The knowledge that isn’t written

the TEA laser, and the papers that didn’t transfer it

today’s seminar

~120 minutes, start to finish

several labs, one set of papers

In the early 1960s, several physics labs tried to build a TEA laser from the published literature alone. The papers were not wrong, not incomplete in any way a referee would have flagged, and not withheld.

what actually predicted success

Harry Collins, studying the attempts, found success correlated with one thing: contact, direct or one step removed, with a lab that already had a working laser. Papers travelled freely. The ability to build the thing they described did not.

never on the record to begin with

This is the sharpest version of tacit knowledge the course will use: not a capability that was lost, but one that was never on the record to begin with. There is no rupture point in this week’s diagram, because nothing here ever broke.

A capability with no record track beside it
A capability with no record track beside it
The TEA laser, by the numbers

discussion

  • If the papers weren’t wrong, incomplete, or withheld, what exactly did they lack?
  • What did “contact with a working lab” supply that the paper itself couldn’t?
  • Is there a skill you have that you could not fully hand to a stranger in writing?
  • If some knowledge is tacit by its nature — never fully write-downable — is “we should have documented it better” ever a complete answer to a lost capability?

activity: write the unwritable instructions

Setup (5 min). Pick a skill you have that you could not fully hand to a stranger in writing — tying a specific knot, a debugging habit, a way of reading a room.

Pairs (20 min). Write instructions for it in five minutes. Trade with a partner and try to follow each other’s instructions literally. Note the exact line where it breaks down.

Share-back (15 min). One breakdown per pair, read aloud.

debrief

Compare what went missing across the room’s instructions against what Collins found missing from the laser papers. Group the misses into kinds: a physical calibration nobody thinks to state, a judgement call, a corrective for when it goes slightly wrong.

wrap-up

Some knowledge was never on the record to begin with — the gap here isn’t a documentation failure, it’s a different kind of thing than a document can hold. Week 3 flips this failure around: a record that is complete and still doesn’t transfer. Week 4 gives you a case built to test today’s open question directly.