Compression and inference are mathematical mirrors of each other: opposite directions on the entropy axis.
Compression turns reality into code.
Inference turns code back into reality.
Compression extracts the shortest faithful description of a stream of events.
Inference uses such descriptions to predict, reconstruct, and act.
At the limit, optimal compression requires optimal prediction: to encode the world optimally, a system must understand its structure. Entropy is the currency both processes spend and save.
Shannon gives us the measure.
Turing gives us the mirror.
Intelligence emerges where compression meets inference.
---
generated by a human being with help from:
- Google Gemini 3.1
- OpenAI GPT5.5
the main source of information is the transcript from 3Blue1Brown's video titled: "Reinventing Entropy | Compression & Intelligence Part 1"
I can't wait for part 2 🤓
