Provenance Is the Product

Every tool below is unglamorous, and each one costs you something small at the moment you use it. What you buy is the ability to answer one question later: where did this come from?

The bottleneck in practitioner research was never the writeup. It was provenance — which study said what, under what conditions, with what sample, and whether you read it or only read about it. That work is invisible while you are doing it and expensive when you find out you skipped it. The five tools here all operate on that problem.

Zotero

A reference manager, open source, run by a nonprofit. The browser connector captures citation metadata and, where the publisher allows, the file itself, at the moment you find the thing rather than three weeks later when you are trying to reconstruct where you saw it.

What it prevents is the half-remembered citation. Anyone who has written something long has had the experience of knowing a claim is real and being unable to find its source again. Zotero makes capture cost about two seconds, and two seconds is the only price cheap enough to become a habit. The catch is that its metadata is only as good as what the publisher exposes. Book chapters and grey literature arrive wrong often enough that you should check every entry you intend to cite. I check all of mine, and I still find errors.

Connected Papers

Discovery through citation structure instead of keywords. You give it one paper and it builds a graph of work related by citation pattern rather than by sharing your search terms.

This addresses a problem most people do not know they have: your search vocabulary is a product of the literature you already know. When a field names your concept differently — and across education, HCI, and organizational studies it almost always does — keyword search cannot show you what you are missing, because you cannot search for a word you have never encountered. Citation structure has no such limit. It is a starting move rather than a survey, though. The graph will hand you an old, heavily cited paper with the same confidence it hands you a current one, and it has no opinion about which one the field has moved past.

Taguette

Qualitative coding, free and open source. Import documents, highlight passages, apply codes, export the coded excerpts.

The obvious failure it prevents is coding in a word processor with colored highlights and then losing the ability to answer "show me every passage I marked as this." The less obvious one is the site licence. The established packages are good, and they are priced for institutions, which means the practitioner doing research on their own time either does without or does the work worse. Taguette is deliberately minimal — no inter-rater reliability statistics, no automated coding, thin visualization. For a small interview study that is enough. For a team project with reliability requirements it is not, and you should know which one you are running before you start.

The Open Science Framework

A project registry with preregistration. You write down what you plan to do and when, and it is timestamped.

What this protects against is not fraud. It is the ordinary, nearly invisible drift where you find something interesting in the data and, by the time you write it up, sincerely remember having expected it. Preregistration does not forbid exploratory work. It keeps the distinction between what you predicted and what you discovered legible — to a reader, and more importantly to you. The honest caveat is that it costs real effort up front, and much practitioner research is exploratory by design, where the payoff is smaller. Use it when you have a hypothesis. Skip it when you are describing something.

Zenodo

Permanent deposit with a DOI, operated at CERN. Upload a dataset, an instrument, a poster, a preprint, and receive an identifier that will still resolve in ten years.

This one prevents the dead link. Your conference slides on a university server that gets restructured. Your survey instrument in a shared drive you lose access to the week you change jobs. Nobody builds on work they cannot retrieve, and the standard practitioner archive — a personal cloud folder — has a half-life of roughly one employer. Deposit is permanent by design, which is worth understanding before your first upload: you cannot quietly remove a version you later regret. New versions supersede old ones, and the old one stays visible. That is the feature, not a defect, but it does mean the first upload deserves a moment's thought.

What they have in common

Every one of these makes something slower at the point of capture in exchange for making something recoverable later. Zotero spends two seconds now to save an hour in eight months. Preregistration spends an afternoon to defend a finding against your own memory. Deposit spends ten minutes to keep work retrievable after you leave the institution that hosted it.

None of that is visible in the finished product, which is exactly what makes it easy to skip and expensive to have skipped. A reader who trusts a piece of research is not really trusting the conclusion. They are trusting that the path back to the evidence is still intact and still walkable. Provenance is not the paperwork surrounding the work. In practitioner research it is a substantial part of the work itself.