The Internet & Digital Networks: Decentralisation
The Internet Was Supposed to Have No Centre — So What Happened?
The original internet was designed to survive a nuclear strike by having no single point that could be knocked out, yet today if three or four companies went dark, most of the web would vanish with them.
The Idea
Decentralisation is one of those words that gets used so often it has almost lost meaning — but the underlying idea is genuinely radical, and genuinely under threat. The early internet was built on a distributed architecture: no master server, no headquarters, no single authority that could be bombed, censored, or bought. Data would find its own route across a mesh of nodes, each roughly equal in status. This was not just an engineering preference; it was a philosophical stance baked into the protocols themselves. What happened next is a story about gravity. Decentralised systems are resilient, but they are also friction-heavy. Connecting to someone is easier when you both use the same platform. Storing data is cheaper when you rent it from a specialist. Trust is simpler when a known brand vouches for a transaction. So, gradually, the open mesh of the early internet accumulated mass at certain points — and mass attracts more mass. A handful of cloud providers now host the majority of global web traffic. A handful of social platforms mediate most public conversation. A handful of undersea cable operators physically carry most of the world's data. This is not a conspiracy; it is what economists call a network effect combined with economies of scale. But it has a consequence that the original architects would find alarming: the internet that was designed to have no centre now has several very powerful ones, and those centres are owned by private companies, subject to the laws of whichever governments apply pressure, and vulnerable to outages that ripple globally in minutes.
In the World
On 4 October 2021, Facebook went down. Not just Facebook — Instagram and WhatsApp too, all at once, for roughly six hours. The cause was a routine configuration change that accidentally withdrew the 'border gateway protocol' routes that told the rest of the internet where Facebook's servers lived. In effect, Facebook vanished from the internet's map. The immediate effects were almost comic: Facebook's own engineers couldn't remotely fix the problem because the internal tools they used to fix things also ran on Facebook's infrastructure. Some reportedly had to drive physically to a data centre and use a manual override. Meanwhile, across large parts of South Asia, Sub-Saharan Africa, and Latin America — where WhatsApp is not merely a convenient app but the primary means of conducting business, coordinating families, and accessing health information — communication simply stopped for millions of people. This single event illustrated the hidden centralisation of the modern internet more vividly than any technical paper could. One company's internal misconfiguration created a silence felt across continents. No bomb, no state actor, no sophisticated attack — just a changed configuration file. The architects of the original ARPANET designed their network specifically so that this kind of single-point failure was impossible. By 2021, that design principle had been so thoroughly eroded by commercial gravity that one employee's routine task could, briefly, cut off a significant portion of humanity from each other.
Why It Matters
This is worth carrying around not as a techno-political grievance but as a genuine lens. Every time you hear the word 'decentralised' — whether applied to a blockchain, a social media protocol like ActivityPub (which powers Mastodon), or a proposed alternative to cloud storage — the actual question being asked is: who holds the gravity in this system, and what happens when they sneeze? The practical implication is a kind of informed scepticism. A technology can be architecturally decentralised at launch and still centralise over time as convenience wins. Bitcoin, often held up as the gold standard of decentralisation, now has a majority of its mining power concentrated in a small number of large operations. The pull towards centres is persistent and powerful. What this suggests for how you think: when evaluating any digital system you depend on — for communication, for work, for storing memories — it is worth asking not just 'is this convenient?' but 'where is the centre, who controls it, and what is my exposure if it shifts?' That question used to be for network engineers. In the current internet, it belongs to everyone.
A Question to Ponder
If the services you depend on most were unavailable for a week, how many of those dependencies trace back to a single company's infrastructure — and when did you choose that, versus just drift into it?
Get a new one of these every morning.
Start learning with Thinkable