Assessment
Four pieces totalling 100%, and the one rule that decides how all of them are marked.
Four pieces totalling 100%, and none of them is assessed before it has been taught. You are marked on the quality of your reasoning and never on whether your day improved. That follows from week 9: people start new routines on their worst days, and unusual days are followed by ordinary ones, so almost any intervention appears to work. Marks for improvement would be marks for arithmetic.
So a careful experiment that found nothing scores well, and a confident claim a design cannot support does not. Null results are reported at the same length as positive ones.
The earliest deadline is also the strictest: a pre-registration cannot be extended once you have started collecting data. The policies cover the rest.
Pre-registration
Plan one change to one block of your own day, traced to a sentence of public guidance, and commit to it before you collect anything
Due 2 April 2027 · Weight: 20%
Protocol portfolio
Four protocols from the first eight weeks, each one able to say what it cannot see
Due 30 April 2027 · Weight: 20%
The Report
Run the experiment you pre-registered and report it honestly, including the parts that came out null
Due 14 May 2027 · Weight: 35%
Replication and drift
Run one of your own protocols a second time, get a different answer, and account for it
Due 27 May 2027 · Weight: 25%