88. Right at the moment, if a URL begins with https: you get TLS and if it’s just http: you don’t. But why should it be that way?

There are proposals to deploy technology where, even when you hit an http: link, the infrastructure quietly turns on TLS.

Now, it might not be quite first-class TLS; while the privacy promise would still stand, your confidence about who you’re really talking to might not be as strong. But it would still drive up the cost of pervasive surveillance. Maybe by really a lot.

Stay tuned. It might turn out not to work, and the greybeards might nitpick it to death. But maybe not.

The IETF matters · It’s horribly flawed in lots of ways, but it’s essential in getting the Internet built and maintained. I’m lucky to have had a chance to help out in tiny little ways here and there, and hope to again.


Comment feed for ongoing:Comments feed

From: Dave Walker (Nov 11 2013, at 05:01)

Video of the session is up at http://www.youtube.com/watch?v=oV71hhEpQ20&feature=share&t=23m30s . I watched it the other night; Schneier excels himself with one of the most awkward-truth-dense presentations (and this is meant as a compliment) I've heard in a very long time. Recommended to any readers who haven't seen it yet.


From: Fazal Majid (Nov 11 2013, at 08:15)

Before making TLS the default, we should fix glaring bugs in it:

1) cipher suites lacking perfect forward secrecy

2) the broken certificate authority system, as evidenced by the Diginotar and Comodo fiascos (and the NSA ones we probably don't know about)

3) known attacks like BEAST

Most of the fixes, e.g. TLS 1.2, are stymied due to compatibly issues with the installed base. The CA issue is probably the toughest one. Perhaps we should require certificates signed by at least two CAs, one from the West and one from Russia or China. It would require joint agreement from the NSA and FSB to issue a forged one.


From: John Cowan (Nov 11 2013, at 09:29)

Given how lax CAs are about issuing certificates to anyone in any name at all (hey, nobody is paying them for certificates they refuse to issue), you should have zero confidence that you are talking to the Right Party based on TLS.


From: len (Nov 18 2013, at 10:38)


Good. That's the right lesson. Your credibility goes up a few hundred per cent.


author · Dad · software · colophon · rights

November 10, 2013
· Technology (85 fragments)
· · Internet (114 more)
· The World (117 fragments)
· · Politics (165 more)

By .

The opinions expressed here
are my own, and no other party
necessarily agrees with them.

A full disclosure of my
professional interests is
on the author page.