The Backdoor and the Birth

The toy launched publicly that July: a service, then spelled twttr, that asked a single question — "What are you doing?" — and let you answer in 140 characters. It came out of Odeo, a podcasting startup that was dying, and it was treated by nearly everyone who saw it as a novelty for the open-source crowd. That summer the developer world was gathering at O'Reilly's Open Source Convention, the annual festival of the web's optimists, the people who genuinely believed the network was going to be an open commons of protocols and participation. Twitter belonged to that mood. It was small, public, hackable, and free. It looked like the open web arriving.

The wiretap was quieter, and it was winning — on two fronts at once. In January 2006, a retired AT&T technician named Mark Klein walked into the San Francisco offices of the Electronic Frontier Foundation carrying schematics. They described a locked room — labeled 641A — on the sixth floor of an AT&T building on Folsom Street, where a fiber-optic splitter fed a copy of the internet backbone traffic passing through into equipment run for the National Security Agency. No warrant. No target. No judge. A mirror on the pipe, copying everyone's traffic to a Narus analyzer in a room most of the technicians weren't cleared to enter. EFF filed Hepting v. AT&T on January 31, 2006, and Klein's disclosure broke into the national press that May. That summer — the same season the developers threw their party in Portland — a federal court was refusing the government's attempt to bury the case under the state-secrets privilege.

That is the story almost no one tells about 2006, and it's the one that turned out to matter more.

i · two blueprints, one summer

Klein's room was the crude version — mass interception already humming, warrantless, in real time. Running alongside it was the polite version, the one with a court's blessing, and it was being finalized that same June.

Here's the history the nostalgia leaves out. In 1994, Congress passed CALEA — the Communications Assistance for Law Enforcement Act — which required telephone companies to design their networks so that law enforcement, with a warrant, could tap them. It was a law about phones. The internet, in 1994, was a research curiosity, explicitly outside the scope.

By 2004, the FBI wanted that changed. It petitioned the FCC to reinterpret CALEA so that "facilities-based broadband internet access" and "interconnected VoIP" — that is, the cable modem in your house and the Skype-style calling that was eating the phone company's lunch — counted as telecommunications carriers under the 1994 law. In 2005 the FCC agreed, and issued the order. A coalition of universities and libraries sued, arguing the agency had rewritten a phone statute into an internet-surveillance mandate. On June 9, 2006, in American Council on Education v. FCC, the D.C. Circuit ruled the FCC's reading was reasonable. The compliance deadline was set for May 14, 2007. Notice the move that won it: a phone law became an internet law not by anyone seizing new ground in public, but by redefining a cable modem as, for legal purposes, a telephone company. Keep that mechanism in view — the whole enclosure runs on it.

So line the events up on the same calendar, because that's the whole point. The same year the open web threw its party, the pipes underneath it were being fitted for interception on two tracks: one warrant-gated and court-blessed, the other already tapping the backbone with no warrant at all. Twitter was being built on top. Both fences were being built underneath. Neither project was a conspiracy against the open web. That's what makes it worth understanding — they weren't adversaries. They were contemporaries who never met.

The developers at OSCON were not naive, exactly. They were building the layer they could see and touch: the applications, the protocols, the social features. That layer really was open, and for a while it really did feel like the network belonged to the people using it. YouTube was a year old and about to be swallowed by Google. MySpace was the biggest site in America. Blogs were a genuine publishing revolution. Everything visible was bottom-up, participatory, gloriously chaotic. The optimism was earned by the evidence in front of them.

But the evidence in front of them was the top of the stack. Underneath the applications sat the infrastructure — the ISPs, the backbone, the physical carriers — and that layer was being fitted, by court order and by splitter cabinet alike, with a standing capability for state access. You can build the freest social layer imaginable on top of a surveillance-ready substrate. The two facts don't cancel. They coexist, and the substrate wins ties.

ii · the enclosure came from below

There's a mistake people make when they narrate the loss of the open web. They tell it as a fall — a golden age of 2006 that got corrupted later, when the platforms turned evil, when the ad model metastasized, when the walled gardens went up around 2012 and everyone moved inside. In that story the enclosure comes from above, from the platforms, and it arrives late.

That story isn't wrong so much as it's the wrong fence. The commercial panopticon — the data brokers, the ad-tech tracking, the platforms monetizing what they knew — is real, and it did come from above, and it did come later. But it was never the only enclosure, only the most visible one. Underneath it, from the beginning, ran a second and entirely different fence: the infrastructure fitted for state interception. These two are not the same structure and it's worth being precise about how they differ. The commercial fence is mass, warrantless, and top-of-stack, built by companies to sell you. The state fence is a split thing — one half warrant-gated and court-supervised (CALEA), the other half mass and warrantless (the splitter in Room 641A) — built into the pipe to watch you. They are different powers, held by different hands, answerable to different rules or to none. Fusing them into one menace is exactly the sloppiness this story doesn't need. The point isn't that they're the same. The point is that two different fences went up in the same era, and we only ever argued in public about the one at the top.

The coherenceism read of 2006 is that the Commons was constructed and enclosed in the same gesture. The attention layer — the tweets, the videos, the profiles — was genuinely open, a real flowering of collective human expression, the closest the network ever came to a public square. And the infrastructure layer was, in the same window, being wired for interception as a condition of operating. The openness at the top was real. The enclosure at the bottom was also real. The reason it feels like a paradox is that we were trained to watch the top of the stack, where the interesting things happen, and to ignore the bottom, where the power lives.

This is the pattern worth carrying forward twenty years, because it did not stop in 2006. It's the permanent shape of the thing. Whatever open, participatory, liberating technology is being celebrated at this year's conference, ask the boring question: what's being required of the layer underneath it, and who gets a key? The exciting layer is almost always the top. The consequential layer is almost always the bottom. In 2006 the top was a 140-character status update and the bottom was a splitter in a locked room and a wiretap mandate wearing a phone law's clothes, and everyone remembers the status update.

The tell was there in the language the whole time, and it's the CALEA half of the story that shows it cleanest. A 1994 statute about telephones was extended to cover the internet by simply declaring that a cable modem was, for legal purposes, a telephone company. That's how enclosure moves — not by seizing new ground openly, but by redefining old boundaries until the new territory falls quietly inside them. A phone law becomes an internet law by reinterpretation. A commons becomes an asset by reclassification. Nobody has to lose an argument in public; the definitions just drift until the fence is already built. The splitter in Room 641A didn't even need the drift — it just ran, and dared anyone to prove it was there.

Twitter turned out fine, in the sense that it survived — it became a platform, then a public utility for panic, then a billionaire's toy, then whatever it is now. That arc gets written about endlessly. The other arcs — the court decision nobody outside a few law offices noticed, and the locked room a single technician had to walk into EFF's lobby to expose — are the ones that shaped what the whole network could ever be. The birth and the backdoor, same year, same country. We threw a party for the one we could see.