Beneficial and Harmful Effects

A navigation app is built to save drivers time, and for most users it does exactly that. But the same rerouting that helps one driver can push hundreds of others onto a quiet residential street that was never meant to carry through-traffic. The benefit and the harm come from the same mechanism: the app optimizes for the individual asking for directions, not for the street it sends them down. That is how one effect splits into a benefit for some and a harm for others — the software does one thing, but that one thing lands differently depending on where you stand.

Scale changes the picture. When a thousand people use the app, a few extra cars on a side street barely register. When a billion people use it, the same routing logic can reshape traffic patterns across entire cities, overwhelm streets with no capacity for it, and create harms no single user could see from inside the app. Effects that are invisible or trivial at small scale can become the dominant story at large scale, even though the underlying code never changed.

None of this makes the app badly designed. A design flaw means the software fails at the job it was built to do — crashes, gives wrong directions, drains the battery. An unintended effect is different: the software works perfectly and still causes harm nobody planned for. This distinction connects to the rest of the unit, where the same pattern shows up in who gets access to a network, whose data trains a model, and who a security system protects or excludes. The code can be flawless and the impact still uneven.