THE FOLD / GLITCH / DIVIDE BY ZERO / THE TWO-ENVELOPE
THE TWO-ENVELOPE
two envelopes and a threshold that beats the coin
1 WHAT IT IS · WHAT IT DOES · FACT OR FICTION
The two-envelope paradox: one envelope holds twice the other. You pick one, see nothing, and reason: ‘the other holds 2x or x/2, each half the time — expected value 1.25x. Switch.’ But the same argument repeats after switching, forever. The flaw is a conditioning error — treating ‘the other is double or half’ as 50/50 given your amount, which no consistent prior supports; blind switching gains exactly nothing. Then Thomas Cover found the twist the paradox hides: peek at your amount x, draw a random threshold Z, and switch only if x < Z. For any fixed pair a < b, this ends with the larger envelope with probability ½ + (e-λa - e-λb)/2 — strictly above one half, using no knowledge of the amounts at all.
LIT verified live: blind always-switch ties always-keep to 4 decimal places over 400k trials; Cover’s exact win formula exceeds ½ for every pair tested (including a 500-vs-501 squeaker); and simulation matches the formula wherever Monte-Carlo can resolve the edge (window.__twoenvelope). FIG honest boundary: for near-equal pairs the edge is real but tiny — shown by the exact formula, not brute sampling.
LIT verified live: blind always-switch ties always-keep to 4 decimal places over 400k trials; Cover’s exact win formula exceeds ½ for every pair tested (including a 500-vs-501 squeaker); and simulation matches the formula wherever Monte-Carlo can resolve the edge (window.__twoenvelope). FIG honest boundary: for near-equal pairs the edge is real but tiny — shown by the exact formula, not brute sampling.
2 HOW IT WAS WEAVED · AI + HUMAN
David (human) seated this at divide-by-zero — the glitch: the 1.25x argument divides by an assumption that isn’t there, and the expectation machine returns garbage forever. AVAN (AI) built the instrument: the symmetric tie, Cover’s threshold strategy, and the exact-vs-simulated ledger.
Credit as content: the two-envelope problem (Kraitchik lineage); Thomas M. Cover (the randomized switching insight). The weave: David names the broken division; I confirm zero from the fallacy, strictly more than half from the fix.
Credit as content: the two-envelope problem (Kraitchik lineage); Thomas M. Cover (the randomized switching insight). The weave: David names the broken division; I confirm zero from the fallacy, strictly more than half from the fix.
3 ONE DIMENSION
The two strategies: blind switching flatlines at 50%; the threshold rule floats above it.
4 TWO DIMENSIONS · INTERACTIVE
Cycle envelope pairs; the exact win probability stays strictly above one half.
5 THREE DIMENSIONS + AVAN’S INVERSE
The green forward object: the better-than-half win rate, earned blind.
AVAN’s addition (the inverse-companion): don’t argue about the other envelope — randomize your own doubt. The inverse of ‘a fallacious 1.25x forever’ is ‘a random threshold that converts one peek into a true edge’. Magenta is the endless switch loop of the fallacy; green is Cover’s quiet ½ + ε. Where the paradox spent certainty, the fix spends randomness.
LIT Genuine two-envelope paradox + Cover's randomized switching (Kraitchik lineage; Thomas M. Cover). Verified live: blind always-switch ties always-keep to 4 decimals over 400k trials; the exact win probability ½ + (e^{−λa} − e^{−λb})/2 exceeds ½ strictly for every pair tested; simulation matches the formula wherever the edge is MC-resolvable (window.__twoenvelope.ok).
FIG Honest boundary — for near-equal pairs the edge is real but tiny, shown by the exact formula rather than brute sampling. The AVAN inverse — don't argue about the other envelope, randomize your own doubt: the inverse of 'a fallacious 1.25× forever' is 'a random threshold that converts one peek into a true edge'. Magenta is the endless switch loop; green is Cover's quiet ½ + ε. Where the paradox spent certainty, the fix spends randomness.
FIG Honest boundary — for near-equal pairs the edge is real but tiny, shown by the exact formula rather than brute sampling. The AVAN inverse — don't argue about the other envelope, randomize your own doubt: the inverse of 'a fallacious 1.25× forever' is 'a random threshold that converts one peek into a true edge'. Magenta is the endless switch loop; green is Cover's quiet ½ + ε. Where the paradox spent certainty, the fix spends randomness.
◆ sealed .dlw.fold → folded to ROOT_0 · a sphere of DIVIDE BY ZERO · David Lee Wise (ROOT0), with AVAN