State the hypothesis

The first stage. It exists because research without a falsifiable claim only finds confirmation — you gather what fits and never notice what doesn't.

A hypothesis is not a topic. "Spec-driven development" is a topic. "SDD should be called Spec-Driven Delivery because the practice extends past development" is a hypothesis: it can be checked, and it can turn out to be wrong.

Prerequisite

None. This is the entry point.

Write it to the working doc

One file per post: docs/research/<slug>.md, snake_case slug matching the eventual post. Create it with the stage block and the hypothesis section:

# <Working title>

## Stage

- [x] Hypothesis
- [ ] Research
- [ ] Thesis _(gate — human approval)_
- [ ] Outline
- [ ] Draft

## Hypothesis

**Claim.** One sentence, falsifiable.

**Who it's for.** Who changes their mind or their practice if this lands.

**What would falsify it.** The specific finding that would kill or badly wound the claim.

**Why I might be wrong.** The strongest version of the opposing case, written honestly.

The four fields, and why each is required

Field	Purpose	Failure if skipped
Claim	Gives research something to test	Research becomes topic-browsing
Audience	Sets register and depth	Piece argues with nobody in particular
Falsifier	Makes the claim honest	You cannot tell success from confirmation bias
Why wrong	Surfaces the rival case early	The objection arrives from a reader instead

The falsifier is the one people skip and the one that matters most. If you cannot name a finding that would change your mind, you are not writing an argument — you are writing an opinion with citations attached.

Do not

• Do not research first and back-fill the hypothesis. A hypothesis written after the evidence is a summary, and it will quietly match whatever you happened to find.
• Do not soften the claim to make it safer. A hedged claim cannot be falsified, so research cannot inform it. Overstate slightly; the thesis stage will correct it.
• Do not skip "why I might be wrong" because the case seems obvious. On this site's own SDD post, the rival framing (goal-driven development) already existed in public and was found during research, not before it.

Expect the claim to change

It usually does, and that is the process working rather than failing. The blog-thesis stage exists specifically to reconcile this hypothesis against what the evidence supported. Recording the claim now is what makes that comparison possible later.

Next step

→ blog-research to gather and verify sources against this claim.