Most people never notice the invisible layer of trust in every meal — the assumption that the salad was washed by a system they will never see.

In a single week, the CDC was hunting the source of a multi-state Cyclospora outbreak causing what officials bluntly called explosive diarrhea, the FDA raised a potato-chip recall to its highest risk tier, and a cruise ship docked in San Francisco with norovirus aboard. American food safety had a bad seven days.

Cyclospora is a parasite spread through fresh produce, and it has a specific talent: it shrugs off the chlorine washes that kill ordinary bacteria, which is why it is a summer regular tied again and again to imported greens and herbs. The recalls and the outbreak are the visible tip of a system built to find contamination by counting the people it has already made sick.

The case for the system is that this is surveillance working. The outbreak was detected, the recall was escalated, the public was warned.

The case against is that detection after the fact is not prevention. An infrastructure that identifies contamination by tallying the ill is structurally reactive, and every cut to public-health capacity makes it slower to notice, slower to trace, slower to warn.

Foodborne illness does not spread its harm evenly. It falls hardest on the immunocompromised, the very young, and the old — the people for whom explosive is not a punchline but a hospitalization.

The salad is safe until it is not, and the only thing standing between those two states is a surveillance system the country underfunds until the week it fails. Building health-data infrastructure that prevents rather than counts is the reform we keep arguing for at Popular Rationalism.

American Life is independent analysis.