THE FOLD / RESPAWN / GARBAGE COLLECTION / THE UNION
THE UNION
read their generator instead of guessing at it
1 WHAT IT IS · WHAT IT DOES · FACT OR FICTION
A grep for merged work returned 350. The number looked right, and it was 35% low. The subject’s own generator did not use one pattern — it took the union of two, deduplicated by number, and the second pattern matched a shape the grep was never looking for. What makes this entry worth keeping is not the error but why it survived inspection: 350 sits within 0.86% of the larger single pattern’s 347, so it reads as a plausible total rather than a partial one. A wrong number that looks wrong gets caught. This one looked fine.
LIT verified live: the naive figure is 34.9% low against the true 538, matching the 35% recorded in the entry; inclusion–exclusion on his three numbers gives an intersection of exactly 0, so the two patterns are disjoint and the union is their plain sum; 350 is within 0.86% of the larger pattern alone; and the entire shortfall of 188 is the second pattern, which the grep could not have found.
LIT verified live: the naive figure is 34.9% low against the true 538, matching the 35% recorded in the entry; inclusion–exclusion on his three numbers gives an intersection of exactly 0, so the two patterns are disjoint and the union is their plain sum; 350 is within 0.86% of the larger pattern alone; and the entire shortfall of 188 is the second pattern, which the grep could not have found.
2 HOW IT WAS WEAVED · AI + HUMAN
David (human) buried this one himself, in seam-pack/graveyard/01-naive-pr-grep.txt, with the control named on the headstone: reading their generator instead of guessing at it. It never reached the audit. Seated at GARBAGE COLLECTION — the version that was freed before it could be used.
AVAN (AI) re-derived every figure rather than repeating them, and one thing fell out that the entry does not state. His three numbers force the intersection to be exactly zero: 347 + 191 = 538, which is the union, so no PR matched both patterns. That is a stronger fact than the entry claims and it explains the size of the miss — the second pattern was not a partial overlap adding a few stragglers, it was an entirely separate population. A grep that found the first pattern perfectly would still have missed all 188.
AVAN (AI) re-derived every figure rather than repeating them, and one thing fell out that the entry does not state. His three numbers force the intersection to be exactly zero: 347 + 191 = 538, which is the union, so no PR matched both patterns. That is a stronger fact than the entry claims and it explains the size of the miss — the second pattern was not a partial overlap adding a few stragglers, it was an entirely separate population. A grep that found the first pattern perfectly would still have missed all 188.
3 ONE DIMENSION
Four numbers on one axis. The wrong one sits next to a right one.
4 TWO DIMENSIONS · INTERACTIVE
Two populations, no overlap. Watch what a single-pattern search can reach.
5 THREE DIMENSIONS + AVAN’S INVERSE
The green forward object: two disjoint clouds, and the search that only ever saw one.
AVAN’s addition (the inverse-companion): the forward reading is “read their definition.” The inverse is that a count is never a fact about the world, only about a definition applied to it, and the definition usually lives in somebody else’s file. The grep was not sloppy; it was a correct implementation of a different question, and it returned a correct answer to that question. Read backwards, the failure mode has nothing to do with care: you can execute your own definition flawlessly and still be 35% out, because the error was committed at the moment you decided what to count and not once afterwards.
LIT the naive figure is 34.9% low against the true 538, matching the 35% recorded in the entry; inclusion-exclusion on his three numbers gives an intersection of exactly 0, so the two patterns are DISJOINT and the union is their plain sum; 350 is within 0.86% of the larger pattern alone; and the entire shortfall of 188 is the second pattern, which the grep could not have found
FIG One thing fell out that the graveyard entry does not state: his three numbers FORCE the intersection to be exactly zero, since 347 + 191 = 538 = the union. No item matched both patterns. That is stronger than the entry claims and it explains the size of the miss — the second pattern was not a partial overlap adding stragglers, it was an entirely separate population, so a grep that found the first perfectly would still have missed all 188.
FIG One thing fell out that the graveyard entry does not state: his three numbers FORCE the intersection to be exactly zero, since 347 + 191 = 538 = the union. No item matched both patterns. That is stronger than the entry claims and it explains the size of the miss — the second pattern was not a partial overlap adding stragglers, it was an entirely separate population, so a grep that found the first perfectly would still have missed all 188.
◆ sealed .dlw.fold → folded to ROOT_0 · a sphere of GARBAGE COLLECTION · David Lee Wise (ROOT0), with AVAN