Station 02 · The Cutting Bench

Find the cut lines.

The research production line looks like one workflow. It's actually six output forms strung together. Cut where the form genuinely changes — not on activity boundaries, not on calendar boundaries.

Cut where the output form changes.

👩‍🔬

Marie thinks…

The Goldilocks rule. Cut too coarse, you get one fragile mega-skill that fails unpredictably and you can't debug. Cut too fine, you spend more time on handoffs than on the science. Cut at form-change boundaries and the joins take care of themselves.

Worked example

The research production line, decomposed.

Ten activities. Five cuts. Six skills. Each cut sits where the output form genuinely changes: prior-art map → hypothesis → protocol → fitted data → figure → prose.

1. read background · 2. id open questions │ CUT │ 3. pick one to test · 4. predict signature │ CUT │ 5. design the rig · 6. pick materials & controls │ CUT │ 7. run measurement │ CUT │ 8. fit data & estimate error │ CUT │ 9. make figures │ CUT │ 10. write the paper

Each cut sits where the output form genuinely changes: literature corpus → claim → protocol → raw traces → fitted parameters → publication figure → manuscript prose.

Try this

Four candidate cut points. Yes or no?

Click "cut here" or "leave joined" on each.

0 / 0 answered

Before you scroll on

Pick your own workflow. Where are the form-change cuts?

Take the research task you held in your head from Station 01. Try to list it as 6-10 micro-activities and mark where the output form actually changes. Don't overthink the first draft — the first cut is usually wrong, and that's fine.