The labs
Eleven labs across weeks 2 to 12, each one finished in the room, with the best ten counting
Due: 24 May 2027 · Weight: 20%
The brief#
Eleven labs, one in each of weeks 2 to 12. You finish in the room.
Each lab hands you a new codebase with a part of it missing, and one source of difficulty isolated so that its effect on a decision is legible. You implement the missing part and submit before the session ends. There is nothing to take home, and nothing to finish afterwards.
The labs do not add up to your instrument. No lab extends the one before it, and no lab’s code is the thing you submit in week 12 — the ideas combine in the assignments, in new codebases written for those. One lab artefact travels literally: the wall configuration from week 11 is submitted with the instrument unchanged. Week 12’s assessed output is an audit and domain statement for a supplied model. It also leaves you with a blank audit plan. You apply that plan to your own instrument and write its separate domain of validity before Friday; the supplied model’s domain is not submitted as your own. What each lab isolates is chosen for what an assignment will later make it collide with.
The schedule#
Week 1 has no lab. From week 2 there is one every week through to week 12, with the two-week break falling between weeks 6 and 7. The date above is the last lab, in week 12.
Each lab’s own brief states what it asks for and what it is marked on. The criteria live there, not here, because they are different every week.
A reference solution is released once each lab closes: the same day, and never before it.
Why ten and not eleven#
A lab is completed and submitted inside the session, so a missed lab cannot be made up. There is no version of it you can do at home, because the constraint that makes a lab finishable in two hours is the specification you were handed in the room.
Dropping the lowest of your eleven marks is what makes that rule survivable. It is not a concession, and it is not a lab you are invited to skip: each of the eleven is worth 2%, ten of them count, and the one that falls away is there so that a single bad week, or a single illness, costs you nothing.
The spec
The brief poses the problem and leaves room for a response. The spec is the fixed part: what a marker checks. Some lines can be checked mechanically; the rest call for judgement.
- each lab is submitted from the room before that session ends
- each lab is marked against the criteria printed in its own brief
- ten of your eleven lab marks count towards the 20%
How it is marked
In-session completion against each lab's own stated criteria, with the best ten of the eleven counting. A missed lab cannot be made up.
There is no second standard here. Each lab is marked inside its session, on what left the room, against the criteria printed in its own brief — and those are written so a marker can tell whether a line has been met without asking you.
What makes a lab convincing is that its own printed criteria are met rather than approximated. Where a lab asks for both a working output and the validation behind it, one without the other does not meet what it asked for — which is a way of reading that brief, not a standard this page adds to it.