# v8 — IDEAS FOR NEW IDEAS

**Not another hypothesis set.** The five expeditions produced 1,500 hypotheses, 150 dossiers and 15
investigations — all of them *claims*. This folder contains the machinery instead: operations that
generate ideas, inventions with a first test, questions with an identified blocker, and candidates for
what is invisible.

---

## What is here

| Folder | Contents | What it is for |
|---|---|---|
| `00_MAP_OF_UNKNOWNS/` | a taxonomy of fourteen kinds of ignorance | diagnosing *why* something is unknown |
| `01_GENERATORS/` | 30 operations (15 from measurement, 15 from theory) | applying to your own problem |
| `02_INVENTIONS/` | 50 devices/processes with mechanism + first test | building something |
| `03_DISCOVERIES/` | 50 open questions with a decisive measurement | deciding what to measure |
| `04_UNKNOWN_TERRITORY/` | 20 candidates for the invisible, each with why it is probably nothing | knowing where to look |
| `05_UNASKED_QUESTIONS/` | 25 questions whose absence has a structural cause | interrogating the field |
| `06_TOOLS/` | brief, build script, index | maintaining the set |

Start with `INDEX.md` for the full title list.

---

## The three categories, and why they matter

**Generators** are the transferable part. A hypothesis dies with its test; an *operation* survives,
because it can be applied to problems that did not exist when it was written. The set includes moves
like: make the blank the sample; compute what the instrument cannot see; read the null space of an
inverse problem; find the floor of an asymptotic series; census the cancellations.

**Inventions** are constrained to have a *mechanism* and a *first test*. The brief forbids "nobody has
tried it" as an explanation for absence — the obstacle must be technical. That constraint is what
separates an invention seed from a wish.

**Unknown territory** is the honest part. Genuine unknown unknowns cannot be listed: anything written
down has, by that act, left the category. So this section does something narrower — it names
candidates whose invisibility has a *structural* cause, and states for each why it is probably
nothing. The five structural blind spots it works from:

1. Where a correction is applied automatically — whatever it removes is not in anyone's data.
2. Where two conventions meet — a silently absorbed factor is invisible on both sides.
3. Where a measurement is a proxy — the proxy's behaviour becomes the object of study.
4. Where an effect is labelled an error — noise, drift, artefact, leakage.
5. Where no incentive exists — no paper, no product, no grant.

All five are properties of the knowledge machine rather than of nature. That is the useful consequence:
**if something is hiding, it is more likely behind a structural blind spot than in an exotic regime.**

---

## How to use it

- **You have a problem.** Read `01_GENERATORS`, apply three operations.
- **You have a laboratory.** Pick from `02_INVENTIONS` by cheapest first test.
- **You have data.** Read `05_UNASKED_QUESTIONS` for combinations nobody has run.
- **You want to know where to look.** Read `04_UNKNOWN_TERRITORY`, honest-status lines first.

---

## What this folder deliberately is not

- **It is not a sixth expedition.** No 300-item hypothesis list, no dossiers. The marginal value of a
  sixth 300 is low; the marginal value of better generators is not.
- **It makes no novelty claims.** No literature search was run for any entry. A reader who finds one
  of these already published has found a defect — and finding defects is the intended use.
- **It does not claim to have mapped the unknown.** The map is of *how things go missing*, not of
  *what is missing*.

---

## Honest limitations

- The generators are only as good as their worked examples; each may fail outside its home field, and
  the "fails when" line is stated for that reason.
- The inventions are seeds, not designs. None has been built or simulated.
- The unknown-territory entries are guesses with a rationale, and 20 of them may all be nothing.
- The questions may already have answers in literatures nobody searched.