This document discusses open identifier infrastructure and describes the ARK (Archival Resource Key) scheme and N2T (Name-to-Thing) resolver. It notes that while URL infrastructure is open, persistent identifier infrastructure is mostly closed. It then outlines how ARKs break with tradition by having a decentralized resolution that allows metadata and relationships without conflict. The document concludes by describing how N2T provides centralized but id-agnostic resolution for ARKs and other identifiers while following open principles for scholarly infrastructures.

1. Names, Things, and Open Identifier
Infrastructure: N2T and ARKs
John Kunze
California Digital Library

2. 2
Open identifier infrastructure?
• URL infrastructure is mostly open
• But “persistent identifier” infrastructure is mostly
closed

3. 3
Non-traditional id persistence
Traditional silos: doi.org, handle.net, purl.org, urn.arpa
2001: ARK (Archival Resource Key) scheme published
• Decentralized resolution shouldn't hurt, ie,
http://soviet.org/ark:/12345/9876
should be the same as
http://russia.org/ark:/12345/9876
• Embryonic and dynamic content
• Provider intention should be queryable

4. 4
Persistence nonsense
URLs aren't long-term, instead use ... (which are URLs)
URLs are locations, instead use ... (which are URLs)
URL hostnames break, instead use ... (which are URLs)
It’s all about hosts – which of these do we trust?
• doi.org, purl.org, handle.net, n2t.net
• bnf.fr, bl.uk, loc.gov, nla.gov.au, nlm.nih.gov

5. 5
ARKs breaking with tradition
• Syntax does not confer persistence
• ARK registry grows monthly, now 360+ insititutions
• ARKs appear along with DOIs in Thomson Reuters
Data Citation Index
• ARK as a collection of ideas for ids

6. 6
ARKs breaking with tradition
• Simple relationships shouldn't hurt
• A/B means A logically contains B
• A.B and A.C means A has two variant forms
• Machine-readable metadata shouldn't hurt
• Méthode ancienne – label: value
• "id?" requests metadata
• "id??" requests more metadata
• no conflict with “content negotiation”
• “Inflections” reduce need to create more ids

7. 7
Breaking with centralized resolver tradition
2007: N2T (Name-to-Thing) resolver introduced
• Centralized resolution for ARKs, but also for any
other identifier type
• Proposed consortium (on hold since 2008)
• Plus suffix pass-through (hierarchy shouldn't hurt)

8. 8
N2T under the hood
• Built using open source components
• Replica running in Edinburgh
• Primary running in California
• Supports ARKs, DOIs, URNs
• Ids from EZID, Internet Archive, and YAMZ.net
• Experimental load of 60 million CrossRef DOIs

9. 9
Summary – ARK and N2T
• Inflections (metadata shouldn't hurt)
• Structure with / and . (relations shouldn't hurt)
• Hostnames inert (decentralization shouldn't hurt)
• Id-agnostic resolution (centralization shouldn't hurt)
• Suffix pass-through (hierarchy shouldn’t hurt)
• Let’s build to the “Principles for open scholarly
infrastructures” (Bilder, Lin, & Neylon)