[adamjohnwright]

Replaces #93 after its review. Belongs to spec 001 (specs/001-biopax-from-graph/): the one difference T016 and T020 recorded without looking into it, the form of an author's name. Documentation only; no code changes.

What was measured

All 15 species, from a private copy of release 98's graph, against release 98's published files. Each PublicationXref was matched to its graph publication (by PubMed id, else ISBN, else title), and its authors compared in order, Person by Person. Names were read only to classify them, and none is recorded.

• One kind, all 872 values: the published value is surname, firstname and the graph's is surname, initial, for the same Person, whose own firstname and initial hold each form. There are no other kinds.
• 62 people, 10 publications, repeated across the species' files: 60 + 67 + 20 + 53 + 7×60 + 4×63 = 872.
• Why: the two writers use different rules.
  • The release's converter wrote the relational _displayName, and the curator tool's generatePersonName puts the first name first.
  • The graph importer rebuilds every Person's display name as surname, initial (ReactomeBatchImporter, "to fix different styles in Person display name"), and has done so since its first commit in 2016.
  • The graph's uniform initials are the importer's doing. The published files show that, in release 98's relational database, only these 62 people held the first-name form.
• Which form is meant: neither tool is at fault, and the exporter writes the graph's display name, as it should. Which form a Person's display name should have is a question for graph-importer's and the curator tool's owners (T022: drafted, not opened).

From #93's review

• The cause corrected: the importer normalises Person names; the data did not change between two readings. Line 108, the section and this description now say so.
• tasks.md: T016's line now has the 4 PublicationXrefs and the cause, and T022 notes the drafted question.
• The July–August detail reduced to "most were created in release 98's curation".
• The 23 → 4 correction keeps a note of where the 23 came from (T016, T016: literature, evidence and curation, as the release writes them #69, before T017 and T018), as the review explained.

Checks

• Prettier passes on specs/.
• Both changed files were searched for every surname and first name involved: none found.

🤖 Generated with Claude Code

[adamjohnwright]

Review: ready to merge once the fixture check is green, with one wording fix advised first.

#93's review is resolved:

• The cause: the importer's rebuild is now the cause, in the section, line 108, T016's line and the PR body. I checked graph-importer's history: ReactomeBatchImporter.java has carried the Person rule ("Different Styles in Person display Name") since its first commit, 5a0cb591 of 30 Jun 2016, so "whichever commit built release 98's graph" holds.
• 23 → 4: corrected in both files, with where the 23 came from.
• T022: notes the drafted question without being ticked.
• The July–August detail: reduced.
• Arithmetic: 60 + 67 + 20 + 53 + 7×60 + 4×63 = 872.

Medium, advised: "so in release 98's relational database these 62 alone held the curator tool's first-name form" says more than the measurement shows. The published files show only the Persons who author literature in them, not every Person in the relational database. Suggest: "so, of the authors the published files name, these 62 alone held the curator tool's first-name form in release 98's relational database".

Low:

• "all but one, a script's, whose display name is its own" sits oddly beside "the importer rebuilds every Person's display name". If the importer's rule applies to every Person, the exception needs its reason, for example a Person with no surname or no _displayName, or it should be stated as observed and unexplained.

Public-repo care: no names, hosts or credentials. The remaining detail ("most of the 10 were created in release 98's curation") is fine.

🤖 Generated with Claude Code