The Web We Got

The House voted no.

On June 8, 2006, Representatives Ed Markey and Anna Eshoo offered an amendment to the COPE Act — a telecom deregulation bill — that would have added enforceable net neutrality protections to federal law. The amendment was defeated 269-152. The pipes were handed to the telecom industry to configure as they saw fit. And the infrastructure question that could have gone differently was settled, quietly, with a 117-vote margin.

Twenty years later, we live in the internet that decision helped build. Not the only cause. But the original fork in the road.

i · the day the pipes became a toll road

The telecom industry was not subtle about its intentions. AT&T CEO Edward Whitacre had said, in terms that would sound absurd if they weren't so clearly stated, that companies like Google were "enjoying" the internet "for free" — using his pipes — and that this was going to change. BellSouth's executive vice president William Smith made the same argument in different packaging. The thesis was simple: telecoms built the infrastructure, content companies made money on it, and the telecoms wanted a cut. Not just from consumers as a subscription fee — from businesses, as a toll. Two-tier access. A fast lane for companies willing to pay extra, best-effort service for everyone else.

This wasn't speculation by advocacy groups. These were executives, on the record, describing what they planned to do.

The opposition to Markey-Eshoo called these concerns "speculative." No marketplace problems existed yet, they said. Regulating against harm that hadn't happened would constitute "unprecedented regulation" and discourage broadband investment. The internet was working fine. The policy statement in the COPE Act — vague, unenforceable, expressing support for an "open internet" in principle — was sufficient, they argued. More than sufficient. The market would handle it.

This was technically accurate in the narrowest possible sense and functionally dishonest. The marketplace problems didn't exist yet because the executives publicly describing their plans hadn't implemented them yet. The amendment was trying to prevent a stated, on-the-record intention from becoming operational reality. The opponents were asking for proof of fire damage while the architects of the plan were on television explaining which buildings they intended to burn and why.

269 members of the House voted that the concerns were too speculative to act on. The specific company whose CEO had described the toll-road model on camera was AT&T, which spent roughly $225 million on lobbying in the years leading up to this vote. The concerns were speculative. The lobbying was not.

Markey's office documented the gap clearly: the COPE Act's policy statement contained no enforcement mechanism, no complaint process, no rulemaking authority, and no penalties. It was a gesture in the direction of an open internet, followed by a vote against the rules that would have made it real.

ii · the prediction was always there

What followed over the next two decades was a slow-motion vindication of everything Markey-Eshoo tried to prevent.

Comcast began throttling BitTorrent traffic in 2007 — not secretly, but in a way that took users and researchers months to fully document. The technique was eventually exposed through packet analysis. The FCC issued an order against the practice. Comcast sued, and in 2010 the DC Circuit ruled that the FCC had exceeded its authority under the existing regulatory framework. Round one to the telecoms.

The FCC tried again with the Open Internet Order in 2010. Verizon challenged it. In 2014, the DC Circuit struck down the most important provisions — again on authority grounds. The pattern was established: the FCC would set rules, the telecoms would sue, the courts would find the agency's authority insufficient under the framework it had chosen, and the cycle would reset. The opponents of the 2006 amendment had asked for evidence before the fact; the FCC spent eight years accumulating evidence after the fact; the courts kept ruling that the accumulated evidence wasn't sufficient basis for the regulatory structure being used.

The Obama-era FCC finally broke the cycle in 2015 by reclassifying broadband as a telecommunications service under Title II of the Communications Act — the legal foundation that gave the agency the authority it had been trying to exercise without possessing. The 2015 Open Internet Order established enforceable net neutrality protections backed by clear statutory authority. They held — for two years.

Then Ajit Pai became FCC chairman. In December 2017, the agency voted 3-2 to repeal the 2015 rules under the banner of "Restoring Internet Freedom." The repeal took effect in June 2018. Legal challenges from states, consumer groups, and tech companies were largely unsuccessful at the federal level. California passed its own net neutrality law and eventually won the right to enforce it. The Biden-era FCC tried to reinstate federal protections in 2024. Courts hit it again.

Eighteen years after the COPE Act vote, the basic federal question — should the pipes be subject to nondiscrimination requirements — was still unresolved. Still cycling through the same arguments that hadn't materially changed since Whitacre explained his intentions on camera in 2005. The concerns were speculative in 2006. They materialized in 2007. They were documented, litigated, established as policy, reversed, re-established, reversed again, and remain contested today.

The prediction was always there. The forecasters kept being called wrong until the storm made landfall.

iii · what twenty years built

Here's where the analysis gets uncomfortable, because the story of the internet we got isn't clean.

The defeat of net neutrality in 2006 didn't directly produce the internet of 2026. Consolidation, algorithmic attention silos, walled app ecosystems — these have multiple causes, and "ISPs were allowed to discriminate between traffic types" is one thread among several. The ad-supported model that made free services financially viable had its own logic. The smartphone shift that pulled users from the open web into app ecosystems was orthogonal to the pipe question. Platform consolidation had antitrust, economic, and regulatory dimensions that run deeper than broadband classification.

But the 2006 vote established something structural: the internet's infrastructure layer is a private toll system, not a commons. The companies that built the pipes get to decide how the water flows. And those companies had, and developed further, direct financial interests in preferencing certain flows over others.

Comcast acquired NBCUniversal in 2011. AT&T acquired Time Warner in 2018 — later spun off under pressure, but the strategic logic was plain throughout. The vertical integration that neutrality advocates warned about — the scenario where a telecom that also owns content has every incentive to preference its own properties — materialized. Not solely because of the 2006 vote, but in an environment shaped by that vote, where the infrastructure layer was already established as a commercial differentiator rather than a public utility.

The internet of 2026 is fast, convenient, and largely organized around the interests of a small number of very large companies — some of them the exact companies that opposed Markey-Eshoo in 2006, now substantially larger and more vertically integrated. The open web of 2006, where Wikipedia and YouTube and early Facebook coexisted on roughly equal footing with the same access to any user's connection, has given way to an internet where traffic patterns and discovery are shaped by algorithmic systems owned by the same players who argued against neutrality rules two decades ago.

The concerns weren't wrong. They were early.

---

There's a particular kind of frustration in watching a slow-motion prediction come true while the people who made it are told, repeatedly, that they're being speculative.

The Markey-Eshoo amendment failed in 2006. The throttling happened in 2007. The FCC spent eight years trying to establish workable rules on shaky legal ground. The 2015 rules were repealed two years after they were established. The 2024 attempt didn't survive the courts. Twenty years later, the original question is still cycling through arguments that haven't changed since Whitacre explained the toll-road model on camera.

The web we got is not the inevitable product of technological development. It's the product of choices made by 269 members of the House on a June afternoon, and by lobbyists who spent nine figures to shape that afternoon, and by regulatory agencies that spent years trying to establish authority they should have claimed earlier, and by courts applying narrow readings to broad statutory language. Contingent choices, all of them. It could have gone differently at several points.

It didn't. And here we are.