Working memory holds roughly three to five items at a time, and once you push past that ceiling, accuracy and mood start to slip together. That’s the wall engineers and clinicians run into every time a manager waves off a request for better tools with the phrase, “just check the numbers”.
The task sounds trivial when it isn’t. Checking numbers under pressure is a working-memory task dressed up as a clerical one, and stacking those tasks day after day is one of the cleaner predictors of burnout in professions that live on calculation.
The nurse titrating a drip at 3am
A night-shift nurse recalculating a weight-based infusion isn’t doing arithmetic. She’s holding the patient’s weight, the concentration on the bag, the ordered dose, the pump’s units, and the last set of vitals in her head at once, while a monitor alarms two beds over. Any one of those items alone is easy. Together they crowd out the safety check that should catch a decimal error.
The load isn’t the maths. It’s the switching. Every interruption resets the mental scratchpad, and the calculation starts over from a slightly worse position than the time before. By hour ten, the cost of “just double-check it” gets paid in fatigue, not seconds.
The process engineer sizing a pump between meetings
An engineer asked to confirm a pump selection between two status calls will pull up flow, head, specific gravity, and efficiency, then try to hold all four steady while a Slack thread lights up on the second monitor. The formula is well known. What turns a five-minute check into half an hour of low-grade anxiety is the context switching.
This is where automation earns its keep most, obviously. A vetted reference calculator, or a simple in-house sheet driven by a documented method for centrifugal pump sizing, takes the arithmetic out of working memory and puts it on the page. The engineer’s attention then goes to the judgement calls the formula can’t make: whether the duty point is realistic, whether the fluid data is current, whether the motor margin is honest.
The resident buried in the chart
Electronic health records were supposed to reduce mental load. In practice, they often pile more on. A narrative review in JMIR argues that poorly presented information and cluttered interfaces add extraneous load that hits early-career clinicians hardest, and that this load tracks with burnout.
The pattern is familiar to anyone who has watched a resident hunt for a creatinine trend across three tabs while trying to remember the antibiotic they were about to dose. A small clinical question can still carry a heavy retrieval cost.
Automation helps when the problem is shaped right
Not every calculation should be automated, and not every automation reduces load. Some just move it around. The ones that genuinely help share a few traits:
- Repetitive with stable inputs: Motor sizing, dose-per-kilogram, unit conversions, tank-volume checks. If the shape of the problem is the same every time, a tool should carry it.
- High cost of a small error: Anywhere a misplaced decimal or a wrong unit turns into a hazard, a patient safety event, or a warranty claim, the calculation belongs on paper the machine can read, not in someone’s head at the end of a shift.
- Frequently interrupted: Work that lives on a ward, a shop floor, or a jobsite gets interrupted by design. Anything that has to survive interruption without restarting from scratch is a candidate for a tool.
- Auditable: A spreadsheet with named inputs and a visible formula beats a mental estimate you can’t reconstruct later. Engineering programs at universities like Missouri teach this discipline early: build the sheet once, change the inputs, let the answer update.
Automation backfires when it hides the work
A calculator that hides its assumptions is worse than doing the maths by hand. So is a dashboard that presents ten numbers when the decision needs two. Both raise extraneous load rather than lower it, and both chip away at the practitioner’s sense that the tool is on their side.
The fix isn’t more software. It’s better-designed software, with fewer fields, clearer units, and a visible trail from input to output. That’s the argument a recent review makes about AI and automation in clinical work: the tools that reduce burnout are the ones that cut extraneous load, not the ones that add another screen to check.
Managers miss the real cost of “just check the numbers”
The phrase treats the calculation as free, and it isn’t. Every check pulls from the same limited pool of attention the professional needs for the harder judgment coming next: is this patient stable, is this pump the right choice, is this the right time to escalate.
Automating routine math is a way to spend expertise where it matters. Give the engineer and the clinician tools that carry the arithmetic, and the working memory they save gets reinvested in the parts of the job that actually require a human.
