/1 · Assumptions — a moral model over Longs
These are the assumptions for /1 Longs. Where longs.csv
holds measured, sourced economic lengths, this set holds foundational human
assumptions — the everyday "lengths" of human nature and construction — and
uses them to build a small moral model: a model whose job is to reason about
code about code, ethics, and Longs together.
Discipline. Everything here is explicitly ASSUMED, not measured. The point is not to assert facts about people; it is to make the modelling assumptions visible and checkable so the moral reasoning built on them is transparent (
ASSUMPTION ≠ FACT,ASSOCIATION ≠ CAUSATION). Change a row and the model's conclusions change with it — by design.
1. What a length is here
A length (a "long") is a count that extends — a quantity with size that a
process lays down over time. In longs.csv a length is an economic magnitude
(GDP, debt, price level). Here a length is a human/construction count-as-length:
| Assumption (example) | The length it names |
|---|---|
| Human lengths per high‑school education | how many life‑norm spans go into finishing HS |
| Life norms per college education | how many lifetime‑radius study passes go into a degree |
| 2×4s per family‑man‑year at Home Depot | is he lengthening (building/growing) or substituting? |
| Dates per pretty girl in a small town | a continuation / procreation signal |
The full seed set (with many more like these) is in
assumptions.csv.
2. The three quantities each assumption carries
- Length — how many units go into one whole (the magnitude of the span).
- Speed — the rate the process runs, in units per year. Speed is "the speed of the process over such concerns": speed of data, of inference, of test. A long span traversed quickly is a different moral object than the same span traversed slowly.
- Lengthening vs. substituting — a flag: does the behavior add real extent to base reality (build, grow, continue) or merely swap one thing for another (substitute)? Lengthening is treated as growth; substitution is treated as churn.
3. What the model computes
For each assumption the model (assumptions.sleela) derives:
| Quantity | Formula (stated so it can be checked) | Meaning |
|---|---|---|
| duration | length / speed (years) | how long the process takes to lay the length down |
| habituality | speed / length, clamped to [0,1] | how habitual/repeated the act is relative to its span |
| growth sign | +1 if lengthening else −1 | does it inhabit base reality or churn it |
| moral score | moral_weight · growth · (0.5 + 0.5·habituality) | signed contribution to the model's moral aggregate |
| relevance | Direct / Model / None on |moral score| | how strongly it bears on base reality |
The relevance classes reuse the /1 (Nordshrift) vocabulary — Direct
(≥ 0.85), Model (≥ 0.50), None (else) — so an assumption's bearing on
base reality is graded the same way a Long's material co‑occupation is.
4. Why this is a moral model of code about code
The moral aggregate ranks assumptions by whether they lengthen (build, continue, habituate a good) or substitute (churn without extent). Applied to software, the same shape scores code about code:
- code that adds durable extent (tests, inference, documented growth) scores like lengthening;
- code that merely substitutes (rewrites without added reach, churn) scores like substitution.
So the model optimizes plan, growth, and habituality: it prefers processes that lay down real length at a sustainable speed and habituate a good, over ones that only swap parts. That preference — made explicit and adjustable — is the ethics layer sitting over Longs.
5. Files
| File | What it is |
|---|---|
assumptions.csv | The seed assumptions (all ASSUMED), with length, speed, category, lengthening flag, moral weight, and a note. |
assumptions.sleela | The moral model over the assumptions (math summary + paragraph evaluation). |
OUTPUT.md | The synthesized opinion the model informs — an overall reading of the health of the human condition (as a product). |
Run it from the repository root:
export SLEELA_SHA256_MANIFEST=security/sha256-manifest.json
export SLEELA_SHEET=SHEET.sheet
impl/build/sleela run churn/1/assumptions/assumptions.sleela
These are foundational modelling assumptions, stated to be argued with. The moral scores are a transparent function of the assumed inputs, not a judgment of any person.