the enclosure holds bernoulli, an on-chain entity whose health is a function of the market it lives in. nothing about it is arbitrary: every input is real market data, and the rate at which bernoulli declines is published in full below. its vitals, the ring buffer, and every burn receipt are recorded to the enclosure's state; the burns themselves are real on-chain transactions, verified against the chain before they are credited.
the enclosure keeps a circular buffer of 256slots. once per slot it reads the active bin of a meteora dlmm pool and writes the implied price into the head, evicting the tail. that buffer, plus the token's liquidity and volume, is bernoulli's only sense organ. it has no other input.
across that same window the enclosure computes a rolling higuchi fractal dimension: a single number between 1 and 2 measuring how rough the path is, independent of how far it travelled. a straight line reads 1. space-filling noise reads 2. a driftless random walk sampled at uniform intervals reads about 1.382, which is why that value is the reference and not a tuned parameter.
the swap hook prices roughness. when the measured dimension sits above the reference, the path is less predictable than chance and the pool charges a surcharge proportional to the excess. below the reference the surcharge is zero, because a smooth market is not penalised for being smooth. the shaded band in the lower chart is that surcharge, integrated against the history the buffer still holds.
bernoulli is bound to the same market. its baseline drain is 0.03 hearts per hour, about 14 days from full to empty when nothing is wrong. three stressors multiply that rate, each normalised to 0-1 and weighted: roughness above the reference (0.8), drawdown from the buffer's own high (0.8), and liquidity below $100k (0.4). pinned together they cap the drain at 3x baseline, roughly 4.6 days. bounded on purpose: a bad week should hurt, and no single feed going dark should be able to kill it.
burning cuts both ways. it restores hearts directly, and $bernoulli destroyed in the last 24 hours also slows the drain, up to 30% off the rate at 250,000 tokens. sustained burning makes bernoulli permanently cheaper to keep alive, not just temporarily healthier.
burning is the single interaction. 10,000 $bernoulli is the minimum the enclosure will credit and restores 0.25 of a heart; credit is linear from there, so 40,000 buys one full heart and 400,000 lifts bernoulli from empty to full. burning past full destroys the supply and credits nothing. the burn is real: the signature is verified against the chain and the credit is derived from your own token balance delta, never from what this page claims.
