• observation: new people come into community, discuss new ideas, discussion dies down
  • but we also look at these new tech & discuss their applicability
• "phone" is identified by several IDs
• hashes talk about accession, not about for example location
• CIIDRA = IDs for construction industry
• "immutability" in programming isn't necessarily useful for persistence
• issues with DOI: What does it mean to "own" an identifier?
  • DOIs can be handed over to other organisation
  • ownership is multi-faceted: content, domain, infrastructure to keep it available, etc.
• same for "trust": intention vs. execution
  • "distributed": membership organisations, server locations, decision taking
• efforts of hacking around social issues are doomed
  • governance needs human interaction
• blockchain for PIDs is antisocial & elitist

• aligning incentive between policy creator & follower needed, like in Hamurabi's code

• data silos: maybe we know them, but like a city covered by fog, we can't easily look deeply => solution ideas: NCATS Data Translator Knowledge Map
• How many FTEs are spend reconciling IDs from different sources?
  • [ ] considered packaging & publishing scripts to avoid re-scripting?
• no qualifiers for relatedTo means merging opposing data becomes common
  • something similar to https://github.com/pandoc-scholar/pandoc-scholar#semantic-citations needed
  • instead: https://github.com/monarch-initiative/kboom
• http://journals.plos.org/plosbiology/article?id=10.1371/journal.pbio.2001414
• [ ] Why PIDs needed for licenses? creativecommons.org/publicdomain/zero/1.0 is URL & ID in one, no?
• http://reusabledata.org/

• difficult to calculate impact / ROI of big scientific machinery

• p hacking or "garden of forking paths" washes random cases of significances up as scientific findings
• remedy: pre-registering studies & methods
• allTrials in medicine, now social sciences
• discussion at BITSS: which PIDs to use?
• pre-registrars require rather many fields, compared to repositories
• still, people do fill these out => several hundred studies pre-registered each
• sounds like RIOjournal, but community-owned registrars have better reputation there => better to introduce PIDs there, than to introduce them to new journal

• conference acronyms not unique
• some conferences don't have publishers
• some are predatory / fake / mock

• audience consensus: little discussion necessary about whether blogs should have DOIs, concrete implementations needed
• manual DOI registration possible, but of little use
• API call better, but integration from blogging software necessary
• SEO also requires exposing metadata => interests align
• schema.org/BlogPosting suggested
• https://blog.datacite.org/ uses JSON-LD
• https://wordpress.org/plugins/schema/
• sitemap needed so that indexer can find the posts to then extract metadata
• beta testers from institutions that already register DOIs welcome

• building tech, syntax & metadata is comparatively easy
• How to create social context for acceptance of identifiers?
  • [ ] But why was that possible for several PIDs, but not for keeping URLs stable in the first place?

A link would be useful here, ideally to a persistent version of the policy.

These keywords should be linked to persistent identifiers (rule number 5).

Researcher - author

University ID

Grant number - use of this can link actions back to requirements and support monitoring / compliance checks

Research funder - link to funder ID

It might well have been better to get each of these documents their own unique identifier rather than bundling them all together under this DOI.