EP.007 — 2026-07-13
Trace Upstream · Case file EP.007
ONE ORDINARY BAG OF SALAD 0 CASES · 31 STATES · SAME SOURCE SURVEILLANCE OFF · SINCE 2025 IN A CALM YEAR, WATCHING FEELS LIKE PAYING PEOPLE TO STARE AT NOTHING. THIS IS THE WEEK THE CALM ENDS.
TRACE UPSTREAM
3,000 sick,
no source.

Somewhere out there is a bag of salad that looks completely ordinary. And right now, food like it has made more than three thousand people, across thirty-one states, sick — and it seems to trace back to a single source.

Here is the strange part. The people whose entire job is to find that source can't do it — because a year ago we quietly switched off the system built to look. In a calm year, watching for outbreaks feels like paying people to stare at nothing. This is the week the calm ends.

01
The Event
The event · July 2026

Three thousand cases. No source.

3,000+CASES SINCE MAY 31–32STATES 86HOSPITALIZED · NO DEATHS 31×MICHIGAN vs A NORMAL YEAR SOURCE, WEEKS IN: STILL UNKNOWN — no farm, no supplier, no product named CYCLOSPORA — A PARASITE THAT RIDES IN ON FRESH PRODUCE (RASPBERRIES · BAGGED SALAD · CILANTRO · BASIL) SUMMER 2025 — THE SURVEILLANCE NETWORK WAS SCALED BACK THE OLD LIST (ACTIVELY TRACKED): EXPERTS WARNED IT WOULD MAKE OUTBREAKS HARDER TO CATCH. A YEAR LATER — HERE'S ONE.
Fig. 1 — the outbreak, and the sensor turned off a year before it · sources: CDC, NPR, NBC, Forbes, TIME

The illness is cyclosporiasis — a parasite that travels on fresh produce, with nausea and diarrhea that can last weeks. Since May: 3,000+ cases across 31–32 states, 86 hospitalized, no deaths. Michigan alone has 1,500+ in a state that normally sees about 50 a year — roughly 31× normal. Weeks in, no farm, supplier, or product has been named.

Here's what makes it a system story. In summer 2025 the CDC scaled back its foodborne-surveillance network; tracking this parasite became optional, leaving only salmonella and E. coli watched everywhere. Experts warned it then. A year later, here's an outbreak that's hard to catch.

02
The Pivot
The pivot · Not "who's slow"

We keep switching off the watchers.

"THE AGENCIES ARE JUST SLOW." PRODUCE OUTBREAKS ARRIVE EVERY SUMMER — THIS IS ROUTINE, NOT A FREAK EVENT. SO WHY DO WE KEEP SWITCHING OFF THE SYSTEMS THAT WATCH FOR RARE DISASTERS — UNTIL THE DISASTER?
Fig. 2 — a recurring event, met with the sensor already off

It would be easy to blame slow agencies and leave it there. But produce outbreaks arrive every summer — this is routine, and investigators have done it many times. What's different this time is that the tool they'd normally use to trace the source was switched off before the outbreak began. Which points at a pattern this channel keeps finding under the news: we keep switching off the systems that watch for rare disasters, right up until the disaster arrives.

03
The Design Flaw
The design flaw · Surveillance is a sensor

Without it, one outbreak looks like 3,000 accidents.

MICHIGAN1 case TEXAS1 case SURVEILLANCE LINKS THEMsame fingerprint → ONE source SENSOR OFF → 3,000 CONNECTED CASES LOOK LIKE 3,000 SEPARATE ACCIDENTS WHY THE SENSOR IS ALWAYS CUT FIRST: WHEN IT WORKS, OUTBREAKS STAY SMALL — NOTHING DRAMATIC HAPPENS. SO IT LOOKS LIKE PAYING TO WATCH FOR NOTHING. YOU CAN'T SEE A DISASTER THAT NEVER HAPPENED. ONE BATCHone processor IN 30 STATES WITHIN DAYS — THE FOOD OUTRUNS THE INVESTIGATION SAME TRAP AS A BUILDING CODE (EP.001): PAY NOW, THE REWARD IS A RARE DISASTER THAT QUIETLY NEVER ARRIVES.
Fig. 3 — the sensor that turns scattered cases into one source · and why it's first to be cut

Surveillance isn't paperwork — it's a sensor. One case in Michigan and one in Texas are just two unrelated bad days; surveillance is what links them by genetic fingerprint into one source. Switch it off, and 3,000 connected cases look like 3,000 separate accidents, and the pattern pointing at one farm never forms. And it's always cut first because its value is invisible: when it works, outbreaks stay small, so it looks like paying to watch for nothing — you can't see a disaster that never happened.

On top of that, one contaminated batch from a big processor is in 30 states within days — the food outruns the trace-back. It's the same trap as a building code: pay now, and the reward is a rare disaster that quietly never arrives, so it gets shaved down until the ground moves.

04
What If
What if · Design the blindness out

Three ways to stop being blind.

FIX A · EUROPE — ONE ALERT, EVERY BORDER, BY LAW A FLAGGED PRODUCT EVERY COUNTRY, IN HOURS+ trace one step back / forward, by law TRADE-OFF: EXPENSIVE, BUREAUCRATIC, NEEDS CENTRAL COORDINATION A 50-STATE PATCHWORK RESISTS FIX B · PUT THE COST ON THE SELLER (LESS GOVERNMENT) BUYERS + INSURERS DEMAND IT"digitize your chain, or you're out" SUPPLIERS TRACE, FAST TRADE-OFF: ONLY WHERE A BIG BUYER EXISTS · CAN ROT INTO BOX-TICKING · CRUSHES SMALL FARMS FIX C · SHORTEN THE CHAIN — SHRINK THE BLAST RADIUS GROWN + EATEN IN ONE REGIONsource obvious within a day vs a processor shipping to 30 states TRADE-OFF: LESS EFFICIENT, COSTLIER, CAN'T GROW EVERYTHING EVERYWHERE ALL YEAR EUROPE'S SPEED · A BUYER'S LEVERAGE · A SHORT CHAIN'S SMALL FOOTPRINT EACH ONE DESIGNS THE BLINDNESS OUT — INSTEAD OF HOPING SOMEONE SPOTS THE DANGER IN TIME. NONE OF THEM NEEDS BETTER PEOPLE.
Fig. 4 — three architectures, three trade-offs · dashed green = revision markup

Three very different systems answer this, none of them just "spend more on the same monitoring." Europe runs one rapid-alert network — a flag in one country reaches every border in hours — with a law requiring every business to know one step back and forward; trace-backs take a day, not weeks. The cost: expensive, bureaucratic, hard to coordinate across fifty states. Or use less government: make the seller liable and let big buyers and insurers demand traceability — faster than any regulator, but only where a powerful buyer exists, and it can crush small farms. Or shorten the chain so a local outbreak stays local and the source is obvious — at the cost of efficiency.

The close · What are we not watching?

A prevented crisis leaves nothing to point at.

3,000SICK — CHEAP TO SWITCH OFF NOW · RUINOUS TO GO WITHOUT LATER EVERY QUIET SAFEGUARD RUNS INTO THE SAME TRAP — · DISEASE SURVEILLANCE · BRIDGE INSPECTIONS · AIRCRAFT MAINTENANCE · THE EMERGENCY STOCKPILE NOBODY HAS TOUCHED IN YEARS WHAT IS NOBODY WATCHING RIGHT NOW — BECAUSE, FOR THE MOMENT, NOTHING BAD IS HAPPENING?

Those three thousand people aren't sick because anyone is stupid. They're sick because turning the watching off is easy and cheap today, and living without it is ruinously expensive the moment an outbreak slips through — a cost that only shows up later. And it's bigger than one salad: disease surveillance, bridge inspections, aircraft maintenance, the emergency stockpile — every safeguard that works quietly looks like waste, because a prevented crisis leaves nothing to point at. So the calmest moment is when the watching is easiest to cut. The real question isn't who was too slow to find the lettuce. It's about you: what is nobody watching right now, because for the moment nothing bad is happening?

TRACE UPSTREAM

Not who's to blame — how it's built. The full interactive blueprint, with the parts that didn't fit the video, lives on this page.

Watch on YouTube Subscribe
?
3,000 sick
no source
OFF
We turned
it off