Enabling Agricultural Traceability through Data Interoperability

There is some overlap between the proposed artefacts. For testable and interoperable specifications, three things should suffice: (1) a suite of JSON schema (2) with terms drawn from a JSON-LD @context which maps to (3) existing well established web vocabularies (or new ones created where there are gaps)

very much agree that any selection of external standards must include confidence that the standards are open, transparent, well governed, and durable.

I'd suggest that it should be preferred (but not mandated) that external standards are freely accessible (ie no paywall). I'm aware that this would impact ISO and SAI standards - hence the suggestion that this should not be a mandate because some ISO / SAI standards will be essential.

Agree that the best way to do this is via JSON-LD @context mappings. But whilst it's important to leverage (ie map to) existing an ubiquitous vocabularies wherever possible (eg schema.org), I would advise caution about ensuring that the mapping is accurate. It is better to define your own precise new vocabulary item than to link to an existing one that is "similar but not the same".

Always better to re-use well tested and adopted standards. But please make sure that the standards are (1) open / accessible and also (2) aligned with EPCIS traceability events and (3) published in a normative form (ie schema and or JSON-LD/RDF vocabulary)

As a tool to map exact or similar meanings across multiple vocabularies, an ontolgy can be valuable. Especially in the current state of play with sustainability criteria where there is a bewildering array of terminology used in various standards and regulations (SBTI, IFRS, ESRS, to name a few). However this can be a costly exercise and it may be worth collaborating with international organisations to maintain such mappings globally rather than only by Australia for Australia. In general, AU should use ontologies to map to its own regulations and standards rather than map between international regulations and standards.

I would suggest to add the United Nations Transparency Protocol (UNTP) - https://uncefact.github.io/spec-untp/. It is still under development but will be in final draft state by end July 2024 and will provide a sound foundation for Australian Agriculture traceability standards. By developing Australian standards as a UNTP extension, we will better support export consignments because the core traceability information will be interoperable across industries and across borders.

There is no problem with multiple location identifiers. Or multiple product identifiers. There will always be many the requirement for any identifier scheme to be part of the framework should be that it be "DRV" - ie "Discoverable, Resolvable, and Verifiable". Discoverable (D) so that the identifier can be easily found within an information set such as via barcode scan or standard data element. A resolvable (R) identifier is one which has a predictable way to construct a URL that points to further information about the subject of the identifier (using ISO-19875 for example). A verifiable (V) identifier is one where the owner of the identifier (eg the farm with the GLN or PIC) can prove that they are the owner of the identifier - eg via W3C DID identity binding.

Yes - explicit requirement that any identifier scheme must be discoverable, resolvable, and verifiable. THis is fundamental to successful and confident traceability.

The DID core specification is fine and it provides a standard data model for any DID. However the W3C leaves the development of specific DID methods to the market. The result has been an explosion of blockchain related DID schemes that might best be described as "me too cryptocurrency ponzi schemes". Accordingly any inclusion of DIDs into the framework should be specific about DID method. We recommend only did:web at this point in time.

Generally mandates of standards are bad. However given the fundamental importance of identifiers (of products, locations, and entities) in the traceability framework (as "signposts" to verifiable data) - the mandate should be limited to registry operators (eg GS1 GTINs and GLNs, Livestock NLIS ID, state PICs, even ABN in the ABR) and should require that all identifiers in the framework are discoverable, resolvable, and verifiable.

Note that there is also a potential governance relationship to international standards. For example, UNTP will define an extensions methodology that allows specific jurisdictions and/or industry sectors to extend UNTP to meet their specific requirements whilst maintaining cross-border and cross-industry interoperability. Accordingly the governance framework could show a link to any international standards governance group upon which Australian Agriculture traceability standards are built.

UN

Big question
- On Commercial challenges, my experience is that solution vendors in the traceability space have shifted their attitude over the last two years from "everyone has to use my system" to "if I don't make my system interoperable then I wont have a business". Therefore proprietary system lock-in risk is reducing. There should be a simple requirement for any technology system which is to implement testable compliance with the interoperability framework. In that case it wont matter which platform or service a given actor chooses - so long as it is interoperable. many of the other commercial challenges are solved if this one is solved.
- A commercial challenge that is not listed but should be - is the (lack of) financial incentive for suppliers to implement traceability & improve ESG performance. Why should I do this if I'm not getting paid more for my products? The way to fix this is to ensure that the buyer is incentivised to demonstrate an improving trajectory on their sustainability disclosures. If the market (or regulatory non-compliance) will punish brands that don't improve their ESG performance then those brands will seek out suppliers that can improve their performance- especially for scope 3 emissions.
- Another incentive that is being offered in some economies is preferential access to trade finance when ESG criteria are met. These incentives are often available through regional development banks such as EBRD or ADB.

On leadership challenges, there must be national leadership from government on standards. This deosnt mean that the standards must be mandated, only that it is obvious to the implementer community which standards are the ones with the weight and significance and durability and open governance - to justify investment in implementation. More than that, there needs to be some kind of "kick-start" to trigger a flood of implementers. Once a few significant actors implement a specific standard, others follow. Government is in a good position to kick start this implementation via grant finding that is accessible only to the first "x" (eg 10) implementers.

On coordination challenges, the best architecture is one that minimises the need for coordination. each actor should be able to (and get value from) implementing the framework without needing to collaborate with upstream and downstream actors. The UNTP has this principle at heart. Another key principle is that the framework should not dictate which information must be shared and which is private. It is my experience that different industry sectors and also different actors within a sector can have very different attitudes to the question of balancing confidentiality and transparency. Therefore the framework must allow each actor to choose their own balance. let the market pull drive increased transparency over time.

As discussed above - commercial incentives and international interoperability.

I put No because I agree 95% but there is one role missing and there was no way to comment if I said yes. Anyhow, the missing role is registry operators. These are very fundamental in the system (PICs, NLIS, ABNs, GTINs, etc). They have a responsibility to make the identifiers they register discoverable, resolvable and verifiable (see earlier answer to understand what this means).