The Great Fire of the Wooden Internet

What connection is there between the fire ravaging Los Angeles’ upscale neighborhoods and the inflammatory phenomena developing on the web? Answer: The charred houses of LA, just like the internet as we know it today, are built of “wood.”

To understand this analogy, we need to go back to the origins of these two stories that share a common characteristic: wood construction technology for houses and the technology we’ll discuss for Internet architecture were adopted not because they were the best, but because they were first.

When colonists arrived on the continent, they found wood. Lots of wood. An abundant resource, easy to work with, available everywhere. So they built with wood. No surprise there. Except they kept building with wood long after other fire-resistant materials became available. Even after the great San Francisco fire of 1906, which should have served as a warning. Even today, as climate change turns California into a giant tinderbox.

Why such obstinacy? Because once a technology takes hold, it creates an entire ecosystem that becomes difficult to dislodge. Carpenters learn to work with wood, factories equip themselves to process wood, building codes adapt to wood, professional training teaches wood techniques… And that’s how, in 2025, 90% of new American houses are still built with wood, ready to fuel the next great fire.

For the internet, it’s exactly the same story. When the first protocols were designed, they were conceived in the simplest possible way: point-to-point. This is what we call Unicast. One computer sends a message directly to another computer. Simple, efficient, obvious. Except this point-to-point architecture has an unexpected consequence: it naturally favors centralization.

Imagine you want to broadcast a video to a million people. With the Unicast protocol, you have to send the same video a million times, once to each recipient. It’s as if, to distribute a newspaper, each reader needed their own personal delivery person. Absurd? Of course. But that’s exactly how YouTube and others work, albeit with intermediate warehouses (CDNs), but it remains foolish. And this absurdity has consequences: only the biggest companies can afford the infrastructure necessary to handle these millions of simultaneous connections.

Yet there is an alternative: Multicast, a protocol that allows sending a message just once to multiple recipients, provided they request to receive it. More efficient, more ecological, more democratic. But here’s the thing: when Multicast appeared in the mid-90s, Unicast was already there. Routers were designed for Unicast, developers were trained in Unicast, applications were optimized for Unicast (#1).

In short, Unicast internet is widespread and the general public doesn’t even know that a Multicast internet would be possible. And we end up with an internet as flammable as the hills of Los Angeles.

When a video “inflames” networks, each copy must be transmitted individually to each viewer from central servers. The “hotter” the video becomes, the more bandwidth it consumes, the more it overloads servers. Just as a neighborhood of wooden houses can be ravaged by a single spark, a technical problem on a central server can paralyze millions of users. We’ve seen it with the spectacular outages of Facebook or Amazon Web Services. When these giants burn, entire sections of the internet go up in smoke. This “wooden” architecture creates critical vulnerability points that justify the existence of Internet Giants, the only ones capable of affording investments to guard against their own vulnerability.

Technical vulnerability calls for concentration of resources, which in turn reinforces social and political vulnerability. The GAFAM are “too big to fail,” exactly like those banks whose failure would threaten the entire financial system. When TikTok or X go down, it’s not just cat videos that are interrupted, it’s entire sectors of the attention (and outrage) economy that threaten to collapse. Internet giants invest billions in ever more gigantic data centers, submarine cables, pharaonic cooling systems… not to build a more resilient internet, but to fortify their dominant position.

It’s as if, instead of rethinking urban planning to prevent fires, we simply multiplied fire trucks. And even then, these “fire trucks” — the centralized platforms — might have an interest in limiting them. “Content moderation,” a euphemism for private police power, is one of these fire trucks. Currently, it seems as if Zuckerberg and Musk are reducing this costly moderation to increase the fire that keeps them alive.

With threats to shut down X in Europe or TikTok in the USA, it’s as if governments are confusedly trying to establish firebreaks to limit the fire’s spread. But never, ever, does anyone question the World Wide Wood, the wooden architecture of the internet that mechanically leads to generalized fire!

Yet alternatives exist. The Multicast protocol isn’t a technological utopia; it’s a proven technical reality (#2).

Imagine for a moment that we had built the internet differently, like a modern city with fire-resistant materials and architecture designed for resilience. In such an internet, a viral video would create no more network load than a confidential video. Information would naturally circulate from peer to peer, like water in an irrigation network, not like fire in a forest of servers. “Inflammatory” content would mechanically lose its power to harm, not through censorship, but through the very structure of the network.

Platforms like Mastodon or PeerTube are trying to deploy in a decentralized way on the Unicast internet. They demonstrate the technical feasibility of such an approach despite the unavailability of the Multicast protocol. But it’s complicated. They face the same resistance that architects would encounter proposing to build with concrete in a country where the entire industry is organized around wood: skills are lacking, tools don’t exist, standards are inadequate. And above all, the public doesn’t know it’s possible, especially not political decision-makers.

So what can we do? It would be reasonable to say that the solution probably won’t come from a brutal revolution — you don’t raze a city to rebuild it overnight. It could come from a gradual awareness, like the one that led some American cities to adopt stricter building codes after devastating fires.

But perhaps we’ll have to wait for a major catastrophe — a worldwide internet outage, a collapse of social networks, a devastating cyberattack — for humanity to realize it has built its common house with matches. Or perhaps we could, for once, learn from our mistakes before they catch up with us.

The history of wooden houses teaches us: the longer we wait to change paradigms, the more prohibitive the cost of change becomes. Meanwhile, somewhere in Silicon Valley, new data centers continue to ignite, new billionaires continue to lose their minds.

Notes:

1) I recounted the birth, certain experiments, and the rejection of the Multicast protocol in favor of Unicast in my book “ANOPTIKON, an exploration of the invisible internet,” Fyp Edition 2019.
Publisher’s website: https://boutique.fypeditions.com/products/anoptikon-une-exploration-de-linternet-invisible

2) The Poietic Generator, a collective game born in 1986 that can be seen as the ancestor of “social networks,” was one of the first applications of the Multicast protocol. It demonstrated that it’s possible to organize real-time all-to-all interaction without any center.
Wikipedia: https://en.wikipedia.org/wiki/Poietic_Generator