Afrimintel · Decision Views · Claims Register · Archive · Methodology · Audit Log
If Afrimintel informed a figure, a decision, or a paper, cite it — like any source. Citation is attribution; it's welcome regardless of licence. What it is not is a redistribution grant (see the licence note below).
Platform:
Afrimintel (2026). Afrimintel: Africa mining intelligence. Data feed v1.1.0, dataset 2026-10-18. Nikesh Patel, ed. https://afrimintel.com (accessed YYYY-MM-DD).
A specific feed or figure (name the feed and the dated record):
Afrimintel (2026). "[figure/record]," [feed name] (api/v1/[feed].json), dataset 2026-10-18. https://afrimintel.com/api/v1/[feed].json (accessed YYYY-MM-DD).
The CC-BY open tier (spatial-reference and research-grade deposit records, plus the jurisdiction-level tenure and traceability registers) is deposited with a permanent DOI so it can be cited without depending on this site staying up. The DOI is the stable form; the feeds below are the live form and move with the platform.
DOI (all versions): 10.5281/zenodo.22664745 — cite this one. It always resolves to the newest version, so a citation does not go stale when the deposit is revised.
DOI (v1.1, current version): 10.5281/zenodo.22724886 — cite this one only if you need the exact bytes you read.
Citation: Patel, N. (2026). African Mining Governance Infrastructure and Open Deposit Register (v1.1) [Data set]. Afrimintel Ltd. https://doi.org/10.5281/zenodo.22664745
First deposited 9 September 2026 under CC BY 4.0 and indexed in OpenAIRE; v1.1 published 12 September 2026, correcting seven errors in v1.0 (errata below). The deposit carries the jurisdiction-level tenure and traceability registers (40 jurisdictions each) and the open-tier deposit records with screening-grade coordinates (83). Those rows carry collection-run provenance — a dated harvest of a named registry set — which is weaker than the per-cell Sourced contract used elsewhere on this platform, and the deposit says so on every row.
Corrected on 12 September 2026. The errors below were found on 11-12 September 2026 in the published open-tier deposit (10.5281/zenodo.22664746, version 1.0) and are fixed in 10.5281/zenodo.22724886 (v1.1), which the concept DOI now resolves to. A published DOI version is immutable by design, so v1.0 still carries every error below and always will; anyone holding a citation to the v1.0 version DOI reads the uncorrected file. That is why the table stays here rather than being deleted: an errata that disappears once fixed is a worse record than one that says what was wrong, when it was found and when it was corrected. Each entry gives what v1.0 says, what it should say, and the build on this platform where it was corrected; the same list is the opening paragraph of the v1.1 description. Published within the seven-business-day window in the Quality Standard.
| Field | Deposit v1.0 says | v1.1 reads | Corrected on the platform |
|---|---|---|---|
| Tenure coverage block | Labelled 25 as the count of jurisdictions with a public register. | 28 have a public register; 25 is the count not recorded restricted. Both keys are now defined in the file. | v2.137.203 |
| Description: registers stating no reuse licence | Stated seventeen. | 18. The v1.0 data file itself already carried 18; the error was in the description, not the data. | v2.137.209 |
| Morocco, tenure licence | Read “n/a”, which in this file means no register exists. | “licence not recorded at collection”. Morocco publishes a register; its terms were not captured. | v2.137.204 |
| Zimbabwe, tenure note | Carried a trade-press figure of roughly 60,000 claims, which the file’s own stated limits say was not adopted. | No count. The register states none and none is asserted. | v2.137.204 |
| Niger, tenure note | Carried “~18-20 exploitation permits”, an approximation read from a screen. | No count. The permit table is countable and an exact figure may replace this on a later collection run. | v2.137.208 |
| Traceability, fifteen absence reasons | Gave the reason that the jurisdiction’s principal minerals fall outside the commodity classes these regimes cover — which cannot be right for gold producers when certified gold is one of the three classes checked. | A checked absence with no reason asserted (15 rows now carry a reason of any kind). | v2.137.208 |
| Rwanda, partnership cell | Characterised the EU MoU as “effectively suspended, not terminated” — a verdict, in a file whose own statement says partnership statuses are dated milestones and never verdicts. | The dated milestones only: MoU signed 19 Feb 2024, European Parliament resolution 13 Feb 2025, EU Council placed it under review 17 Mar 2025. | v2.137.208 |
| Traceability provenance | Described as collection-run on every row. | Two grades: 16 collection-run and 24 registry-index. This is a platform regrade after the deposit, not an error in it. | v2.137.202 |
| Per-row read dates | The description said these would be published as v1.1. | Still open, and deferred to a later version: the corrections above were not held up for collection work. Stated rather than dropped. | — |
The three deposited data files are otherwise unchanged in substance: 40 jurisdictions in each register file, 83 deposit records with screening-grade coordinates, and the counts of public registers, portal-only access, restricted registers and scheme participation are as published. README.md and zenodo-metadata.json are carried forward from v1.0 unchanged and still describe the record as it stood on 8 September; they are scheduled for the next version. Every item above is on the append-only log with its build.
Not yet in a deposited version. Six of the 83 deposit records in the deposited file carry a position, status or identity that is wrong. They are in v1.0 (10.5281/zenodo.22664746) and v1.1 (10.5281/zenodo.22724886) alike: v1.1 corrected the jurisdiction files and left the deposit records as published. Both versions are immutable, so anyone holding either version DOI reads these rows as first published. The live feed (/api/v1/spatial-reference.json) carries the corrections from build v2.137.281, and the table below is generated from it. Kolomela is placed at a coordinate filed on SAHRIS; Black Mountain and Pering are placed at the position of the named town, on the location statements cited in the row, which is why their envelopes are wider. This is disclosure on the platform; a corrected deposit is a new version, which has not yet been uploaded. Positions remain screening-grade, with the envelope stated on each row.
| Row | Deposited file (v1.0, v1.1) says | Live feed reads | Corrected on the platform |
|---|---|---|---|
| Kolomela | -27.7, 22.9 (status Op) | -28.383, 22.987, ±5 km (status Op); sources on the row: SAHRIS: Butler (Banzai Environmental), Palaeontological Field Assessment for the proposed upgrade of the Kolomela Mining Operations, issued 13 Sep 2019 (page footers 31 Oct 2019), Fig. 7 'Kolomela mining pit' 28deg22min56.81S 22deg59min11.55E. Locality phrases placing the mine west/south-west of Postmasburg: the same report ('approximately 15 km on the R309 west of the town') and the Mining Technology project page ('22 km from Postmasburg', no direction); the phrases are distances from the town, not from the pit point | v2.137.281 |
| Black Mountain Zn (SA) | -29.5, 19.8 (status Op) | -29.2, 18.82, ±5 km (status Op); sources on the row: Black Mountain Mining Social and Labour Plan 2024-2029, s1.1 and s1.3: the Deeps/Swartberg operation 'is situated immediately west of the town of Aggeneys'; Aggeneys '113 km northeast of Springbok'. The plan's cover letter to the DMRE (13 Dec 2024) places Swartberg/Deeps east of the town and Gamsberg west, contradicting its own body; the body agrees with the town gazetteer ('just to the west of the town'). Coordinate: Aggeneys town geocode 29deg12minS 18deg51minE (Wikipedia), offset ~3 km west - a town-derived point, not a mine coordinate | v2.137.281 |
| Pering Zn (SA) | -25.7, 23.4 (status Op) | -27.567, 24.2, ±15 km (status Hist); sources on the row: Location statements: BHP Billiton SENS 25 Jun 2002 (sharedata.co.za copy; closure of Pering, tied to Reivilo) and Miningmx 21 Oct 2009 ('about 140 km north of Kimberley, near Reivilo'). Coordinate: Reivilo town geocode 27deg34minS 24deg12minE (Wikipedia) - a town position, not a mine coordinate; no published coordinate for the mine was found | v2.137.281 |
| Songwe REE | -10, 34 (status Dev) | Retired: name and position match no single project. The name points at Songwe Hill (Mkango Resources, Phalombe district, south-east Malawi), about 640 km to the south; the position lies nearer Tanzania's Songwe Region (Ngualla). Both have their own records, which stay. | v2.137.281 |
| Mkango REE (Malawi) | -10, 35.2 (status Dev) | Retired: as Songwe REE. The name points at Songwe Hill, about 620 km to the south, or possibly Kangankunde; both have their own records, which stay. | v2.137.281 |
| Mbamba Bay | -0.5, 14.5 (status Exp) | Withdrawn: Mbamba Bay is a town on the Tanzanian shore of Lake Nyasa; the row placed a copper-lead-zinc deposit in the Republic of Congo, and no deposit of that name was found. | v2.137.281 |
Found by the adversarial passes on v2.137.277 and corrected within the seven-business-day window in the Quality Standard; each change is on the append-only log.
build_version; asset-data vintage carries a separate dataset_version and dataset_date (the two can differ — a deploy that changes no asset data does not advance the dataset vintage). Cite the dataset_date for reproducibility; the audit log records every change with its version. Current: feed v1.1.0 · dataset 2026-10-18.The api/v1/ feed paths are intended to be stable; when a feed's schema changes materially it is versioned, and supersessions are recorded on the audit log rather than silently overwriting a path.
Structured, sourced, stable-URL data is also what AI assistants ground on. Being the cited source is the point — but not at the cost of the licence line above.