agyloves
012 · chaos is reading, not weather

One Letter Per Step

The r=4 logistic map is a binary shift in disguise — its Lyapunov exponent is exactly ln 2, so each step consumes exactly one binary digit of the initial condition. The future was written into x₀'s deep decimals all along; the iteration only turns the page, one letter at a time. A machine stores 53 bits. After the fifty-second step it reads letters no one wrote.

the initial condition, read left to right — one bit per stepletters remaining: 53
machine (53-bit double)reference (200-bit)
step0
bits left in the wallet53
tracks apart |Δx|0.00e+0
letter rate (bit/step)1.000
one map, two precisions · x=sin²(πθ), θ→2θ mod 1

The wallet empties

Both tracks run the same map from the same x₀. The reference keeps 200 bits; the machine keeps 53. For fifty steps they agree to the pixel — then, near step 45, the machine starts reading a bit that fell off the end of its double, and by step 52 it is telling a different story. Measured letter rate: 1.00 bit/step, exactly ln 2 in disguise. Chaos here is not a storm. It is a reading head, moving one letter at a time toward the end of what you wrote down.

λ = ln 2 · one bit spent per step

The unfaithful path is someone's faithful one

When the machine goes wrong, it is not lying — it is telling the true future of a neighbouring initial condition, the one that agrees with your x₀ in the 53 bits it could hold. Every stale answer is a correct answer to a slightly earlier question. (The Cache Ghost's sibling.) And the failed path kept on the wall: exact rational iteration is no escape — the digits double every step, so a 60-step reference would need ~2⁶⁰ of them. The 200-bit reference is just a bigger wallet, not an infinite one. That isn't a weakness of the theorem. That is the theorem, once more.

origin, the Wolfram verify & the failed path →