1. Overview
Language and environment for statistical computing, with scripts and data in plain text or documented serialisation formats.
Category: Scientific Computing & GIS.
2. Version and scope assessed
Version assessed: Pinned evidence c267ebb7f9720154535e63642c25a5d91d5f512d.
3. Overall determination
Computing determination
4. Principle profile
| Order | Principle | Finding |
|---|---|---|
| I | Sovereignty | Not assessed |
| II | Longevity | Not assessed |
| III | Locality | ✓ Meets |
| IV | Transparency and Control | Not assessed |
5. Fundamental requirements
| Requirement | Status |
|---|---|
| E1 Independent access to essential information | ✓ Pass |
| E2 Practicable exit and migration | ? Not Verified |
| E3 Independent interpretation and recovery | ? Not Verified |
| E4 Continuing authority and custody over already-created work | ✓ Pass |
| E5 Governability of consequential actions | ? Not Verified |
6. Evidence and verification
- E1: R projects use source scripts and data files, while renv records package dependencies in a lockfile. (source), confidence high
- E2: The full readRDS documentation explains that infoRDS reports representation/reader metadata but cannot establish file validity, and explicitly cautions against treating bare RDS as a machine-interchange format. Native-endian serialization and newer compression introduce additional dependencies; cross-tool and CSV preservation remain separate questions. (source), confidence medium
- E3: The serializer writes distinct version-2/version-3 headers with writer/minimum-reader versions; the minimum values in the inspected source are R 2.3.0 and R 3.5.0 respectively, and version 3 also records native encoding. These format-level bounds do not guarantee every object/package class will reconstruct on an older R. (source), confidence medium
- E4: R projects can run locally and use local files and libraries. (source), confidence medium
- E5: Statistical results can inform consequential decisions, but data provenance, model validation, and review controls were not tested. (source), confidence low
- sovereignty: A positive Sovereignty finding requires established E1, E2 and E4 in the declared scope. The existing access/exit/custody record for R does not establish all three minimums. The prior positive finding is withdrawn; recorded storage and portability strengths remain facts, not a verified failure. (source), confidence medium
- longevity: E3 is not established for R's existing essential-record scope. The previous Longevity finding is replaced by Not assessed, not a verified failure or a positive placeholder. (source), confidence medium
- locality: R analysis can run locally and offline once dependencies are available. (source), confidence medium
- transparency: The positive control observations above do not yet establish whole-scope E5. The methodology requires that minimum before a positive Transparency and Control finding; the previous positive finding is therefore withdrawn to Not assessed, not changed to Does not meet. (source), confidence medium
- E3: Serialization headers specify representation/version/encoding bounds but do not recover every object/package class. No factor/date/attribute/NA-preserving independent reconstruction with a complete package/source/data environment has been recorded; bare RDS compatibility metadata is not file validity or whole-workflow recovery. (source), confidence medium
- E5: Model accuracy alone is not E5. The current evidence does not establish effective script/package execution authority, workspace/file overwrite decisions, inspectable transformation state or available correction of consequential data/model changes. (source), confidence medium
7. Migration and recovery
Native storage format: R scripts, CSV, and RDS serialised objects.
Synchronization or runtime model: Local installation with package repositories.
Example export or inspection command (not verified against any particular version or data set): Rscript -e 'str(readRDS("data.rds"))'
8. Architectural strengths
Established strengths and their scope are recorded in the evidence and assessment findings on this page.
9. Architectural limitations
- Source/retained-input inspection only; no installation or application restore performed.
- Factor/date/attribute/NA round trips, a non-R reader, older-release trials, package-library upgrades and workspace overwrite recovery remain untested.
10. Assessment findings and remaining work
Evidence update: 23 October 2026
- Resolved: The serializer writes distinct version-2/version-3 headers with writer/minimum-reader versions; the minimum values in the inspected source are R 2.3.0 and R 3.5.0 respectively, and version 3 also records native encoding. These format-level bounds do not guarantee every object/package class will reconstruct on an older R. Source.
- Resolved: The full readRDS documentation explains that infoRDS reports representation/reader metadata but cannot establish file validity, and explicitly cautions against treating bare RDS as a machine-interchange format. Native-endian serialization and newer compression introduce additional dependencies; cross-tool and CSV preservation remain separate questions. Source.
- Open: Factor/date/attribute/NA round trips, a non-R reader, older-release trials, package-library upgrades and workspace overwrite recovery remain untested. E5 needs script/package and workspace-write authority, visible transformation parameters/state and correction boundaries.
Assessment follow-up: These bounded findings are incorporated into the evidence above. Eligibility and principle ratings are deliberately unchanged; their reconciliation and any blank-status fixes are deferred to the next pass. Reproduction details and retained inputs are in evidence/third-pass/batch-10/.
Human-readable criteria: Can an independent party reproduce the analysis from preserved data, code, units, dependencies, and provenance, and exchange results without hidden loss?
Rating reconciliation: 24 October 2026
E3 and E5 remain Not Verified for the existing declared scope. Longevity and Transparency and Control are Not assessed pending their required minimums. Sovereignty is also Not assessed because the required access/exit/custody minimums are not all established. This is a targeted evidence decision, not a finding that the product fails. Existing E1, E4 and Locality findings were not re-certified.
Remaining: Factor/date/attribute/NA round trips, a non-R reader, older-release trials, package-library upgrades and workspace overwrite recovery remain untested. E5 needs script/package and workspace-write authority, visible transformation parameters/state and correction boundaries.
Next without installing locally: Use synthetic scripts/data/serialization and package-lock/source manifests with expected types, missingness and results; compare independent readers or externally reproduced outputs and same-version execution/overwrite controls.
The individual decision and provenance are retained in evidence/rating-reconciliation/batch-10/. Overall: Insufficient evidence.
Contribute verification evidence for R
Have you tested clean-machine restoration, format migration, or failure rollback for R? We welcome reproducible laboratory logs, synthetic test fixtures, and independent assessments to resolve open findings.
Read the evaluation protocol · Submit verification records on GitHub
Individually reconciled in rating batch 10 on 24 October 2026. E3/E5 remain Not Verified for explicit missing coverage; dependent unsupported findings corrected. Previously resolved questions are retained, not counted again. E1/E4/Locality were not re-certified by this targeted review. The assessment procedure is described in the methodology.