Written on the 25th of October 2025.

We must decentralise the internet further

The internet has become as essential as electricity, and now we face a concerning phenomenon. Its profound fragility. On the 20th of October, many experienced this not as a theory, but as a reality. I noticed it again last week when colleagues of mine were unable to use Canva due to an outage. This outage at Amazon Web Services (AWS) wasn't a minor glitch, it was a global, cascading failure. Many systems froze. Communication apps went silent. Smart homes became dumb, turning coffee machines into unresponsive bricks. This event is a stark reminder of the invisible and brittle infrastructure most of us have come to lean upon.

This fragility doesn't stem from a single error, but from a systemic design choice we've all quietly accepted. We've traded the internet's original, resilient design for the irresistible economics of convenience. The backbone of the 21st-century global economy now rests almost entirely on a handful of server racks, owned by a tiny oligopoly of 'hyperscaler' companies like Amazon, Microsoft, and Google. I see this in my own work where we once carefully selected multiple hosting providers. I've since watched them get bought out, all falling under the same massive parent group. We've traded diversity for a fragile monoculture. Now that we are unhappy with some of the services they have provided that we've come to rely on we are left without alternatives.

The irony here is almost painful. The internet I was taught about, born from ideals, was the antithesis of this. It was a more resilient, more distributed web designed with less central control, fewer points of failure. And I'll be the first to admit, I'm part of this convenience trap. This very website you're reading is hosted on Cloudflare's infrastructure, which has had, and will undoubtedly continue to have, their own outages. I've become dependent, and you probably have too. What I'm not saying is that if I were to host it elsewhere there would be no downtime. Of course it will still go wrong elsewhere. However there it might not go down together with half the internet.

That recent AWS outage wasn't just a technical failure but it's a warning. The internet's evolution from a resilient, decentralised network into a fragile, centralisd oligopoly is a dangerous detour, not a final destination. This centralistion has not only created systemic technical risks but has also enabled unprecedented levels of censorship and actively stifled the 'permissionless innovation' that defined the early web. To reclaim the internet's promise of freedom, resilience, and opportunity, we must consciously and actively work to rebuild its decentralised foundations.

Link

The Internet that was

The AWS outage was a warning shot, but what exactly did we lose that made us so vulnerable? To answer that, we have to look back at the original architecture, which wasn't just a technical blueprint. It was more, a revolutionary political statement about how information should flow. The fundamental principles of freedom and resilience weren't abstract ideals tacked on later, they were embedded directly into the code of the network's foundational protocols.

The most crucial of these was the 'end-to-end principle.' In simple terms, the core of the internet was designed to be 'dumb.' Its only job was the humble but critical task of passing data packets from one point to another without inspecting them, censoring them, or optimizing them. All the 'intelligence', the applications, the data, the content, and the complex logic, lived out at the 'edges' of the network, on the computers of its users and service providers. This meant the power to create and communicate belonged to the people at the periphery, not the machines and companies in the centre. This 'dumb network' design naturally resulted in resilience.

Link

How we got here

This decentralised utopia was quietly paved over by the irresistible economics of convenience, creating the centralised model we are trapped in today. The shift was not forced by malice but was a slow, rational creep driven by consumer demand for frictionless experiences.

First came the rise of platforms with their 'walled gardens.' Around the dot-com boom and bust, a new generation of services emerged, such as search engines, social media, and e-commerce, that were infinitely easier for the non-technical user to access. Why manage your own web server or email client when Google offered a flawless, free service? Why learn HTML when Facebook offered a ready-made platform for your social life? These platforms (Google, Facebook, and Amazon) created a powerful gravitational pull by offering 'free' services in exchange for a simple currency: user data. They convinced us to surrender our digital autonomy by promising a frictionless experience, effectively creating vast, proprietary 'walled gardens' where they set the rules.

This set the stage for the tipping point, the cloud revolution. The true shift was the launch of Amazon Web Services in 2006. For companies, managing your own physical servers dealing with power, cooling, security, and maintenance was costly and inefficient. AWS changed the calculation. Over time it became exponentially cheaper, faster, and more reliable for any company to simply rent world-class infrastructure from Amazon than to build and maintain their own. This was a purely rational, logical business decision for millions of companies, but it simultaneously accelerated the concentration of the internet's physical infrastructure into the hands of a few others.

The final consequence of this trade-off is what can only be described as a new feudalism. When companies abandoned their own servers, they became digital 'tenants' on a 'landlord's' property. This means that if Amazon chooses to enforce a rule or raise its prices, the tenants have little recourse. Even more concerning is the fate of the user. Our data, our social connections, and our content are no longer things we own on our personal computers. They are locked inside these centralised platforms. We don't own our digital identity, we are merely given conditional access to it, subject to the changing whims and policies of the 'digital sovereign' that hosts our account.

Link

The dangers today

The AWS outage was merely the most recent visible symptom of the deep-seated dangers of our new centralised reality. The convenience we were promised, and which was eagerly accepted, has now delivered a hidden cost, turning our digital infrastructure into a liability rather than an asset.

The most immediate risk is technical fragility, or what is often termed cloud concentration risk. When such a massive portion of the global internet runs on the infrastructure of just a few companies, a single configuration error or bug in one data centre region can cascade across the globe, crippling businesses and services around the world. We have built a large portion of the 21st-century world upon a handful of single points of failure. The original internet was developed to route around destruction, the centralised cloud is designed for efficiency, but that efficiency is the enemy of resilience. When everything is interconnected in one monolithic system, everything can fail at once.

This concentration has created political and social chokepoints. Because the vast majority of human communication and commerce flows through a handful of corporate-owned platforms (Amazon, Apple, Google, Facebook), an organisation need only pressure a few to de-platform them, shut down applications, or restrict information flow. Corporate gatekeepers can now act as private ministries of culture and commerce. For instance, Apple's App Store has absolute power to decide who can and cannot publish software, effectively imposing its own moral and economic rules on millions of developers and billions of users worldwide, often without transparency or recourse. Now that legislators are trying to change this for the better, we are seeing Apple drag their feet to prevent this from happening.

Centralisation has severely stifled innovation. The 'permissionless frontier' is over. Today's innovators must build on platforms that are often their direct, existential competitors. How much friction does it create when a company like Netflix must run its primary streaming service on Amazon's AWS, enriching its greatest rival? Moreover, the financial tariff on innovation, such as the infamous 30% 'app store tax' charged by the gatekeepers on all digital transactions, acts as a massive deterrent, strangling small developers and startups and cementing the competitive advantage of the established giants. We are paying for convenience with our opportunity.

Link

A way forward

The challenge of centralisation might seem insurmountable, given the inertia and scale of the hyperscalers. However, a return to a decentralised internet is not a dream. It is a tangible goal being actively built today with old ideas and a new generation of technologies. This movement is focused on leveraging the original genius of the internet, placing power and intelligence back into the hands of the individual user and the periphery of the network.

The first step in this reclamation is re-decentralising application ownership. For too long, we have been forced into an all-or-nothing trade, handing over our most sensitive data to gain access to a service. The necessary counter-movement is built around concepts like Self-Sovereign Identity (SSI). The idea is simple: digital identity systems should allow users to own their digital identity and data outright. This means applications would run across a distributed network, not a central server, and the user would grant services permission to use their data for a specific purpose, rather than handing over permanent, unconditional ownership. This shifts the fundamental dynamic from feudal lord and tenant to two consenting parties.

Next, we must decentralise the online spaces we inhabit. The problem isn't the existence of applications like social networks, rather it's that a handful of companies own the entire network. Recent attempts to fix this, such as the AT Protocol, developed for and used by Bluesky, show promise, but they often fall short of true decentralisation. For instance, while the protocol is open, the core identity on Bluesky is still managed and ultimately owned by Bluesky itself. You can take this into your own hand, but it is too difficult for most users and will be left out of the network in some ways. This means that, despite the open protocol, the company retains the ultimate authority over your place in the public square. A far superior model is that of email. A personal email server, which you and I can host ourselves, can send a message to a Gmail user because both platforms adhere to a shared, open protocol: SMTP. Federated social media systems should operate the same way, allowing thousands of different servers to truly interconnect. No single company should own the town square, instead it should be a borderless space where different instances can communicate freely, much like the original internet was designed.

Link

The internet can be rebuild

The AWS outage was not a technical failure, it's a loud and philosophical warning. The internet's current continued trajectory toward centralisation is not some inevitable law of nature, but a correctable mistake, the consequence of billions of rational, short-term decisions made in favour of frictionless convenience. We are straying from the robust, resilient design of what the internet can be, and are continually lured by the 'free' and easy services offered by centralised giants. We are now paying the cost in systemic fragility, the loss of digital freedom, and the choking of permissionless innovation.

But the story does not end with the reign of the hyperscalers. The original ideals of the internet—openness, resilience, and true user control—are not dead. The tools to rebuild a more distributed web already exist, and more ideas and technologies are being developed every day. From identity systems that put the individual back in charge of their data to network protocols that reject central ownership of the public square, the architecture of a better web is emerging at the edges.

The future does not have to be a handful of corporate-run digital kingdoms. The choice we face is simple yet profound: continue as tenants in a new digital feudalism, or become active citizens in a rebuilt digital republic. By consciously choosing to build with, support, and use these emerging decentralised technologies, we can reclaim the internet's promise. We can build a global network that is once again a true network of networks. A permissionless, antifragile, and open frontier for human collaboration, one that is owned by all of its users, not just a few.

Link

Further reading