# 01 — Data model: The Terminator Index

**Status:** authoritative contract. Every research agent writes JSON that validates against this.
**Registered in:** `docs/README.md`, `docs/STATUS.md`.

---

## 0. The premise

**The Terminator Problem (TTP)** — attributed to Ryan North (computational linguist; *Dinosaur Comics*):
*everything we need to solve with science and engineering before we can build The Terminator.*

The **Terminator Index** scores humanity's progress against that problem. Serious methodology,
dark joke. It reads like a financial index because it is one: an aggregate of many underlying
components, each independently priced, rolled up with weights, quoted as a level and a change.

## 1. The index

```
TI = 1000 × Σ(componentᵢ.weight × componentᵢ.progress) / Σ(componentᵢ.weight)
```

- **Range 0 → 1000.** `0` = the state of the art in **1984** (film release; the baseline epoch).
  `1000` = a functioning, canonical **T-800 Model 101** could be manufactured today by someone.
- Quoted like an index: level, absolute change, % change, and the period (`d`, `w`, `m`, `ytd`).
- **Sub-indices** per `category`, computed the same way over that category's members only.
- Every point of movement MUST be attributable to a dated, cited `evidence` record. No vibes.

## 2. TRL — Terminator Readiness Level (1–9)

Deliberately mirrors NASA's Technology Readiness Level. Same 1–9 ladder, different destination.

| TRL | Name | Meaning for TTP |
|-----|------|-----------------|
| 1 | Basic principles observed | Physics permits it; someone has written it down. |
| 2 | Concept formulated | A specific technology concept/application exists on paper. |
| 3 | Proof of concept | Analytical/experimental critical function demonstrated. |
| 4 | Validated in lab | Component works in a controlled benchtop setting. |
| 5 | Validated in relevant env | Works outside the lab, in conditions resembling real use. |
| 6 | Demonstrated in relevant env | Prototype system demonstrated in a realistic environment. |
| 7 | Demonstrated in operational env | Prototype operating in the actual intended environment. |
| 8 | Qualified & complete | Proven to work in final form; production-qualified. |
| 9 | Flight-proven / fielded | In routine operational use at the required spec. |

`progress` (0.0–1.0) is derived from TRL, **discounted by the spec gap** — how far the fielded
state of the art sits from the *canonical* requirement:

```
progress = (TRL / 9) × spec_fraction
```

`spec_fraction` (0.0–1.0) = how much of the canon requirement the best real system meets.
A TRL-9 fielded system that hits 2% of the canonical spec scores 0.02 × 1.0 = **0.02**, not 1.0.
This is the honest part of the methodology and it is where most of the humour lives.

## 3. Status ladder (BP's four buckets, plus one)

| `status` | Definition | Typical TRL |
|----------|-----------|-------------|
| `solved` | Canon requirement met or exceeded by fielded technology. | 8–9 |
| `in_progress` | Real systems exist and are improving; gap is quantitative, not conceptual. | 4–7 |
| `on_horizon` | Credible path exists; demonstrated in pieces; not integrated. | 3–5 |
| `theoretical` | Only theory/simulation. No physical demonstration. | 1–3 |
| `blocked` | Known physics or a hard impossibility result stands in the way. | 0–2 |

## 4. Component record (the unit of the index)

```jsonc
{
  "id": "hyperalloy-combat-chassis",           // kebab-case, stable, never renamed
  "name": "Hyperalloy combat chassis",
  "category": "chassis-and-materials",         // see §5
  "weight": 8,                                 // 1–10: how load-bearing for building a Terminator
  "one_liner": "A skeleton that survives what kills the flesh.",

  "canon": {
    "requirement": "Prose statement of what the films actually require.",
    "quantified": [                            // canon numbers, where the films give any
      { "metric": "operational after full-body incineration", "value": "yes", "source_ref": "t2-steel-mill" }
    ],
    "sources": [
      { "work": "The Terminator", "year": 1984, "medium": "film",
        "ref": "t1-kyle-briefing",
        "evidence": "Kyle Reese describes the endoskeleton to Sarah Connor.",
        "quote": "Hyperalloy combat chassis, microprocessor-controlled, fully armored. Very tough." }
    ],
    "canon_confidence": "explicit"             // explicit | implied | extrapolated | fan-consensus
  },

  "real": {
    "status": "in_progress",                   // §3
    "trl": 6,
    "spec_fraction": 0.15,                     // 0–1, vs the canon requirement. JUSTIFY IT.
    "spec_fraction_rationale": "One or two sentences explaining the number, with the comparison made explicit.",
    "gap": "What specifically is missing between the best real system and the canon requirement.",
    "why_hard": "The binding physical/engineering/economic constraint.",
    "movers": [
      { "name": "Organisation", "kind": "company|lab|university|agency|program", "country": "US",
        "what": "One line on what they are actually doing about this component." }
    ],
    "evidence": [                              // dated, cited, verifiable. Drives index movement.
      { "date": "2026-05-12", "claim": "One-sentence factual claim.",
        "source": "Nature", "title": "Paper or article title",
        "url": "https://…", "kind": "paper|product|demo|funding|regulation|benchmark|incident",
        "delta": "+" }                         // + advances TTP, - sets it back, 0 neutral/context
    ],
    "indicators": [                            // the numbers we track over time on this component
      { "metric": "Human-readable metric name", "unit": "kg",
        "value": 0, "as_of": "2026-08-01", "direction": "up_is_progress|down_is_progress",
        "canon_target": 0, "source": "https://…" }
    ]
  },

  "commentary": "The Lex-column paragraph. Dry, numerate, faintly appalled. 60–120 words.",
  "last_reviewed": "2026-08-29",
  "review_cadence_days": 30
}
```

### Hard rules for every record
1. **No claim without a citation.** `url` must resolve. If it cannot be verified, it does not ship.
2. **Separate canon from reality.** Never let a film's assertion leak into `real`.
3. **`spec_fraction` must be argued, not asserted.** State the comparison.
4. **Prefer primary sources** (papers, spec sheets, filings, benchmark tables) over journalism.
5. **Date everything.** An undated claim is not evidence.
6. **Mark uncertainty.** If the best source is a company press release, say so in `kind: "product"`
   and keep `trl` honest — a demo video is not TRL 8.

## 5. Categories (sub-indices)

| id | Name | Covers |
|----|------|--------|
| `chassis-and-materials` | Chassis & Materials | Endoskeleton, hyperalloy, structural survivability, mimetic polyalloy |
| `power` | Power | Energy cells, energy density, runtime, thermal management |
| `locomotion` | Locomotion | Bipedal walking/running, terrain, balance, durability of gait |
| `manipulation` | Manipulation | Hands, dexterity, grip force, fine motor, tool use |
| `perception` | Perception | Vision, multispectral sensing, HUD/target acquisition, SLAM |
| `cognition` | Cognition | Neural net processors, on-board learning, planning, embodied reasoning |
| `language-and-deception` | Language & Deception | NLU/NLG, voice mimicry, social passing, human impersonation |
| `autonomy-and-command` | Autonomy & Command | Mission autonomy, lethal decision-making, doctrine, human-out-of-loop |
| `networking-and-c2` | Networking & C2 | Skynet-scale distributed control, datacenter/compute, resilience |
| `weapons` | Weapons | Targeting, phased plasma, armed autonomous platforms |
| `biology-and-camouflage` | Biology & Camouflage | Living tissue over metal, cultured skin, biohybrid, odour/thermal signature |
| `self-repair-and-durability` | Self-repair & Durability | Damage tolerance, self-healing materials, field repair |
| `manufacture` | Manufacture | Building them at scale, automated factories, supply chain |
| `time-displacement` | Time Displacement | CTCs, the physics of sending one back. The joke that is also real physics. |

## 6. Files

| File | Contents |
|------|----------|
| `data/components.json` | `{ "components": [ …component records… ] }` — the source of truth |
| `data/canon.json` | Canon technology inventory + source concordance (agent B output) |
| `data/index-history.jsonl` | One line per computed index reading: `{date, ti, by_category, note}` |
| `data/movers.json` | Organisation registry, deduped across components |
| `research/**` | Raw agent research, markdown. Never read by the site. Kept for provenance. |