Underlying theory
The Apollo Formula
C = M × (1 / D) × S
Shannon measured the box. I measured what's inside.
Compression isn't about the signal; it's about the meaning. As semantic distance D approaches zero - when a representation is its meaning - compression approaches infinity. LoreTokens are a constructive demonstration that D can be made small enough, on real corpora, to get compression ratios Shannon-coding can't touch.
- C compression of meaning (not bits; meaning-per-symbol)
- M mutual context (what sender and receiver already share)
- D semantic distance (how far the signal sits from the meaning it carries)
- S scope (how much of the knowledge domain the representation covers)