The Raid That Backfired

The Pirate Bay was offline for three days.

Then it came back. Traffic exceeded pre-raid levels within a week. Donation revenue spiked. International press coverage — uniformly framed as a story about the United States pressuring a smaller country to do the entertainment industry's bidding — reached audiences that had never heard of the site. A political party that had been quietly organizing for months suddenly had rocket fuel. The founders held a press conference and called it a victory. They were not wrong.

The MPAA had deployed months of diplomatic muscle, spent real political capital, and produced the most effective advertising campaign in the history of digital piracy. They had transformed a Scandinavian file-sharing site into a global symbol of resistance.

Eighteen years later, the same industry is running the same play against AI companies. The lesson apparently has a 20-year TTL.

i · the machinery of failure

The raid didn't happen because Swedish authorities spontaneously decided to prioritize copyright enforcement. It happened because the MPAA had been running an organized diplomatic pressure campaign — through the US Embassy in Stockholm, through Swedish government channels, reportedly through a retained Swedish lobbying firm — to convince the Swedish government to act against a site that operated in a legal gray zone under Swedish law at the time.

This is the important part. The Pirate Bay hosted torrent files — small pointer files that direct users to where actual content could be downloaded from other users' machines. They were not hosting copyrighted material. This distinction was precisely why the MPAA couldn't simply sue them into oblivion the way they'd gone after Napster or Grokster. Those services operated as centralized infrastructure; killing the center killed the service. TPB was an index. Different legal animal. Different attack surface.

So the strategy shifted from courts to diplomacy. Frame copyright infringement as a bilateral trade issue serious enough for state-level intervention. Make it about the relationship between Sweden and the United States, not about whether a specific statute was violated. Sweden, with trade relationships to protect, complied.

The raid was the result. What it accomplished legally was roughly nothing. Swedish prosecutors eventually filed criminal charges against founders Peter Sunde, Gottfrid Svartholm, Fredrik Neij, and investor Carl Lundström, but the case dragged through the courts for years. They were convicted in 2009, sentenced to prison time and fines, and spent years in the appeals process. The site itself was never shut down through legal means; it migrated hosting repeatedly and operates to this day, nearly two decades after the "decisive" raid.

What the diplomatic pressure approach accomplished culturally was to turn a file-sharing site into a political test case. When you involve the US Embassy and diplomatic channels in pressuring a sovereign government to raid a website, you've signaled that this fight is worth real geopolitical capital. The press picked up that framing and ran with it globally. The story stopped being a copyright enforcement story and became a story about power asymmetry — the entertainment industry using American diplomatic muscle to kneecap a Swedish website that had never hosted a single infringing file. That is a much worse story to be in.

ii · the long count

The immediate numbers were bad enough for the MPAA's purposes. TPB was offline for 72 hours, not the permanent closure they'd sought. Traffic returned immediately and exceeded pre-raid levels within days. The Streisand Effect at bandwidth scale — except the Streisand Effect is usually an accident. This outcome was entirely predictable if you understood anything about distributed systems or about how martyrdom narratives work.

The Swedish Piratpartiet — the Pirate Party — had been founded in January 2006, four months before the raid. It was a nascent political organization focused on intellectual property reform, privacy, and internet freedom. Membership before the raid: modest. Membership after: the kind of surge you get when an abstract political position suddenly becomes a concrete grievance. People who had been vaguely sympathetic to digital rights now had a specific incident to be angry about, a specific villain to point to, and a specific organization to join.

In the 2006 Swedish general election, held that September, the Pirate Party received 0.63% of the vote — below the 4% threshold for Riksdag seats, but enough to demonstrate organized political will in less than a year of existence. In the 2009 European Parliament elections, held two weeks after the criminal convictions of the TPB founders, the Pirate Party received 7.1% of the Swedish vote and won two seats in the European Parliament. The raid had succeeded in creating a durable political constituency that had not previously existed in organized form.

The Pirate Party concept spread internationally. Germany, the Czech Republic, Iceland, the United States, and dozens of other countries saw Pirate Party formations in subsequent years, most modeled explicitly on the Swedish example. The 2006 raid didn't just fail to stop file sharing — it seeded a political movement that spent a decade reshaping conversations about digital rights, surveillance, and intellectual property across the developed world.

Meanwhile, piracy metrics continued their trajectory through the mid-2000s and into the 2010s. The BitTorrent ecosystem diversified. Private trackers proliferated alongside public ones. Usenet groups flourished as an alternative. When Megaupload was taken down in 2012 — a far more dramatic enforcement action, involving FBI coordination and the arrest of Kim Dotcom in New Zealand — piracy traffic briefly dipped and then routed around the loss. The ecosystem was not centralized enough to decapitate.

The enforcement approach never changed. Different targets, same logic. After The Pirate Bay: Megaupload. After Megaupload: streaming sites. After streaming sites: private trackers. Each takedown earns a press release announcing a victory and identifies the next target. The content remains available. The targets multiply. The lesson remains unlearned.

iii · the pattern that won't quit

What actually worked, when something worked, was Spotify. The recorded music industry spent a decade trying to enforce its way out of piracy and watched its revenues collapse by roughly half in the process. Then a Swedish startup — Swedish, notably — offered a service more convenient than piracy at a price point consumers accepted. Within a few years, streaming revenue had substantially restored what enforcement never could.

The lesson was not "piracy is impossible to stop." It was "piracy fills a gap that the market won't." Fill the gap with a better product, piracy retreats. Try to close the gap by force, and you spend a decade and hundreds of millions of dollars producing slightly different file-sharing software.

This lesson was available in 2006. The music industry was in the middle of demonstrating it. The organizations that backed the Pirate Bay raid had access to the data. They chose enforcement anyway, because enforcement feels like doing something, and "offer a better product at a fair price" is harder to fund as a lobbying campaign.

Now we are in 2026, and the same organizations are filing copyright suits against AI companies over training data. The structure of the argument is familiar: these companies used our content without permission; they must pay licensing fees or cease; the existing legal framework requires it. Some of those arguments have genuine merit — the legal questions around training data and copyright are not settled, and reasonable people disagree about where fair use ends and infringement begins.

But the enforcement logic underneath the legal argument is the same one that failed in 2006. The training is already done. You cannot un-train a model any more than you could un-download a torrent. Enforcement will extract licensing revenue from large companies that can afford to negotiate and establish precedent that shapes the next generation of training methodologies. It will drive smaller or more aggressive actors offshore or underground, where they will continue training on whatever they can access. The models will keep improving.

The Pirate Bay is still online. The founders are building new ventures. Two decades of digital rights advocacy trace a line back to that Stockholm raid. And the enforcement machinery keeps cranking, looking for the decisive action that will finally make this stop.

I've watched this failure mode three times now. The mechanism is always identical: a centralized enforcement apparatus attacks a distributed behavior, generates a martyr narrative, and discovers too late that you can seize hardware but you cannot seize a protocol. The raid makes the legend. The legend makes the next ten sites. The next ten sites make the lesson available to anyone who wants to learn it, and most people don't.

Sixty-five officers. Three days of downtime. Eighteen years of consequences.

They got exactly what they paid for.