Ranking methodology v1.0
The Top Robots Index is computed from the entity graph by an open, deterministic formula — the same code that builds this site. No editorial thumb on the scale, no paid placement, ever.
The four pillars
| Pillar | Weight | What it measures |
|---|---|---|
| Deployment scale | 30% | Commercial status tier (commercial 1.0 · limited 0.6 · announced 0.3 · R&D 0.15 · discontinued 0.05) plus a bonus of 0.08 per tracked deployment / launch / expansion / production event, capped at +0.3. The pillar is then capped at 1.0, so the event bonus only moves robots below commercial status. |
| Specs | 25% | Confidence-weighted spec coverage: each manufacturer-verified spec counts 1.0, press-reported 0.75, unverified 0.4 — summed and saturating at 10 specs. Rewards documented, sourced capability over marketing claims. |
| Commercial traction | 25% | Maker scale (public company 0.85 · subsidiary of a major 0.7 · else log₁₀ of disclosed funding ÷ 10, capped 0.9 · 0.3 when neither public/subsidiary status nor a funding total is disclosed) + 0.15 if the robot has a genuine public price + 0.05 if funding events are tracked. Capped at 1.0. |
| Recency | 20% | 1 − (months since the newest tracked event ÷ 36), floored at 0. A robot with no news for 3 years scores zero here. |
overall = 100 × (0.30·deployment + 0.25·specs + 0.25·traction + 0.20·recency)
Data-integrity rules
- Every spec carries a source URL, a retrieval date and a confidence rating (verified = manufacturer-published, reported = reputable press/analyst). Unsourced specs cannot enter the graph — the build fails.
- Prices are only shown when genuinely public, with a "last checked" date. We never fabricate prices, ratings or review scores.
- History is append-only: status and spec changes become dated events, never silent edits.
- Affiliate relationships never affect scores — the formula has no editorial or commercial input.
Known limitations (v1.0)
- Deployment scale uses status tiers and event counts, not audited unit volumes — makers rarely publish those. As disclosed shipment data improves we will weight actual units.
- Specs measures documented coverage, not head-to-head performance; cross-sector spec comparability is inherently limited.
- Because the deployment pillar is capped at 1.0 and commercial status alone already scores 1.0, deployment and launch events do not currently differentiate robots that are already commercial — the bonus only separates robots below that tier. A future version will weight disclosed unit volumes so the commercial tier spreads out.
- The graph currently tracks 50 robots (352 specs, 243 citations). Coverage grows continuously; scores are comparable within a quarter.
Versioning
Methodology changes are versioned (this is v1.0, effective Q3 2026) and noted here. Quarterly editions are archived so citations remain stable.