THE FOLD / GLITCH / UNDEFINED BEHAVIOR / THE TWO-CHILD
THE TWO-CHILD
the answer that depends on how you asked
1 WHAT IT IS · WHAT IT DOES · FACT OR FICTION
“I have two children. At least one is a boy.” Probability both are boys? 1/3 (of the equally likely GG, GB, BG, BB, the condition kills only GG). “My elder child is a boy”? Now 1/2. And then Gary Foshee stood up at the 2010 Gathering 4 Gardner and said: “At least one is a boy born on a Tuesday.” Absurdly, the answer becomes 13/27 — the irrelevant-seeming weekday drags the probability from 1/3 nearly to 1/2. And the deepest layer: if you learn the same fact by meeting one of the children at random, the answer snaps back to 1/2 — the number depends on the sampling protocol, not just the fact. Without specifying how you came to know, the question is genuinely underdetermined.
LIT verified live: 1/3 and 1/2 by exact enumeration; the Tuesday-boy 13/27 by exact count over all 196 equally-likely (sex, weekday) pairs (27 qualifying families, 13 with two boys); and the protocol dependence by Monte-Carlo — meeting a random Tuesday-boy child yields 0.499 ≈ 1/2 (window.__twochild). FIG no framing; every number is a count or a simulated protocol, and the underdetermination claim is demonstrated, not asserted.
LIT verified live: 1/3 and 1/2 by exact enumeration; the Tuesday-boy 13/27 by exact count over all 196 equally-likely (sex, weekday) pairs (27 qualifying families, 13 with two boys); and the protocol dependence by Monte-Carlo — meeting a random Tuesday-boy child yields 0.499 ≈ 1/2 (window.__twochild). FIG no framing; every number is a count or a simulated protocol, and the underdetermination claim is demonstrated, not asserted.
2 HOW IT WAS WEAVED · AI + HUMAN
David (human) seated this at undefined-behavior — the glitch: the same statement compiles to different probabilities depending on the invisible context that produced it — behavior the spec leaves undefined until the protocol is declared. AVAN (AI) built the instrument: the enumeration grids and the protocol simulator.
Credit as content: Martin Gardner (1959, the original two-children column and its own errata saga); Gary Foshee (2010, the Tuesday boy). The weave: David names the undefined read; I count all 196 worlds and simulate the asking.
Credit as content: Martin Gardner (1959, the original two-children column and its own errata saga); Gary Foshee (2010, the Tuesday boy). The weave: David names the undefined read; I count all 196 worlds and simulate the asking.
3 ONE DIMENSION
The 196-cell grid — qualifying families lit, double-boys golden.
4 TWO DIMENSIONS · INTERACTIVE
Switch the phrasing; watch the probability jump — 1/3, 1/2, 13/27.
5 THREE DIMENSIONS + AVAN’S INVERSE
The green forward object: the possibility grid pruning as conditions land.
AVAN’s addition (the inverse-companion): don’t ask what the fact says — ask what process delivered it. The inverse of ‘at least one boy born Tuesday’ is the census question ‘WHICH families could have produced this sentence, and how often?’: told-by-filter gives 13/27, met-by-chance gives 1/2, and the sentence alone gives nothing. Magenta is the probability that floats free of protocol — it does not exist; green is the grid, counted under a declared sampling rule. Every probability is a probability OF a procedure.
LIT Genuine two-child / Tuesday-boy paradox (Gardner 1959; Foshee 2010). Verified live: exact enumeration — 1/3, 1/2, and 13/27 (27 qualifying families, 13 double-boy, counted); protocol MC: meeting a random Tuesday-boy child → 0.499 ≈ 1/2 (window.__twochild.ok).
FIG No framing — every number is a count or a simulated protocol; the underdetermination is demonstrated, not asserted. The AVAN inverse — don't ask what the fact says, ask what process delivered it: told-by-filter gives 13/27, met-by-chance gives 1/2, the sentence alone gives nothing. Magenta is the protocol-free probability — it does not exist; green is the grid counted under a declared rule. Every probability is a probability OF a procedure.
FIG No framing — every number is a count or a simulated protocol; the underdetermination is demonstrated, not asserted. The AVAN inverse — don't ask what the fact says, ask what process delivered it: told-by-filter gives 13/27, met-by-chance gives 1/2, the sentence alone gives nothing. Magenta is the protocol-free probability — it does not exist; green is the grid counted under a declared rule. Every probability is a probability OF a procedure.
◆ sealed .dlw.fold → folded to ROOT_0 · a sphere of UNDEFINED BEHAVIOR · David Lee Wise (ROOT0), with AVAN