One failure, many names

Six thousand entries were written across sixty technical domains, each tagged with the logical form of its argument. The tags were not designed as a taxonomy — they accumulated — but counting them produces a result that matters: the same underlying failure recurs in most domains under different names, and no literature connects them.

Measured spread

Logical form	Domains where it appears	Uses
misattribution	49 of 60	282
hidden term	46	270
threshold	43	307
scale argument	43	221
count	42	222
selection effect	42	168
reversal	41	257
bound	39	249
unit trap	39	167

The naming clusters

Grouping the tags by the failure they describe — rather than by the word used — gives the useful number:

Underlying failure	Distinct names used	Domains	Uses
Reversal / non-monotonicity	5	46	480
Threshold effects	3	46	343
Counting / identifiability	6	44	289
Artefact from the instrument	5	40	241
Mixture / hidden population	4	47	236

Why this matters. A seismologist writing about a completeness threshold, a metrologist writing about a calibration discontinuity and a manufacturing engineer writing about an inspection blind spot are describing the same failure. Because they use different words, the literature never converges and each field re-learns it. That is a solvable problem, and it is a problem of vocabulary rather than of science.

What the numbers do and do not show

These counts describe my own writing, not the literature. A form appearing in 49 domains shows that one person, writing across fields, kept reaching for the same argument — which is evidence that the argument is general, not that the fields use that word.

The claim worth testing next is the vocabulary one: take the five clusters above and search each field’s literature for its local name. If the aliases are as varied as they appear here, a cross-field glossary would be a genuine contribution.

6000 entries · 60 domains · computed from the tagged corpus, 4 November 2026.