The direct answer
What changed after RCI Data Brief 001? Between immutable catalog v0.4.0 and v0.4.5, the number of products and manufacturers stayed fixed at 75 and 25. The catalog added 35 source-linked claim IDs across 19 products, modified 43 existing claim records across 11 products and removed zero claim IDs. The union touched 25 exact products.
Did evidence quality improve? In a bounded sense. Independently verified claims increased from one to six because five exact-model experiment or characterization records were added. Conflicts increased from three to six and unknowns from fourteen to eighteen because later audits exposed incompatible source values and missing comparison fields instead of hiding them. More C and U records are not deterioration when they replace false certainty.
Did publication growth produce search discovery? Not yet at a level RCI can generalize. At the pre-publication cutoff, Google had confirmed only the homepage indexed, while Bing had accepted submissions without an indexing conclusion.
Catalog v0.4.0 / Data Brief 001 through catalog v0.4.5 and site v0.5.2, before publication of RCI 023. Publishing this page creates RCI 023 and is not counted inside its own pre-publication operating snapshot.
How the delta was calculated
RCI compared the complete claim arrays in the immutable v0.4.0 catalog and v0.4.5 catalog by stable claim ID. An ID present only in v0.4.5 is added; an ID present in both but with different serialized claim metadata is modified; an ID present only in v0.4.0 is removed.
A new claim ID is a new source-linked evidence field, not a new product, shipment, market event or independent validation. A modified record means serialized claim metadata changed; it does not by itself prove that the earlier value was wrong or that hardware changed.
- Added: a new field-level record with its own evidence state and source trail—not necessarily a new product or a newly published manufacturer specification.
- Modified: at least one value, condition, note, evidence state, source object, verification date or related claim field changed.
- Removed: a stable claim ID disappeared from the later catalog. The observed count is zero.
Five releases added depth in different ways
| Release step | Added | Modified | Removed | Evidence delta M / V / D / C / U | Editorial focus |
|---|---|---|---|---|---|
| v0.4.0 → v0.4.1 | +5 | 0 | 0 | +0 / +5 / +0 / +0 / +0 | Five exact-model independent evidence records across actuation, perception, compute and power. |
| v0.4.1 → v0.4.2 | +12 | 36 | 0 | +11 / +0 / +0 / +1 / +0 | LiDAR revision identities, condition-preserving range fields and one additional current-source conflict. |
| v0.4.2 → v0.4.3 | +2 | 0 | 0 | +0 / +0 / +0 / +0 / +2 | Two explicit continuous-torque unknowns where only peak, stall or other non-equivalent boundaries were public. |
| v0.4.3 → v0.4.4 | +14 | 1 | 0 | +10 / +0 / +0 / +1 / +3 | Actuator backlash, accuracy and related motion-error fields, including one preserved same-page mass conflict. |
| v0.4.4 → v0.4.5 | +2 | 6 | 0 | +2 / +0 / +0 / +1 / -1 | Depth-camera lens and error-metric boundaries, including one conflict promotion and one unknown resolved to manufacturer evidence. |
The 36 modified records in v0.4.2 primarily reflect LiDAR revision and condition normalization across Ouster, Hesai and Livox records. “Modified” therefore cannot be translated into “36 corrections.” The record-level JSON and CSV expose the changed field names so readers can audit each case against both frozen catalogs.
The evidence mix moved, but manufacturer claims still dominate
| State | v0.4.0 | v0.4.5 | Delta | Interpretation |
|---|---|---|---|---|
| M · Manufacturer claimed | 525 | 548 | +23 | Twenty-five new M IDs were partly offset by three M → C transitions and one U → M transition. |
| V · Independently verified | 1 | 6 | +5 | Five bounded exact-model records were added; they do not verify whole products. |
| D · RCI derived | 1 | 1 | +0 | The single explicit RCI derivation was unchanged. |
| C · Conflicting | 3 | 6 | +3 | Three prior manufacturer records were promoted to conflicts when current official surfaces disagreed. |
| U · Unknown | 14 | 18 | +4 | Five unknown IDs were added and one prior unknown gained manufacturer evidence. |
The four evidence-state transitions were deliberately asymmetric: Orbbec Gemini 335 range, CubeMars AK70-10 mass and Hesai QT128 mass moved from M to C; the ZED X Mini 2.2 mm accuracy field moved from U to M after a lens-specific official error definition was added. No claim moved from M directly to V.
Non-joint systems received almost half the new claim IDs
| Hardware layer | Components | Claims v0.4.0 | Claims v0.4.5 | Delta | Current V / C / U |
|---|---|---|---|---|---|
| Motion & actuation | 18 | 130 | 147 | +17 | 1 / 1 / 8 |
| Perception & sensing | 21 | 146 | 162 | +16 | 3 / 3 / 1 |
| Compute & control | 17 | 123 | 124 | +1 | 1 / 1 / 4 |
| Power & energy | 11 | 79 | 80 | +1 | 1 / 0 / 2 |
| Interfaces & integration | 8 | 66 | 66 | +0 | 0 / 1 / 3 |
Perception and sensing added sixteen claim IDs and motion and actuation added seventeen. Main compute and power each added one independent record. Interfaces added no new IDs in this window. This is depth work on a fixed 75-product cohort, not evidence that one hardware layer has a larger market or higher engineering importance.
No stable claim IDs were silently removed
The delta contains 78 record-level change rows: 35 added and 43 modified. Every row names the exact product, field, evidence before and after, changed metadata fields, current source URL and current verification date. There are no removed rows.
Zero removals is a revision-control result, not a promise that RCI will never retire a record. A future removal must appear in a versioned delta with a reason; correcting a bad URL, separating a revision or promoting a conflict does not justify making the historical claim disappear.
Prepared outreach is not completed validation
At the cutoff, RCI had prepared 16 product-level fact-check packets covering 13 responsibility parties and fixed a first wave of 10. Sent packets: 0. Received responses: 0.
Zero responses is not a response-rate result because outbound fact-checking had not started. Public editor identity, a brand sender and an archive destination remain send gates; personal account details are not a substitute.
Publication and indexing are separate funnels
| Layer | Observed result | What it proves | What it does not prove |
|---|---|---|---|
| Public corpus | 22 publications and 149 sitemap URLs before RCI 023 | Pages and files were public, crawlable and enumerated. | Discovery, indexing, ranking or citation. |
| Google Search | Homepage confirmed indexed; eight research requests accepted | At least one URL was in Google and selected requests entered its queue. | That every sitemap URL or requested page was indexed. |
| Bing / IndexNow | Five research requests accepted; latest IndexNow batch HTTP 200 | The submission mechanisms accepted the URLs. | Discovery, crawl completion or Bing indexing. |
The operating implication is simple: more pages are not the next bottleneck. RCI must now earn external verification, independent results and third-party references while preserving stable URLs and continuing index observation.
Download and reproduce Data Brief 002
Download the immutable catalog release delta JSON and record-level CSV. Stable aliases are current JSON and current CSV. Reproduce the comparison from the frozen v0.4.0 and v0.4.5 catalogs.
Suggested citation: Robot Component Index. “Robot Component Catalog Delta: 35 Added Claims, 5 New Independent Records, 0 Removed IDs.” RCI 023 / Data Brief 002, research dataset v0.1.0, 2026-08-10. https://robotcomponentindex.com/research/catalog-release-delta-data-brief/
For the evidence-state definitions, use the methodology. For material publication changes, use the change log. For version-aware BibTeX and machine citation records, use Cite RCI.