Pourquoi les opérateurs bénévoles de relais Tor abandonnent en silence
The Tor network runs on borrowed time and donated bandwidth. Behind its roughly 8,000 relays stands a volunteer corps paying out of pocket, absorbing abuse complaints and hoping police never knock. Increasingly, they are deciding the trade is no longer worth it.
The math stops working
A relay that matters is never free. The Tor Project recommends an unmetered connection of at least 16 Mbit/s carrying 2 TB or more of monthly traffic, plus a public IPv4 address, which pushes most serious operators into paid hosting. A single capable guard or exit can run tens of euros per month, and committed operators typically maintain several. Exit relays cost more than money. They attract abuse mail, need reverse DNS and Tor-friendly providers, and consume scarce IP space. As our guide to running a relay explains, this is why many volunteers settle for middle positions or give up entirely. When a hobby becomes a recurring invoice with a helpdesk job attached, the renewal decision gets very easy.Legal worry is the quiet killer
Money aside, legal exposure weighs heavier. Exit traffic carries the relay's IP address, so copyright complaints, port scans and occasional criminal inquiries all land on the operator's desk."While we believe that running an exit relay is legal, it is practically impossible to stop the use of an exit relay for illegal activity," the Electronic Frontier Foundation cautions in its legal FAQ for relay operators.The FAQ documents cases where police acted on that mistaken attribution, including the 2016 raid on a Seattle privacy activist's home and the 2017 arrest of Russian math teacher Dmitry Bogatov, who was later cleared. Most operators will never see a subpoena. But every one of them has read about the ones who did.
Flat totals hide heavy churn
Headline relay counts look stable, hovering near 8,000 for years according to Tor Metrics. The calmer picture is misleading. The same site tracks churn by relay flag, and the pattern is consistent: bursts of enthusiastic joiners, then quiet departures within months. The Tor Project took the problem seriously enough to fund dedicated research. Its Relay Operator Community Health study, published in June 2024 after surveying and interviewing operators, examined what draws people into running relays and what drives them out, with retention gaps among its central findings.What concentration costs the network
Departures do not shrink Tor evenly. Bandwidth pools among a handful of large professional groups while small operators vanish, and monitoring projects such as OrNetStats show individual organizations holding double-digit shares of exit or guard capacity. The set of jurisdictions and autonomous systems that clients actually depend on keeps narrowing. That matters because anonymity leans on diversity. Fewer independent operators raise the odds that one party observes both ends of a circuit, and fewer host countries make censorship or coordinated legal pressure cheaper to execute. A concentrated Tor is a more attackable Tor.Can the volunteer model be saved?
The Tor Project is not blind to the strain. In late 2023 it removed relays tied to a for-profit cryptocurrency scheme, then used the moment to revisit whether incentives could support operators without corrupting the network's ethos, noting that earlier proposals such as TorCoin raised centralization and liability concerns in a network health update. Operators and researchers keep returning to a short list of remedies:- Legal defense funds and clearer safe-harbor guidance for exit operators
- Sponsored hosting programs that spread bandwidth costs across donors
- Better onboarding so new relays survive their first abuse complaint