DMCA vs NTD: how takedown law decides what completes
Retention gets all the attention, but the legal regime a backbone runs under quietly decides which articles vanish. The same NZB loses different parts on a DMCA feed than on an NTD one — and that gap is the whole case for diversity.
Ask why a download failed and most people reach for retention. It’s the wrong first suspect. An article inside retention can still be gone — not because the spool aged out, but because somebody sent a notice and the backbone removed it. Which notices land, what they target, and how fast they’re processed is decided by the legal regime the backbone operates under. There are two that matter, and they punch different holes in the same NZB.
Two regimes, two different holes
Every major feed sits under one of two takedown systems. Both are “notice and takedown” at heart; the difference is who sends the notices, what they go after, and how the provider has to respond.
DMCA — the US Digital Millennium Copyright Act (17 U.S.C. §512). A backbone with US infrastructure keeps its safe-harbour protection only by removing access to specific infringing articles once it receives a valid notice. In practice that means a steady, largely automated stream of notices from rightsholders and their anti-piracy agents, each naming message-IDs to pull. US backbones process a lot of it.
NTD — the European notice-and-takedown model, most visible in the Netherlands. Dutch providers remove on notice too, but the notices come from a different ecosystem: the anti-piracy foundation BREIN is the prominent sender, and since the EU’s Digital Services Act became applicable in 2024, the notice-and-action obligations of EU hosting providers are shaped by that framework rather than by US copyright statute. Different senders, different priorities, different cadence.
Neither regime is uniformly “stricter.” They remove different articles, identified by different parties, on different timelines. That is the point.
Why the same NZB completes differently
A takedown targets a message-ID — one specific article. A 50 GB release is tens of thousands of articles, and a notice campaign rarely removes all of them cleanly: it removes the ones it was told about, on the backbone it was sent to.
So when your NZB hits a DMCA backbone, the missing articles are the ones US rightsholders reported there. When the same NZB hits an NTD backbone, the missing articles are the ones BREIN and EU notice-senders reported there. The two gap sets overlap, but they are not identical — and where they differ, one backbone holds exactly the articles the other lost. That non-overlap is what a second server on the other regime fills. It’s also why “just buy more retention” doesn’t fix a completion problem caused by takedowns: the article was deleted, not aged out.
It maps straight onto the backbones
This isn’t abstract. The regime is a property of where the servers physically sit, and the community’s backbone map labels it:
- DMCA-side feeds: Newshosting’s US servers, Giganews, the US side of UsenetExpress, and the Abavia/XS News group (which leans on US backfill).
- NTD-side feeds: the Dutch independents — Eweka, Tweaknews, Usenet.Farm, and ViperNews — all carry NTD notices.
Two details make this sharper than “US vs EU”:
- Split-regime backbones. UsenetExpress is DMCA on its US servers and NTD on its NL backfill — a single backbone straddling both. Usenet.Farm and ViperNews run NTD up front with DMCA on the deeper backfill. The regime can even depend on how old the article is.
- One owner, two regimes. Inside Highwinds/Omicron, Eweka operates under NTD while Newshosting’s US servers operate under DMCA — same corporate group, two different takedown streams. It’s a reminder that the legal regime, not the brand or even the owner, is the variable that moves completion.
What this means for your setup
The block-account and “who owns whom” arguments both end here. The reason a second server lifts completion is not merely that it’s a different backbone — it’s that the most useful second server is under a different takedown regime, so it was scrubbed by different notices.
The practical rule: pair across regimes. A DMCA primary (say, a Newshosting-family unlimited) plus an NTD fill (a Dutch backbone like Usenet.Farm, or a Blocknews-style block) covers far more than two DMCA brands ever could. If your primary is already an NTD Dutch provider like Eweka, flip it: add a DMCA-side fill. The gain shows up exactly where you’d expect — on older releases and on heavily-reported content, the two categories takedown campaigns hit hardest.
This also reframes the block-account guide: when it says “buy a fill on a different backbone,” the sharper version is “buy a fill under a different takedown regime.” And it’s the mechanism beneath who owns whom — two brands on the same regime were scrubbed by the same notices and fail together.
Caveats, because law is messy
- Takedowns hit articles, not releases. A “removed” release is usually one with enough missing articles that PAR2 can’t repair it — partial loss, not a clean delete. A different regime often supplies just enough of the missing pieces to push repair back over the line.
- Reposting muddies it. Popular content gets reposted, often obfuscated or password-protected, which resets the clock until the new posts are reported too. Completion on a given release drifts over time on both regimes.
- Regimes evolve. The DSA is still bedding in across the EU, US notice volume keeps climbing, and a backbone can change how aggressively it processes either. Treat the DMCA/NTD split as a durable structural fact, but not a fixed scoreboard.
- It’s inference, not a published spec. Providers don’t advertise their takedown internals; the regime labels come from legal notices, server location, and the community map. Verify before you build a setup around one assumption.
The headline number on a provider’s page is retention. The number that quietly decides your success rate is which legal regime has been editing that spool. Buy your second server on the other one.