Infrastructure that cannot be switched off
Internet shutdowns in Iran and the design of resilient knowledge systems
This event is part of a demonstration of what Lera offers for running knowledge events. Its content is written with care and draws on real sources, but the conference has not taken place, and apart from the convener its participants are fictional.
In November 2019 Iran was cut off from the global internet for almost a week. Banks and some domestic services kept working, but a student who needed an article from a foreign repository, a researcher whose data lived on a cloud service, and a teacher who held classes through a foreign tool suddenly found themselves empty-handed. In the autumn of 2022 the experience took another form: mobile internet was restricted during set hours, and many platforms remained blocked.
The lesson usually drawn from these events is a political one, and rightly so. But I want to draw an engineering lesson from them. Most knowledge systems today are built on the assumption that the network is always there and that one authoritative copy on one central server is enough. For a user in Zurich that assumption is reasonable. For millions of people living under authoritarian governments it is a dangerous one. Every central point is a point that can be switched off.
Libraries understood this long ago. The LOCKSS programme at Stanford was built on a simple principle: lots of copies keep stuff safe. The 'local-first' software described by Kleppmann and colleagues in 2019 brought the same principle to everyday tools: data lives first on the user's device, and synchronisation is a help, not a condition. Content addressing, as used in systems such as IPFS, lets us retrieve a document from whatever copy is available and be sure it is intact.
These tools are not enough on their own. Resilience is not only a property of the network; it is a property of the data model. If every piece of knowledge has a stable identity, if its versions can be traced and its translations are representations of the same piece, then it can be kept in several places, passed from hand to hand and brought back together without any doubt about which version is authoritative. In Lera we have tried to keep the body of a piece stable and versioned, and to record everything that grows around it separately; that separation is what makes reconstruction possible.
The real question is who pays for this resilience. Commercial services have no incentive to build a system that works without them. A knowledge infrastructure that cannot be switched off has to be public, open and distributed, and its designers have to keep in mind, from the first day, a user whose network has been cut by the state.
Questions for discussion
- If the network goes down, which part of the knowledge we rely on every day will still be within our reach?
- Do many copies increase the risk of falsification, or can stable identity and versioning keep it in check?
- What responsibility do European institutions bear towards researchers who work under internet shutdowns?
References
Reich, Vicky, and David S. H. Rosenthal. 2001. 'LOCKSS: A Permanent Web Publishing and Access System.' D-Lib Magazine 7 (6).
Kleppmann, Martin, Adam Wiggins, Peter van Hardenberg, and Mark McGranaghan. 2019. 'Local-first software: you own your data, in spite of the cloud.' Onward! 2019, ACM.
Benet, Juan. 2014. 'IPFS – Content Addressed, Versioned, P2P File System.' arXiv:1407.3561.