# Algorithms broke the internet. We are rebuilding it from the ruins of news feeds.

#security #opensource #news #node

We spend hours in feeds that know everything about us, yet leave us in absolute emptiness. Modern social networks have turned into dopamine factories: algorithms push us into comfortable information bubbles, feed us rage-bait or addictive content, and breed isolation.

Labels dictate what we think. A person's name, a brand's status, or a loud headline replace real context.

But what if a social network could work differently? What if it wasn't a static grid of posts, but a living, evolving mind woven from human associations, knowledge, and connections?

That is why I built ABNoise.

What is ABNoise?

It’s a node-based, decentralized platform where the name of a node is secondary. Instead, the context is dynamically shaped by the community through directed graph connections, and every node is created by people.

Technical stack & architecture:

Node-based Platform & Social/Knowledge Network: Built for meaning-driven exploration rather than endless scrolling.
Crowdsourced & Dynamic Metadata: Every member can reshape, contribute, and expand nodes.
Lightweight Core: No heavy search engines or complex infrastructure (Elasticsearch, etc.). It runs on a clean relational stack with SQLite and custom graph relations.

Quick Guide & Tips

I tried to keep it simple: there’s a main page where you can search for nodes by their connections, and you’ll get it fast. Users can freely edit any node, add tags, and comments. There are also built-in sections like Favorites, Recent, and Settings.

I recommend checking out a couple of specific pages because they make the concept super clear. For instance, I suggest searching for "Donald Trump" or "OpenAI".

But wait—how are you going to reach them? Just type "Donald Trump" in the search box, right? Wrong! Trump's node could be named literally anything: D. J. Trump, just "Trump", or "Trump Donald".
Instead, try a different approach: think of what naturally pops into your head when you think of Donald Trump. A burger? Could be a good example, but no. The United States, right? Add it to your search filter. What else? Maybe "President"?
Once you search that: look at the referencing nodes list—you'll probably find him right there as one of the first ones in the list. You can use the exact same logic to figure out how to find OpenAI.