Rules and guidance for re-publishers of data

[timgdavies]

Linked to the discussion in #130 and exploration of the technical requirements for an OCID registration tool, we have been exploring whether any special fields are required to indicate that a publisher is 're-publishing' open contracting data.

E.g. when a publisher is not the original source of information, what should they do to indicate this.

This might apply to platforms like OpenOpps.com which are aggregating from many sources.

[timgdavies]

See also #325 which handles related issues for cases of systems republishing OCDS from multiple sources.

[jpmckinney]

1. I recommend that the publisher be part of each release in Add publisher field (release schema) #325, to make it clear who published the release. Releases should be treated as immutable. Re-publishers can add their own metadata in the package – though, what is a use case that relies on information about the re-publisher?
2. The relevant comment from External project identifier #130 is External project identifier #130 (comment). However, I think the authority that registered the OCID prefix should be the only publisher authorized to either mint new OCIDs or delegate such responsibility. Others should not be minting new OCIDs for which they have no authority. If delegated, all minters should agree on a methodology for what suffix to use Multiple Organizations publishing data about the same process (OCID prefix) #461 (comment)
3. Further to (2), we should clarify that OCID prefixes establish an identifier series, for which there is one publisher with the authority to mint OCIDs (and to allow others to do so). Anyone can publish a release using an existing OCID. Having publisher on the release as in (1) would allow users to exclude publishers they don't trust.

This can be written up on a new guidance page, and also integrated into the OCID prefixes page.

[yolile]

Observatorio Fiscal (a CSO) from Chile is adding more information to the existing Chile Compra OCDS publication with data from the Ministry of Works. For processes that already exist in Chile Compra, they are adding new releases (planning and contract implementation in this case), using the same ocid as Chile Compra, and using the release/publisher object to identify each release's publisher. For contracting processes that are not in Chile Compra, they are publishing their information using a new ocid and prefix created for them.

[duncandewhurst]

Some content related to OCID assignment from #1646 (comment):

The publishers would need to agree the process and responsibilities for minting OCIDs. For example: who can mint OCIDs; if an OCID cannot be calculated deterministically from the input data, then how to lookup whether an OCID for a given process already exists, and how to share the new OCID for others to find. They would also need to implement the agreed approach, whether this means reporting a new OCID to a central registry, or requesting the OCID for a given procedure from a shared "OCID issuer" service, etc.