Waiting for experiment
Every strategy faces identical generated stores, making the comparison fair. Adjust any assumption and the experiment reruns.
Each round generates four visible queues. Pick one before the hidden cashier speeds, coupon snags, and actual service times are revealed. This is the single-trip version of the question: a smart choice improves your odds, but cannot guarantee a win.
This is a discrete-event simulation: each shopper has a randomly generated basket, each cashier processes one item at a time, and payment adds a fixed cost. It repeats the same decision hundreds or thousands of times to separate a good rule from a lucky guess.
Add all visible items ahead, then add one payment-and-bagging overhead for each shopper. Choose the smallest total. This beats simply counting people whenever baskets vary substantially. If baskets are hidden, the model automatically falls back to their expected size.
If every basket is drawn from the same distribution and you cannot see its size, each person represents the same expected amount of work. The shortest line is then the rational choice—though chance can still make it lose on any single trip.
A single line feeding the next free cashier pools risk. One slow basket delays everyone slightly instead of trapping one unlucky line. It generally reduces the long tail and makes waits fairer, even if its mean can resemble a well-chosen separate line.
Basket size uses an over-dispersed random distribution. “Unpredictability” controls its spread: low values cluster shoppers near the average; high values create more tiny baskets and occasional overflowing carts.
Service time equals items × seconds per item + fixed pay/bag overhead, with modest random noise. The cashier-speed toggle gives lanes persistent speed differences, reproducing the frustrating line that looks short because its cashier is slow.
All four rules are scored on the same generated queue in each trial. The bars show mean wait; the dashboard adds the 90th percentile, win rate, regret, and a 95% confidence interval for your selected rule.
Colored circles are shoppers; the number inside is basket size. Cash registers sit at the top, queues advance toward them, and the orange outlined shopper is “you.” Speed changes playback only, never the calculated outcome.
Pressure compares arriving work with total scanning capacity. Above 100%, shoppers arrive faster than the checkouts can serve them, so a continuing queue would grow without bound. Picking cleverly cannot fix an understaffed store.
Real lines include coupons, age checks, equipment failures, queue switching, express limits, and correlated family baskets. Use the result as a decision principle: estimate work when visible; otherwise minimize people; prefer pooled queues.
Typical store is a balanced baseline. Basket lottery creates extreme basket variation. Friday rush pushes incoming work above capacity. Express test opens a 12-item lane and shrinks baskets so you can measure when eligibility matters.
When enabled, lane 1 accepts shoppers with 12 items or fewer and is populated only by eligible baskets. A simulated shopper with a larger basket cannot choose it. This often makes express best for small trips, but a long express queue can still lose.
The snag controls add a configurable chance that each shopper triggers an extra random delay. Snags are not visible when a line is chosen, so they widen the wait distribution, increase tail risk, and demonstrate why the apparently optimal line sometimes loses.
The challenge creates a fresh four-lane store on every round. You see shopper counts, total items, and express status, then choose. It reveals hidden service times, the optimal line, time lost, and which simple heuristic would have made the same choice.
The 90th percentile describes a bad-but-common wait. Standard deviation measures volatility, tail risk reports waits above ten minutes, and regret measures extra time versus a clairvoyant choice. The 95% interval shows Monte Carlo precision, not real-world certainty.
The verdict names the lowest-mean strategy for the current controls. It also explains hidden baskets, express eligibility, and overload. Every slider reruns matched trials locally, so conclusions and uncertainty always describe the displayed setup.
Choose the smallest sum of item-scanning time plus one fixed transaction cost per shopper.
Choose the shortest line. Identically distributed random baskets give every shopper equal expected work.
A single shared line is most reliable because it automatically sends the next shopper to the first available cashier.