Photo by Enkhjin photography on Unsplash

At my local supermarket, I watched a cashier scan the same yogurt five times. Not five different yogurts — the same one, passed over the glass again and again, beep after beep, while the register clearly had a button to type a quantity once and be done.

I filed it where I usually file these things: bad software. Jason Fried is right that the world is full of applications with miserable UX, and here was another one. A system so poorly designed it forced a person to repeat an action a machine could resolve in a single keystroke. Provably slower. Obviously worse. Case closed.

So I asked the store manager to explain the clumsy interface. He did not reach for a policy or a statistic. Roughly paraphrasing, what he said was: we do not want to make the cashiers’ life too easy.

There is no fraud in this story. Nobody is stealing. The shortcut simply makes the cashier’s job easier while making the store’s data worse — an ordinary, well-meaning slide toward the path of least resistance. And the quantity button, the faster path I was mourning, is exactly what opens that slide.

Picture a promotion: five yogurts of any flavor for a discount. The thorough way to ring it up is to scan each flavor’s barcode. The quick way is to scan one flavor and hit “×5.” The customer pays the same either way, so who would notice?

Inventory cares. With the shortcut, the store’s records say it sold five strawberry and zero of the other four flavors. The total is right; the per-flavor breakdown is fiction — and the breakdown is what feeds the next restock order and the flavor-level demand forecast. One tired cashier taking the quick route corrupts the numbers those decisions run on. And notice which number moves. Add the button and checkout gets faster — something a store almost certainly measures. The bad data sits on no dashboard at all. The visible number improves while the invisible one rots.

Here is the part that took me a second conversation to get right, because my first instinct was wrong.

Removing the quantity button does not make the shortcut impossible. A cashier can still swipe one barcode five times. What removing the button takes away is the payoff for the shortcut.

With the button: scan once, type five. Measurably faster than scanning five separate items, especially when a barcode is crumpled or awkwardly placed. The wasteful move is also the fast one.

Without it: swiping one item five times and scanning five separate barcodes take roughly the same effort. The time the shortcut used to save mostly evaporates, and with it the reason to bother. The corner can still be cut — it just stops being worth cutting, so mostly it isn’t.

That is the whole trick, and it is worth stating precisely: a good constraint rarely makes the wrong move impossible. It takes away what made the wrong move worth choosing.

Which is a real trade-off, and the manager had chosen a side. Speed at the register pulls against accuracy in the records, and he had decided the records win. The friction at the register is not a flaw in the system — it is the price of clean data, paid by the one person in the whole chain you happen to be looking at.

The friction is not perfect, and it is more honest to admit where it leaks. When one specific barcode refuses to scan, re-swiping an easier item is still tempting, and some bad data slips through there. But that is an occasional leak, not a systematic one. The quantity button was a universal shortcut, available on every transaction. The bad-barcode trick is a rare exception. Trading the first for the second is a good deal.

Engineers have a name for the family this belongs to. Toyota calls it poka-yoke — mistake-proofing — shaping the process so the right action is the easy one and the wrong one takes deliberate effort. A purist will object that classic poka-yoke physically blocks the mistake, while this design only removes the reason to make it. Close enough cousins. Either way, the clumsy checkout is mistake-proofing wearing the costume of bad UX.

The checkout did not invent this, and its designer at least knew the reason. The next two cases are stranger: the reason is real, but the people keeping the practice alive have forgotten it.

The Bedouin wear black robes in the desert. Black — the color that soaks up heat — in one of the hottest places on earth. It looks like a mistake basic enough to correct on sight. When researchers actually measured it (Nature, 1980), the black robe was no hotter at the skin than a white one. It absorbs more heat, but the loose fit turns the gap between cloth and body into a chimney, and convection carries the heat away before it reaches the skin. Generations of desert living kept an answer that looks like a mistake and is not.

Cassava feeds hundreds of millions across Africa and South America, and in its bitter form it carries enough cyanide to cripple or kill. The traditional preparation is a laborious multi-step sequence — grating, repeated washing, separating, drying and cooking — spread over days, for a root you could, in principle, just boil. In The Secret of Our Success, the anthropologist Joseph Henrich describes Indigenous communities in the Amazon who follow the sequence faithfully, though often no one can tell you which step removes the poison, or that removing poison is the point. The tedium is the safety mechanism: cut the slow steps and the cyanide stays. Where the processing has broken down — during droughts and food shortages, worsened by hungry, protein-poor diets — people have developed konzo, an irreversible paralysis. The cases cluster where the slow steps got cut.

None of these people were optimizing for elegance, and unlike the store manager, most could not tell you what their inefficiency buys. That is the harder case: a practice whose reason has outlived everyone’s memory of it. Which is Chesterton’s Fence — do not tear down a fence until you know why someone built it. The reformer who sees only the obstruction, and not the thing it holds back, is the dangerous one.

So this is the lesson I keep relearning. The reflex to remove an inefficiency is usually right and occasionally catastrophic. Before you optimize the friction away, find out what it is paying for. Some inefficiency is waste. Some is load-bearing. The skill — in product, in process, in culture — is telling them apart, and the tell is a single question asked before you reach for the fix: why is this here?

The cashier was not fighting a stupid system. The system was fighting something I could not see.