The four things worth recording on every run — and why machines can finally capture them consistently.
Capture the source, process, content, and output of every run: where the inputs came from, exactly how the work was done (protocol and software/hardware versions, parameters, timestamps), what actually happened (images, sensor readings, errors, human interventions), and the results. Automation's real advantage here is consistency — a robot records all of it the same way every time, which is the foundation of reproducible, reusable (FAIR) data.
Poor record-keeping is a major driver of the reproducibility problem — only 6 of 53 landmark preclinical cancer studies could be independently reproduced [1]. You can't reproduce what wasn't recorded. The 2016 FAIR principles — data should be Findable, Accessible, Interoperable, and Reusable — set the modern standard, with explicit emphasis on data that machines can find and reuse, not just humans [2].
| Layer | What it captures |
|---|---|
| Source | Where inputs came from — reagent lots, sample IDs, provenance |
| Process | How it was done — protocol version, software/hardware versions, parameters, timestamps |
| Content | What happened during the run — images, sensor traces, errors, interventions |
| Output | The results, linked back to the exact run that produced them |
A tired scientist at 6pm doesn't note that the protocol was version 3, the reader firmware updated last week, well B4 was re-pipetted, and the room was two degrees warm. A robot logs all of it automatically and identically every time — version stamps, timestamps, images, exceptions, and interventions — turning a notebook of intentions into a record of what actually occurred.
Humans record what they meant to do. Machines can record what they actually did.
Consistent machine capture is what makes data Findable, Accessible, Interoperable, and Reusable without a clean-up project later [2]. On the Undergrad, every step is logged automatically — what was done, when, and how — so reproducibility is a property of the run, not a chore after it.
