The Cage and the Key
The Cage and the Key
The sovereign internet is no longer a Russian bill. It is a live system.
The cage
Roskomnadzor’s whitelist filtering is now active in 71 of Russia’s regions, with outright mobile internet blackouts in 68. Moscow entered the shutdown zone last week. Residents in central districts found their connections either dead or restricted to government-approved services: banking apps, state media, telecom portals. The system is being tested on the capital’s metro network. Orders to throttle came, in the words of a telecom source speaking to Kommersant, directly “from above.”
The numbers are no longer theoretical:
- 37,166 hours of intentional internet shutdowns in 2025 alone — first globally, more than triple second-place Pakistan.
- 469 VPN services blocked by Roskomnadzor as of February 2026.
- Telegram banned nationwide; failure rates reaching 95% without VPN.
- $12.5 billion in estimated economic damage from throttling, blackouts, and social media bans in 2025.
- $29 million allocated for an AI-powered censorship system scheduled to go online later this year.
- $37–62 million lost in the Moscow region alone from just five days of disruption.
Russia’s Digital Development Ministry confirmed the whitelist will permit only essential services: banking, marketplaces, mobile operators, email, and digital cash registers. Everything else is blocked by default. Positive filtering. The assumption is no longer that the internet is open and specific sites are banned. The assumption is that the internet is closed and specific services are permitted.
That is the architectural difference between censorship and sovereign infrastructure. Russia has crossed from the first to the second.
VPN connections from inside Russia surged 800% above baseline after the Kremlin escalated its crackdown, according to the VPN Observatory. But the state’s capacity to block circumvention tools is growing faster than the tools can multiply. The Moscow Times reports that life inside Russia’s internet now resembles the circumvention habits of Iranian users: constant protocol-switching, cat-and-mouse with DPI, and a growing share of the population simply unable to reach the global internet at all.
The key
Three days after Russia’s whitelist system went live, the Tor Project opened something different.
On May 19, Tor launched its first Web3 crowdfunding campaign — a crypto-native round using quadratic funding to support ten nonprofit anti-censorship and privacy projects worldwide. The campaign runs through June 18. It accepts cryptocurrency donations and distributes a $115,000 matching pool using the quadratic funding formula: the more individual donors a project receives, the more matching funds it earns, regardless of donation size.
The matching pool sponsors read like a roll call of privacy infrastructure: Cake Wallet (Monero-focused mobile wallet), Zcash Community Grants (zero-knowledge proof ecosystem), Logos (trust-minimized governance), and Octant (Ethereum staking yield for public goods).
The projects on the receiving end include SecureDrop, the whistleblower submission system used by major newsrooms, and OnionShare, which enables anonymous file sharing over the Tor network. The full roster covers censorship circumvention, secure communications, and public-interest digital infrastructure.
The structural shift
This is not Tor accepting Bitcoin on a donation page, which it has done for years. This is a deliberate pivot away from the grant model that has sustained the organization since its founding.
The tension in that model has always been sharp. The Tor Project — a tool used by journalists, activists, and dissidents to escape government surveillance — has historically been funded in significant part by the same governments that sometimes surveil those very users. US government grants have comprised a meaningful share of Tor’s budget. The organization has been open about this tension. The pivot to crypto-native quadratic funding is, in part, an answer to it: if your users are global and your adversaries are sovereign, your funding should not come from sovereigns.
Whether quadratic funding can scale to replace grant revenue is an open question. The model rewards breadth of support over depth of pockets — 500 people giving 5 beats one whale giving 2,500 in matching power. But it is vulnerable to sybil attacks, and converting Tor’s 4.8 million daily browser launches into verified, unique contributors is a different challenge than serving anonymous web traffic.
The campaign is also accessible as a Tor Onion Service — meaning it can be reached from inside the countries that need the tools most. That is not incidental. That is the architecture.
What it means
Two infrastructures arrived in the same week.
One is a sovereign internet built on the assumption that all traffic is guilty until approved. It costs billions. It breaks economies. It works.
The other is a funding mechanism built on the assumption that the people who need privacy tools should be the ones who fund them — and that small, distributed contributions should carry more weight than large, centralized ones. It costs $115,000 in matching capital. It also works.
The cage is bigger. The key is smaller. But the key is being forged in a shape the cage cannot easily replicate, because the cage’s architecture depends on central approval and the key’s architecture depends on distributed participation.
That is the dispatch. The sovereign internet is real. So is the sovereign response.
Sources: United24 Media, The Moscow Times, Kommersant, HRW, CyberInsider, CryptoBriefing, VPN Observatory, Freedom House, Top10VPN. Campaign page: internetfreedom.torproject.org.