Regional case survey#

Week 7 — desk study begins

Three requirements before a case counts#

1. Sourcing — does it name where it came from?
2. Behaviour, not atmosphere — does it describe what happened, not just how it felt?
3. Corroboration — is there more than one account behind it?

A striking case that fails any of these still gets read. It just doesn’t get weight in a comparison.

Failure mode 1: official-sounding, corroborated by nothing#

• The case — facility-scale, with a published containment procedure
• What audit found — no demonstrated causal relationship between the procedure and the anomaly’s actual behaviour
• The lesson — a procedure can read as authoritative while remaining pure narrative, with no working countermeasure behind it

Failure mode 2: structurally uncorroborable by design#

• The case — its own mechanism is “everyone who could confirm it either passed it on or is compromised”
• The problem — no position exists from which an independent witness could ever check it
• The call — read for method, never weighted in a comparison

Failure mode 3: no evidence at any point in its history#

• The case — a device supposedly appeared, caused documented harm, then was removed by unmarked agents
• The check — no unit, registration, or contemporary account has ever surfaced, at any stage
• The distinction — not “hard to verify now”: nothing to verify existed in the first place

Case-format literacy#

The same three questions apply whether a case arrives as a text write-up or a video/software claim: the format changes what a fake looks like, not whether the test still applies.

• Text case: an object motionless only under continuous direct observation by two or more witnesses, contact injuries reported exclusively during momentary lapses.
• Video/software case: a claimed behaviour on period hardware. The sourcing question here is whether that behaviour is even possible on the hardware claimed; footage looking real isn’t sufficient on its own.

The week-2 reveal#

Held back on purpose since week 2: an object with a genuinely affectionate, unchanging baseline that, rarely and only while unobserved, does something else entirely (scavenged materials, crude and violent, then nothing again). Classified persistent and dormant, two tracked channels, not one forced fit.

Why it was held back#

Week 2 taught the classification move on this case without the story behind it, on purpose. You now have the sourcing method too — read this case for what it corroborates, past how striking it looks.

A countermeasure that made things worse#

A secondhand cartridge, an unfamiliar save file, escalating glitches, and the standard fix, wiping and resetting it, is the documented action that made it worse. The obvious fix is itself an action, not a neutral default.

Halloween candy tampering: a legend’s effect as data#

Newspaper survey, four papers, 1958–84: 76 reported incidents, injuries in only 20, no verified stranger-poisoning deaths — and real defensive behaviour every October regardless. The legend’s content doesn’t hold up. Its effect on real behaviour is itself worth treating as data.