Three assignments on one target: reduce the options, commit on evidence, then find out what the commitment cost.

You pick a target before the first assignment in week 4 and keep it for the rest of the semester. The three assignments are the same problem at three stages, and each is marked on your reasoning rather than on whether the thing runs.

Choose that target carefully. Something trivial leaves you with no trade-offs to analyse and very little to write about. Something enormous leaves you unable to measure anything. You also choose your own new requirement in the final assignment, so pick something with room to grow.

The last assignment asks you to mark your own first one. That is not a formality: calibration — knowing when you became wrong, and what evidence should have caught it — is the single largest criterion in the course.