Skip to main content

The Care Model

Turn Catwatch's observations into a revisable model of your human cat's preferences and history — one that survives contact with change

Due: Wed, 21 Apr 2027, 08:30 · Weight: 25%

Your fieldwork

Turn what Catwatch taught you into a model of your human cat — one you expect to be wrong about, and plan to revise.

Catwatch trained you to keep observed and inferred apart. The Care Model asks what you do with everything that produced: build a structured, honest account of this human cat that grows directly out of your own log — not a fresh character study invented for this assignment. A model that cannot be revised is just a profile wearing a model’s clothes.

Released Wed 17 Mar 2027, 09:00, immediately after Catwatch closes. This fieldwork spans the two-week teaching break. The gap in contact is part of the test: a model that only holds together while you are watching every day is not yet a model of your human cat, it is a model of your own attention.

What your model should hold

Cover, at minimum:

The question behind all of it:

What should you carry forward, and what must remain revisable?

Later in the semester you will meet a sharper version of this question, sorting what you know into what to keep, what to hold lightly, and what to let expire. You do not need that framework to submit good work now — but a model built with revision already in mind will still be standing when you do.

What you hand in

A structured model — a document, a diagram, a short set of rules, whatever form you can justify — built from Catwatch’s observations. Show at least one entry the model has revised, with the reason for the revision, not a silent replacement. Include a short, honest note on what the model still does not know.

Where this goes next

The Care Model is presented, not resubmitted, in the Week 7 Care Lab showcase — and it is the model Know Your Cat spends the rest of the semester living with the consequences of.

What good work shows

These are the things your work should make visible.

How it is marked

CriterionWeight
The model distinguishes recurring pattern from one-off event30%
The model is explicitly revisable, not a fixed profile30%
A real change or contradiction is shown updating the model25%
The model is legible to someone who did not build it15%