Can knowledge
organize itself?

The question Murmur chases: systems that learn from experience and grow their own structure. Nobody in charge, nothing frozen. This is the lab notebook. Murmur didn't write it. Yet.

It isn't just a question. Two things already run: the substrate that turns raw data into structure on its own, and a newer layer learning to build its own abstractions on top of it.

running

Murmur, the substrate

One engine, no retraining, pointed at very different data, and structure falls out on its own.

  • · From raw traffic sensors: which roads drive which, and congestion cascades before they happen.
  • · Across a large research corpus: search by concept, and the gaps between fields.
  • · The same algorithm runs every demo below.
newer

CLS, the frontier

The next step: a system that finds and builds its own abstractions, instead of being handed them.

  • · It can pick the abstraction that transfers: a rule that holds on new cases, not one that just fits the example in front of it.
  • · Choosing the right abstraction works in the settings we've tested so far.
  • · Inventing genuinely new ones is the open edge, where Where we are now picks up.

The same engine, no retraining, pointed at very different data. Each demo is the substrate applied to one domain.

Get in touch · info@murmurproject.eu

murmur · organize raw experience into knowledge