Bacon Is Not Plutonium

Here is a list of things officially known to cause cancer in humans: tobacco smoke, asbestos, plutonium, solar radiation, alcohol, wood dust — and processed meat. That is not a tabloid's idea of a scary list. It is, roughly, the membership of a single category used by the world's main cancer agency: Group 1 — carcinogenic to humans. Your morning bacon and the core of a nuclear weapon hold the same rank.

Every couple of years a headline announces that bacon, or coffee, or your phone causes cancer, and the internet splits into two camps that are both wrong. One panics and clears out the fridge. The other rolls its eyes — they said coffee too, it's all junk — and orders a second hot dog. Both reactions make the same mistake: they treat "causes cancer" as a single fact, when it is quietly doing two completely different jobs at once.

One job is whether — is there real evidence this thing can cause cancer at all. The other is how much — by how much it actually moves your odds. The headline collapses them into one word. The whole skill here is prying them back apart. It is easier to see than to say, so here is the same fact, drawn to scale.

The label grades the evidence, not the danger

The agency is the International Agency for Research on Cancer (IARC), the WHO's cancer arm, and its famous groups are the source of nearly every "X causes cancer" story you have ever read. Group 1 means carcinogenic to humans; Group 2A, probably; Group 2B, possibly; Group 3, not classifiable. The categories sound like a danger scale running from lethal down to harmless. They are nothing of the kind.

What the group actually grades is how strong the evidence is that the thing can cause cancer at all — not how much cancer it causes. That single distinction explains the whole absurd-looking list. A substance can earn Group 1 with overwhelming proof of a tiny effect, and sit in the very same group as something with overwhelming proof of an enormous one. Processed meat and tobacco are both Group 1 not because they are equally dangerous, but because in both cases the science is equally sure. Bacon and plutonium are filed together on the strength of the proof, and nowhere near each other on the size of the harm.

The agency could not be more explicit that this is the point. Its own Q&A insists that sharing a group "does not mean that they are all equally dangerous" — and reaches for its own example to make it land: active smoking, second-hand smoke, and outdoor air pollution are all Group 1, and carry wildly different risks. Same strength of proof that each belongs on the list; nowhere near the same danger. Reading the rank as a danger scale is the exact mistake the people who built it keep pleading with us not to make.

If that sounds like a system that could make mistakes, it is — and it corrects them in public, which is the tell that it is grading evidence rather than issuing verdicts. Coffee sat in Group 2B — possibly carcinogenic — for twenty-five years. In 2016 IARC looked again, found the evidence didn't hold up, and moved it out to Group 3 — not classifiable. The thing that landed in 2A that year was not coffee but drinking anything above about 65°C, hot enough to scald. The groups are a live scientific verdict on the evidence, revised when the evidence is. They were never a fixed ranking of how scared to be.

Relative risk is a magnifying glass

When IARC put processed meat in Group 1 in 2015, the number that travelled was this: about 18% more bowel cancer for every 50 grams eaten a day — a daily hot dog, or three rashers of bacon. Eighteen percent sounds like nearly a fifth of something. The question the headline never makes you ask is: a fifth of what?

A percentage with no base rate under it is a magnifying glass held over nothing. That 18% is a relative increase — it says how much a baseline grows, not how big the baseline was. So Cancer Research UK poured it back onto the baseline. Of 1,000 people in the UK, about 61 develop bowel cancer at some point in their lives. Among those who eat the least processed meat it's about 56; among those who eat the most, about 66 — a gap of ten in a thousand. That is what an eye-catching seventeen-or-eighteen percent looks like once it is in bodies: ten people, and the other 934 untouched either way.

A big percentage of a small number is still a small number. The relative figure alone never tells you which you're holding.

Now run the same move on smoking, and watch it break the other way. Smoking doesn't raise lung-cancer risk by a tidy 18%; it multiplies it by fifteen to thirty times — and across a long list of cancers, not one. Same Group 1 label as the bacon. Utterly different size. "Same category" was never "same risk." The grid above floods red for one and shows a thin stripe for the other, and that gap is the answer the word "carcinogen" hides.

Put it in bodies and the gap is just as stark. The WHO links diets heavy in processed meat to roughly 34,000 cancer deaths a year worldwide — real, and not nothing. The figure for tobacco is about a million, perhaps thirty times larger; for alcohol, around 600,000. One Group 1 shelf, and a thirty-fold spread from one end of it to the other.

The dose makes the poison

There is a second thing the label leaves out, and it is nearly as old as medicine. Sola dosis facit venenum — the dose makes the poison — Paracelsus, five centuries ago. Whether a substance harms you depends on how much, how often, for how long. Water and oxygen will kill you at the wrong dose; plenty of poisons are medicines at the right one. A Group 1 stamp says a thing belongs on the chart. It says nothing about where on the dose you actually live.

This is the gap between the classification and your kitchen. The 18% is for 50 grams every day, indefinitely — not the occasional sandwich. Sunlight is Group 1, but a walk is not a sunbed at seventeen. Your real risk is never the label alone; it is potency multiplied by how much you're actually exposed. Strip the dose out and you can make almost anything sound alarming, which is precisely how the scary version of the story gets written.

So is it nothing? No — and that's the calibration

Here is where both camps go wrong. The panicker hears Group 1 and bins the bacon. The cynic hears they were wrong about coffee and decides the whole enterprise is theatre. They are the same error wearing opposite clothes: letting the label be the entire answer, in one direction or the other.

The calibrated read sits in between, and it is genuinely useful. Most everyday Group 1 exposures — the daily drink, the regular bacon — add a little to a baseline you mostly can't control, and a sane response is mild: cut back if it's easy, don't lie awake. But a few — heavy smoking, heavy drinking, a sunbed habit started young — add a lot, and there the same evidence is pointing at a risk worth real action. The label can't tell those apart. The numbers can. That is the whole job: not deciding whether to believe the word, but asking how much, and acting in proportion to the answer.

And to be plain about what this is and isn't: it's not medical advice, and it isn't telling you what to put on your plate. Eat the bacon or don't, pour the drink or don't — that call is yours, and it always was. The only thing on offer here is the decision handed back to you at its real size, so you weigh the thing itself instead of the headline's shadow of it.

The headlines will keep telling you things cause cancer, and most of them will be telling the truth — a little. The trick a word like carcinogen plays is to smuggle a true statement about evidence in as a false statement about size. You can refuse the swap. Hold the claim up, find the baseline under the percentage, ask where on the dose you actually live — and you'll find that most of what you were sold as a verdict was only ever a measurement, waiting to be read to scale. (Run any scary headline through the calculator the next time one finds you; for risks more broadly, Dread vs. Data does the same job.)