> For context, about 1 in 54,000 people will die of a wasp or hornet attack, making that cause twice as risky as this red dye Yes, and we test people for aller

> For context, about 1 in 54,000 people will die of a wasp or hornet attack, making that cause twice as risky as this red dye Yes, and we test people for allergy to these attacks, and often those people carry interventions like EpiPens, etc.—this risk is high enough that we do something about it, no? People clear nests near their homes, etc. Why not do something about red dye? I agree that it's unlikely to affect many people. At a population level, we often take action to further reduce low-risk events because the aggregate benefit is high (2400 cancers prevented) and the cost is low (a few companies spend a few bucks to replace a barely-toxic red dye with a non-toxic one). One reason we do this is that the same trade-off at the individual level is comparatively expensive and wasteful, as the benefits only accrue (and the costs may only be low) if you can aggregate them over a large enough sample size. Consider another example: most people are not affected by vehicle accidents, but the risk is still there and pretty high. At the individual level, there is very little you can do besides "not drive" and "live in a pedestrian area". Historically, few car manufacturers included safety features before they were mandated, because they increased cost and consumers did not perceive the risk as high enough to be worth it. Regulation that mandates these safety measures raises the cost floor, but not by as much because now the extra cost is somewhat amortized over the full fleet for sale. Further, some of these safety features primarily benefit people other than the vehicle owners, like backup cameras. How many people are killed by vehicles backing up? Lifetime risk is also about 1 in 100,000. Yet we still mandate this feature in new vehicles because the cost is comparatively low. There are plenty of these kinds of trade-offs that regulators make at a population level that would not necessarily make sense at an individual level. The individual impact is quite low. But the combined aggregate effect is probably quite large.