publishedresearchresearch
What must a SaSame research dataset disclose?
Dataset pages make measurement inputs understandable without exposing credentials or private customer data.
- Collection
- research
- Updated
- 2026-07-29
- min read
- 2
- Version
- v1
Verified summary
A dataset discloses provenance, collection window, identity keys, fields, exclusions, license and privacy boundaries.
- Observed
- 2026-07-23 14:08 UTC
- Claim strength
- measured
- Dataset scope
- The Registry-Runtime Gap study's discovery dataset: 31,407 endpoint records, 4,162 enumerable tool surfaces, and 78,041 declared tools, collected over the 2026-06-18 to 2026-07-23 window.
- Methodology
- Datasets are versioned and cited by DOI. The dataset's own numerical-reproduction check (
scripts/research/verify-paper-00.mjs) is re-run rather than assumed, and any mismatch is recorded on the dataset page instead of silently corrected. - Limitations
- The published discovered-endpoint-record count (31,407) does not currently reproduce exactly — an independent re-run returns 31,408. This is disclosed on the dataset's own evidence-status record, not hidden.
Measured — backed by a specific dated observation or dataset
01
Purpose
Dataset pages make measurement inputs understandable without exposing credentials or private customer data.
02
Method
Datasets are versioned and related to papers, methodology, observations and corrections.
discovered_endpoint_records : 31,407 (published) -> 31,408 (reproduced) [open, off-by-one]
repeated_observations : 152,781 -> matches
enumerable_tool_surfaces : 4,162 -> matches
declared_tools : 78,041 -> matches
overall_annotation_coverage : 78.2% -> matches03
Boundary and limitations
Dataset size is not the same as unique live products, owners, customers or market demand.
EX
Examples
1- 01
A registry-derived endpoint list is separated from the subset returning a valid runtime response.
RF