Hands Only One Researcher Knows: A Bus-Factor Audit for Bench Methods
A lab can hold a complete written protocol and still have a method that only one person can run. The document may list volumes, temperatures, and timings, yet the procedure works reliably only when a particular researcher performs it. The difference is not usually secrecy. It is the accumulation of small adjustments, tool familiarity, and recovery habits that never made it into the text. For students and collaborators, that gap becomes a reproducibility problem the moment the original operator is unavailable.
The Problem With a Complete-Looking Protocol
Bench methods often fail the transfer test in quiet ways. A custom lysis step may require a specific pipetting rhythm that the protocol describes as “mix gently.” A microscope alignment may depend on a sequence of knob movements that the manual omits because the manufacturer assumes a trained service technician. A cell-strain maintenance routine may include a visual check that one researcher performs without thinking and therefore never records. None of these are exotic failures. They are ordinary forms of tacit knowledge, and they concentrate risk in a single pair of hands.
This is a management issue as much as a documentation issue. When a method has a bus factor of one, the lab is exposed to vacation schedules, turnover, illness, and the ordinary friction of onboarding. The written record looks adequate, which makes the dependency harder to spot. A short audit can make it visible.
A Three-Part Audit for Method Custody
Begin with a candidate list. Ask each lab member to name three procedures they believe they could not hand to a new person without standing beside them for a full run. Combine the lists and remove duplicates. The resulting set is usually smaller than expected and more revealing than a general protocol inventory.
Score each candidate on three dimensions using a simple 1-to-3 scale:
- Tacit cues. How much of the method depends on unstated sensory or timing judgments, such as color changes, bubble patterns, or the feel of a resistance point?
- Tooling dependency. How much does success depend on a specific instrument, adapter, software setting, or bench arrangement that is not described in the protocol?
- Recovery difficulty. If the run begins to fail, how much does recovery depend on the original operator's diagnostic instincts rather than a written decision rule?
A score of 7 or higher marks a method for immediate documentation. A score of 5 or 6 warrants a scheduled review. The scoring is not a precise measurement; it is a way to start a conversation that labs rarely have without a prompt.
Recorded Walkthroughs Without Performance
The first artifact is a recorded walkthrough. The goal is not a polished training video. It is a continuous, narrated run of the method as it is actually performed, with the camera close enough to capture hand placement, tool changes, and intermediate checks. The operator should narrate decisions aloud: why a sample is moved to the back of the rack, why a particular pipette tip is preferred, what a normal color or texture looks like at each stage.
Recording has two practical advantages. It captures details that a written protocol tends to smooth over, and it creates a reviewable object that a second person can pause and question. Keep the recording in the lab's shared repository, label it with the protocol version and date, and pair it with the written method rather than replacing it. A recording without an owner and a review date becomes stale quickly.
The Naive Trial Run
The second artifact is a second-person trial run. Choose an operator who has not performed the method, provide the written protocol and the recording, and ask them to run the procedure without coaching. The original operator observes but does not intervene unless safety requires it. The observer's job is to note every point where the new operator hesitates, asks a question, or makes an adjustment that the protocol does not mention.
This trial is more informative than a demonstration because it exposes the difference between knowing a method and having learned it. Hesitations are data. A pause before a centrifugation step may reveal an unstated balancing convention. A question about waste disposal may reveal a missing safety note. The output of the trial is a short list of amendments, not a judgment about the new operator's competence.
A One-Page Failure Checklist
The third artifact is a one-page failure checklist for the method. It should list the symptoms an operator is likely to see when something goes wrong, the most probable causes, the first corrective action to try, and the conditions under which the run should be stopped and escalated. The checklist should be written for someone who has run the method a handful of times, not for the original expert.
A useful checklist stays short. If it grows beyond one page, it is probably trying to document the entire method rather than its failure modes. Keep the language concrete: “No band after amplification” is more useful than “suboptimal reaction conditions.” Include the name of the person or team to contact when the checklist runs out of options, and review the page whenever the protocol changes.
Making the Audit Routine
The audit works best when it is scheduled rather than prompted by a crisis. A quarterly review of two or three high-scoring methods is enough for most academic labs. Assign an owner to each artifact and record the review date in the same place the protocol lives. When a lab member prepares to leave, the audit doubles as a handover tool: the recording, the trial-run notes, and the failure checklist give the incoming operator a starting point that a binder alone cannot provide.
Some labs will resist the process because the original operator is faster alone. That is usually true in the short term and misleading in the long term. The purpose of the audit is not to slow down expert work. It is to reduce the time it takes for a second person to become independent, and to make that independence visible before it is urgently needed. A method that only one researcher can run is not yet a lab method. It is a personal practice waiting for a second operator to discover its edges.
