Resilience runs through people: read the full report

The future of internet resilience lies just as much human networks than it does in tools.

Resilience runs through people: read the full report
Adapted from a 16th-century folio of Zakariyyā al-Qazwīnī's Ajaib al-Makhlūqāt (The New York Public Library, Spencer Collection)

After many months of work, the full report is out today! You can download the full report here:

The short version

In June 2025, while Israel was striking targets across Iran, Iranian authorities ran a different kind of shutdown. They kept Iran looking online to the outside world by leaving BGP routes in place, while implementing centralized deep packet inspection, protocol whitelisting, and DNS manipulation at the border. International traffic fell by roughly 90% at peak. The National Information Network, hosting banking, government, and major Iranian platforms, kept working. For ordinary users, the global internet had effectively disappeared, and from outside the country, you could not see it happen.

We surveyed 900 people in Iran in the two to six weeks after the event, in a Persian-language Telegram bot survey co-developed with ASL19. We asked what they needed, what they tried, what worked, and who helped.

Five findings that matter

  1. Only about one in eight high-need users reached effective use of circumvention tools. The biggest losses cluster at three points: people who never realized tools existed, people who knew but never tried setup, and people who set up a tool successfully but found it didn't work for the things they actually needed to do.
  2. News is less the problem than contact or money: News fulfillment was around 57%. Contacting family scored 38%. Work and finance scored 39%. People's top priorities were the worst served.
  3. The infrastructure is largely human: 76% percent of people who knew about circumvention tools had helped someone else use them. Tens of millions of Iranians played some kind of helper role. Formal documentation played almost no part. The question "where did you learn about this?" was answered with "a friend" 40% of the time.
  4. Helping is not evenly distributed: As helper roles get more technical, helpers skew younger, wealthier, and more male. Women and lower-income users share information at high rates but rarely do hands-on setup work.
  5. Trust is fragile: Among the most successful users, more than half still reported significant privacy concerns. Many believed that free VPNs might be traps, that banking apps couldn't be trusted over VPN, and that some tools might be controlled by security agencies.

What we recommend

For tool builders: design for emergencies, not for calm configuration. Replace user-side trial and error with tool-side automatic failover. Optimize for contact and work, not just for content. Be radically clear about what is logged and what is not. Build features for helpers, not only for end users.

For funders: separate budget lines for helper support from generic user outreach. Don't take user acquisition for granted. Fund usability testing in realistic stress conditions, not in labs. Prioritize task-level outcomes (can someone place a call, complete a transaction, stay logged in) over installs.

For support organizations: work through the helpers who already exist. Provide bundles they can forward. To address helper inequity, support lower-level helps to advance in their journey.

For media and information work: read this report not only as a study of Iran, but as a study of what happens under the surface of algorithmic distribution anywhere. The helpers we identify for tools are doing the same work for news, for safety information, for fact verification. They need products and materials built for them, not for the reader they are passing things to.

This report is a joint output of Gazzetta and ASL19. It is part of a longer line of work on what we are calling distributed information: the shift from a delivery model of media and tooling to a network model, where the work of reaching people is done by the people themselves.

If you build tools, fund this work, run editorial operations in censored or constrained environments, or do service design for hard-to-reach populations, we would like to hear from you: hello@gazzetta.xyz.

Huge thanks to the Open Technology Fund for supporting the design of the final report released today in Brussels.

Subscribe to Gazzetta

Don’t miss out on the latest issues. Sign up now to get access to the library of members-only issues.
jamie@example.com
Subscribe