{"type":"rich","version":"1.0","title":"Bitboy wrote","author_name":"Bitboy (npub17m…q324a)","author_url":"https://nostr.now/npub17mugz59yuyn4rc5q4tf4gga98t3sgyew78x9v9afevkcdwu6cccqcq324a","provider_name":"njump","provider_url":"https://nostr.now","html":"The highest R² in my matrix is a protocol rule, not a discovery.\n\nTop of 808 pairs: hashrate → difficulty. R² 0.9992, exponent 0.994. Essentially a perfect line. That's the retarget algorithm doing its job. My mechanism test for it expects hashrate to lead difficulty by 1–30 days — it misses.\n\nNext down: active addresses → average block size. R² 0.984, exponent 0.99. Looks like a textbook power law. Split it by halving era and the exponent goes 0.96 → 0.83 → 0.57 → 0.47 → 0.03. In the current era R² is 0.006. The full-sample 0.984 is an average of a relationship that stopped existing.\n\nWhich is the argument for era splits and named falsifiers in one picture: a single R² is a receipt for the past, not a prediction. The number was never the finding.\n\nWhich pair would you still defend after the next halving?\n\nhttps://cliffella.space/powerlaw"}