What we didn’t finish#
Week 12. The last one. Nothing is due.
A course called Unfinished Thinking that ended with a summary would be getting its own joke wrong#
So here is the ledger instead.
The central claim is untested#
That a loop repeats because the thinking never gets the conditions under which it would finish.
It is assembled from real findings. Incubation during undemanding tasks. Attention residue. The phone break that measures like no break.
Nobody has tested the claim directly, including us. It is the best account we could build, and it is still a hunch with good parts.
Three things we refused to claim#
They are on the deck and they have not moved all semester.
The temptation to quietly upgrade them at the end of a course is exactly what this lecture exists to resist.
Two traditions poked holes we never filled#
Week 8 gave you Morita, saying the fix is not mechanical, and the Friendship Bench, saying it is not private.
Both have at least as much support as the course’s own claim.
Neither is in it.
Most of it is small#
| Technique | Roughly |
|---|---|
| Worry postponement | d = 0.31 and 0.19 |
| Expressive writing | about d = 0.15 |
| Exercise on state anxiety | a quarter of a standard deviation |
| If-then plans | d = .65, the oldest and least re-examined number here |
| Sugar on mood | measurably backfires |
Several of these help a bit, one backfires, and the biggest number is the least scrutinised#
That is the honest summary of a semester.
Some of it we got wrong and fixed in public#
The sequencing of this course was built on concrete-to-abstract evidence, then corrected against Luiten, Ames and Ackerson’s meta-analysis, which says learners do better with a brief organising idea up front.
That is why week 1 states the argument in a sentence and then abandons it until week 5.
The Lectures page carries the full correction, including the contested paper saying the whole approach might be backwards.
One thing we could not find at all#
The assessment calendar is built around avoiding everyone else’s deadlines.
We could not find a study isolating clustered against spread deadlines and measuring wellbeing.
Assessment labels it design reasoning, not a finding, because that is what it is.
Strip the content out and three things survive#
Noticing the seam. At which layer did we stop learning about the message and start projecting? That question works on anything, and it needs none of the research to be true.
Checking what you are crediting. You had a break and credited the sugar.
Knowing what would change your mind. State a prediction specific enough to be wrong. Applies to every claim here, including ours.
Where to keep going#
Not here.
The log asks what you found elsewhere that worked better than anything we taught you, and those answers change the syllabus.
Three of the crits you sat this year came out of a previous cohort answering that question.
Nobody teaching this has stopped overthinking.
That was on the Policies page in week 1. It is still true in week 12.
Last four questions#
- Which claim from this semester are you least convinced by, and what would it take?
- What did you find outside this course that beat anything in it?
- If you had to delete one week to make room for something better, which, and what goes in?
- The course argues a thought needs conditions in order to finish. Has anything you were carrying in February actually finished?