Week 10 Seminar: Detection

Every case so far was diagnosed in retrospect. What would evidence of a loss in progress look like?

26 April 2027

Open the slides

The case, or the argument#

Look back across weeks 2 through 9: FOGBANK, the F-1, deskilling, automation surprise, contested craft losses — every one was noticed after the fact, when the capability was needed and turned out not to be there (or, in week 9’s case, argued about after the fact with no live capability to test at all). None of this course’s cases is a story about an institution that caught the gap while there was still time to act on it cheaply.

This week turns the method around: instead of diagnosing a loss that already happened, what would an automation surprise or a FOGBANK-style gap look like while it was still forming, rather than after it had already cost years and a rebuilding budget? The honest answer is that this is harder than diagnosis, precisely because the diagnostic tools this course has built so far — the record track, the capability track — are read from documents that describe a state, not from a live signal that a gap is opening.

What we do in the seminar#

For each of weeks 4, 5, 7 and 8, identify the earliest point at which someone inside the institution plausibly could have noticed the gap forming, using only information available at that time. Mark it on the week’s diagram as an early rupture point and compare it to where the institution actually noticed.

One reading#

Bainbridge (1983) — reread for its practical suggestions about maintaining operator skill under automation; several are, in effect, early-detection measures rather than after-the-fact fixes.

A question left open#

If detection is harder than diagnosis, is it because the evidence genuinely isn’t there before the capability is needed, or because nobody is looking for it — and which of those two is true of your own organisation, if you have one?