A few days ago I published an idea: What if we could fix the internet with a technology that's been around for 40 years? The response was encouraging – so today I want to go one step further and show you not just the why, but the how. Internet Fix 1.0 is no longer just a concept. There's a working implementation, a running server you can try right now, and an open invitation for you to get involved.

The Idea: A Whitelist Proxy for the Web

At its core Internet Fix 1.0 is just a simple proxy server that only allows access to domains that have been explicitly approved by a community. Think of it as a curated layer on top of the existing web – you still use your normal browser, you still access the same internet, but all the noise, spam, and AI-generated slop simply doesn't get through.

Every member of a proxy space can register new domains. If a domain turns out to be harmful or low-quality, the community can vote it out. It ends up on a blacklist – and stays blocked, even across connected proxy spaces.

Just a community, a whitelist, and a proxy.

Community & Federation – The Human Algorithm

No hidden algorithm decides what you see. No corporation controls the feed. Internet Fix 1.0 has no central authority. No single company or server decides what is good content and what is not. Instead, this decision is made by people – by the members of a proxy space.

But it goes further than that. Proxy spaces can connect with each other and form a trusted network. If your proxy space trusts another space, it automatically accepts their whitelisted domains as well. This is the same federation principle that powers Mastodon and the Fediverse – and it means Internet Fix 1.0 can grow organically, space by space, community by community, without ever needing a central authority.

The result is a decentralized web of trust – curated by humans, for humans.

Smart Search – Based on Large Language Models

Finding good content doesn't just mean blocking the bad stuff – it's also about discovering the good. Internet Fix 1.0 comes with an integrated semantic search based on Large Language Model technologies.

Whenever a new domain gets whitelisted, its content is automatically indexed and made searchable. By using LLM technology the search understands meaning, not just keywords – so you find what you're looking for, even if you don't know the exact words.

The entire search pipeline runs on a locally hosted LLM – no cloud API, no data sent to third parties. For example you can use a nomic-embed-text on a local llama.cpp server. Entirely under your control. Full stop.

A Working Implementation – Built on Quarkus & Cassandra

Internet Fix 1.0 is not just an idea anymore. There is a first working reference implementation available as Open Source. Built with Quarkus and Cassandra DB, it proves the concept works in practice.

You can try it right now at proxy.if1-0.org – register, whitelist a domain, and search. No installation needed.

The full source code is available on Codeberg. But here's the important part: this is a reference implementation, not the reference implementation. It's one possible way to build a proxy space – you can implement the same idea in a completely different way. Just support the federation principle.

This Needs You

I can build a proxy. I can write an API. But I can of course not fix the Internet alone.

Internet Fix 1.0 needs developers who believe the web is worth fixing. You don't have to use Quarkus or Cassandra – build your own proxy space in whatever language or stack you love. All that matters is that you support the open API and the federation principle.

If you want to dive in: the reference implementation on Codeberg is a good starting point. And if you want to run your own server, there's an installation guide to get started. Try the running instance at proxy.if1-0.org and let's improve it.

Most importantly: let's talk. Open an issue, write a comment, reach out. This idea only becomes real if people take it seriously.