Quantum computing projects have an odd rhythm. Weeks of reading and circuit design lead to a burst of runs on a simulator or a cloud device, and then everybody scatters to write up whatever happened. Between runs there is rarely time to ask the team how the work itself is going. What follows is a working checklist for pausing and asking. It is a start-stop-continue retrospective tailored to teams working on quantum experiments, with audit questions you can tick off before you write your three lists.
Before you begin: what to gather
You do not need much. Collect these items and keep them on one page or shared folder.
- The list of experiments or runs in the period you are reviewing, with a one-line purpose for each.
- The notes or notebooks where results were recorded.
- Any messages or tickets where the team got stuck or changed direction.
- A note of which tools, simulators or devices were used, without needing to list every setting.
If a fact is missing, write “unknown” rather than guessing. An honest gap is a useful finding in its own right.
Audit checklist, part one: the experiments
Work through each question as a group and tick it, flag it or leave a note. These questions focus on how the experiments were run rather than on the physics.
- Did every run have a written purpose before it started?
- Can someone outside the team reproduce a given run from the notes alone?
- Were the circuits or parameters recorded in the same place as the results?
- Did the team compare noisy and ideal outputs consistently, or only when results looked odd?
- Were failed or inconclusive runs written up with the same care as successful ones?
- Were the same mistakes made twice?
Items one to five are practice checks. Item six is the most revealing. Repeated mistakes mean that a lesson was learned but not recorded, and the fix is almost always a line on the start list.
Audit checklist, part two: the team
- Did anyone wait more than a day for a reviewer or an answer?
- Was the knowledge of one tool or device concentrated in a single person?
- Did the team agree on what counted as a meaningful result before looking at the data?
- Were assumptions written down, or did they live in people’s heads?
- Did the team have a quiet moment to think, or was every day reactive?
The point of both checklists is not to score yourselves. A cluster of ticks suggests a habit worth continuing. A cluster of flags suggests a start or a stop. Notes capture the stories that explain why.
Turning the audit into start-stop-continue lists
Read through the ticks, flags and notes, then sort them. The mapping is usually natural.
- Start usually comes from a missing practice: a purpose line for each run, a shared template for recording parameters, a short weekly note on what surprised the team.
- Stop usually comes from a habit that wastes effort: re-running experiments from memory instead of the notes, comparing outputs informally, keeping results in private notebooks.
- Continue usually comes from a ticked box that the team is proud of: careful write-ups of failed runs, a regular reading session, a habit of pairing on tricky circuits.
Limit each list to two or three items. If you have nine starts, you have a backlog, not a plan. Put the extras aside and look at them next time.
An invented example, to make the checklist concrete
The team below is made up for illustration. A small group had spent a month exploring a hybrid approach to an optimisation problem. During the audit they ticked purpose lines, flagged reproducibility and found that two of the runs could not be repeated because a parameter had been changed in a notebook without a note. Their lists: start a shared run template that records parameters and purpose before the run, stop editing notebooks in place without versioning them, continue the habit of writing up inconclusive runs.
Three lines. The next month, a newcomer reproduced an earlier run in an afternoon, something that would have taken days before. That is the kind of result a start-stop-continue retrospective is designed to deliver: small, concrete and checkable.
Reading that sharpens your lists
If your audit flagged questions about how machine learning and quantum computing relate, this piece on how quantum computing changes machine learning offers framing that may influence what you put on the start list. If the team’s recent experiments dealt with optimisation, a look at quantum algorithms applied to tough optimisation tasks can help you judge whether your stop list should include approaches that sound promising but have not delivered.
A note on borrowing structure from elsewhere
Quantum teams often find inspiration in unexpected places. BestPilotsNow, an aviation discovery site, sorts its pages into areas for pilots, flight schools, courses, resources, forum discussion and blog posts. The takeaway for an experiment team is not about flying at all. It is that sorting material into clear, labelled areas makes it easier to find. Give your run notes, your audit results and your open questions separate labelled homes in the team’s shared folder.
Scoring the audit lightly
If you like numbers, give each audit question a simple mark: green for “we did this reliably,” amber for “sometimes,” red for “rarely or never.” Resist adding up a total. A single red on reproducibility matters more than five greens elsewhere, and a sum would hide that. Instead, look at the reds first and ask what one start or stop would turn each toward amber. Then look at the greens and ask which ones are fragile, meaning they depend on one person remembering. Fragile greens belong on the continue list, with a note about how to make them less fragile.
Keep the marks in the same folder as the lists, so that next time you can see whether the colours moved. Movement from red to amber on two questions in a month is a perfectly good result. Quantum work is slow and uncertain by nature, and the team’s own habits are one of the few things it fully controls.
Checklist for the session itself
- Set a timer: forty-five minutes for a small team.
- Silent writing first, discussion second.
- Describe habits, not people.
- End with owners and review dates.
- Put the lists next to the experiment log.
- Open the next session by reading these lists aloud.
What to do when the lists feel obvious
Sometimes the team looks at its lists and shrugs: of course we should record parameters. That is fine. A habit that is obvious yet not happening is exactly what the retrospective is for. If the lists feel too easy, ask a different question: which of these obvious things have we said before and still not done? Items that recur across sessions deserve a stronger approach, such as building the template, assigning an owner or making the habit part of the review for every run.
Common pitfalls
- Auditing the physics instead of the practice. A retrospective is about how you work, not whether a result is correct.
- Skipping the continue list. Good habits fade when no one names them.
- Collecting evidence for weeks. A thin audit that happens beats a thorough one that does not.
Wrapping up
Reviewing a month of quantum experiments takes no special software. You need an hour, a short audit of how runs and teamwork went, and the willingness to turn what you find into three plain lists. Tick, flag, sort, trim, assign and review. Repeat after the next stretch of runs, and you will find that the team’s habits get steadily better at supporting the science it is trying to do.