# The Twelve-Week Judgment Curriculum

> Juniors used to learn craft by doing the work agents now do. That path is
> closed, so apprenticeship has to move from production to judgment, and judgment
> has to be taught deliberately because it will not be absorbed by osmosis
> anymore. Twelve weeks, one exercise a week, each producing something real.

---

## How to run it

One designer, one mentor, one hour a week plus the exercise itself. The exercise
is done on **real work in flight**, never on invented scenarios. Invented
scenarios teach people to perform judgment rather than exercise it.

Each week has the same shape: the mentor does it first while the junior watches,
the junior does it, they compare, and the difference is the lesson. The comparison
is the whole method. Skip it and this becomes a reading list.

Assessment happens three times: week 4, week 8, week 12. Rubric at the bottom.

---

## Weeks 1 to 4: Seeing

The goal is noticing what is wrong before knowing how to fix it.

**Week 1. Grade thirty outputs.**
Pull thirty real outputs from a live surface. Grade each good, mixed, or bad, one
sentence why. The mentor does the same set independently. Compare.
*Produces:* the raw material for a rubric.
*The lesson:* where their instincts already match, and where they are graded on
things that do not matter.

**Week 2. Find the failure nobody logged.**
Spend the week using the product like a real user in the messiest account they
can get access to. Write down every moment something was wrong and nobody would
have noticed.
*Produces:* five entries for the failure inventory.
*The lesson:* invisible failure is the expensive kind.

**Week 3. Rewrite three principles as constraints.**
Take three lines from an existing design principles doc. Rewrite each as a rule
with a failure condition. Test them with someone who was not in the room.
*Produces:* three checkable constraints.
*The lesson:* what specific means.

**Week 4. Write the range spec for a shipping feature.**
Best case, worst acceptable, refuses to. Then sort ten real outputs against it.
*Produces:* a range spec.
*Assessment 1.*

---

## Weeks 5 to 8: Deciding

The goal is making a call and being able to defend it in writing.

**Week 5. Kill an option.**
Given four credible directions for a real problem, pick one and write why the
other three die. Each rejection names what it would have bought and why that is
not worth the cost.
*Produces:* a decision record.
*The lesson:* a decision is only real if the alternatives were real.

**Week 6. Own a wrong path end to end.**
Take one failure mode and design the whole response: detection, disclosure,
recovery, copy. Ship it.
*Produces:* shipped work on the unhappy path.
*The lesson:* the unhappy path is where the relationship is decided.

**Week 7. Write the refusal set.**
For one surface, every refusal, with the category and the nearest available
action. Read them aloud.
*Produces:* a refusal set.
*The lesson:* boundaries are a design output, and copy is design.

**Week 8. Argue against your own decision.**
Take the week 5 decision record and write the strongest case for the option they
killed. Mentor judges which case is stronger.
*Produces:* the case against.
*Assessment 2.* If the case against is stronger, the original decision gets
revisited, and that is a success, not a failure.

---

## Weeks 9 to 12: Scaling

The goal is judgment that operates without them in the room.

**Week 9. Extract a rubric.**
From the week 1 grading, pull dimensions, apply the cost test, write binary tests.
*Produces:* a rubric v1.

**Week 10. Calibrate it.**
Score five outputs against another person. Classify every disagreement. Rewrite.
*Produces:* a rubric v2 and a disagreement log.
*The lesson:* vague is discovered, not predicted.

**Week 11. Run a review.**
Facilitate a real thirty-minute rubric review. Mentor watches, does not
intervene, debriefs after.
*Produces:* a scored review with a decision.
*The lesson:* running the meeting is a different skill from having the judgment.

**Week 12. Hand it over.**
Write the constraint set, rubric, and failure inventory for a surface, and hand
them to somebody who was not involved. That person works from the documents
alone.
*Produces:* a specification a stranger can execute.
*Assessment 3.* The assessment is simply whether the stranger produced work the
junior recognizes as right.

---

## The assessment rubric

Same four dimensions at weeks 4, 8, 12. Score each not yet / getting there /
solid.

| Dimension | Solid looks like |
|---|---|
| **Specificity** | Writes rules a stranger can check. No principles in disguise. |
| **Cost reasoning** | Can name what a failure costs, in the units the business uses. |
| **Decisiveness** | Makes the call, names what died, does not hedge in the record. |
| **Transferability** | Their judgment produces the same result when applied by someone else. |

The fourth is the one that matters most and takes longest. It is also the only one
that distinguishes judgment from taste.

---

## What this replaces

It replaces the years spent producing screens to absorb craft by repetition. The
production path taught judgment as a side effect, slowly and unreliably, and it is
gone.

The honest tradeoff: the old path also taught craft in the hands, and this does
not. Juniors on this curriculum will be better at specifying and worse at making
than the previous generation at the same stage. Whether that matters depends on
your product, and it is worth deciding on purpose rather than discovering later.

---

## The one thing to do this week

Run week 1 with one junior on your team. Thirty outputs, graded independently,
compared. It takes ninety minutes and it will tell you more about where their
judgment actually is than a year of portfolio reviews.

---

_From "The Ladder, Rebuilt", chapter 19 of The Design Operating Model._
_falkster.com/design/the-ladder-rebuilt_