Find one named institutional user (the actual JOSS gate)

[shivamlalakiya]

Why this is the gating item

Everything else about this project is in reasonable shape: ~7,900 lines of source, ~8,000 of tests, 1,700 tests passing, 95%+ coverage, 43 documentation pages, four external contributors. None of that is what a JOSS editor screens on for a solo project.

The screen is research application and evidence of use. JOSS explicitly accommodates a single author, but only with "other evidence of community engagement, either in the repository or evidenced in the paper". Current signals: 6 stars, 5 forks, 0 watchers, 0 dependents, 0 Discussions, and no named user anywhere.

More concretely: there is no validation on real donor data anywhere in the repository. The one real dataset that ships (load_ciob_fundraising) has no donor rows, no gift amounts and no labels, and no estimator is ever fitted on it. Every number in the README, in docs/explanation/benchmarks.md and in paper.md is synthetic. A named institutional user is the cheapest thing that changes that, because it converts "plausible tool" into "tool someone runs".

What counts

Any one of these, in rough order of value:

• A nonprofit, university advancement office or AMC foundation willing to be named as using it, even for exploratory analysis.
• A researcher citing it in a paper or preprint.
• A public post, talk or notebook by someone who is not the maintainer.

Concrete next step already drafted

There is a "who's using this?" Discussion drafted and still unposted, and 0 Discussions exist. That is currently the only instrument by which a user could signal existence. Posting and pinning it costs minutes.

This needs a person to say yes, so it cannot be closed by a PR.

[shivamlalakiya]

Re-scoping note, not a close.

This issue frames the JOSS gate as "find one named institutional user". That is still the highest-value route and it is worth pursuing. It is not the only one, and reading the live JOSS text rather than its reputation turns up two cheaper doors.

submitting.html:

There must be evidence that the software is being used for research at minimum by the developers themselves, and ideally by others.

Reviewer checklist:

Research impact statement (required section): Does the paper include a "research impact statement" section providing evidence of realized impact (publications, external use, integrations) or credible near-term significance (benchmarks, reproducible materials, community-readiness signals)?

So research use by the maintainer is explicitly the floor, and an archived reproducible benchmark is a named alternative to publications. scripts/leakage_experiment.py (#101) is already half of that; its one disqualifying weakness is that it runs on synthetic data only.

That weakness is now #124, which replicates the leakage finding on KDD Cup 1998 and archives the result. It needs nobody's permission and it is weeks rather than a quarter.

Two caveats worth stating so nobody over-reads this:

• The checklist "or" is a reviewer's judgement call. Desk rejection happens earlier, against submitting.html, which is the stricter text. So Validate on real donor data: replicate the leakage experiment on KDD Cup 1998 (the tractable JOSS impact gate) #124 pairs with a preprint, tracked in JOSS submission: tracking issue for the four desk-reject gates (target late October 2026) #127.
• This issue stays open. A named institutional user is what makes every subsequent venue cheaper and is the only route to a citation that is not the maintainer's own. It is just no longer a blocker for submission.

Sequencing and the rest of the gates are in #127.