But there isn’t one size of food producer. I would think there are diminishing marginal returns on food safety after a certain size. At that point perhaps the blast radius of mistakes offsets the advantage of scale. It’s a false dichotomy to consider only national producers and local producers.

Why though? There’s nothing inherent about speeding production up, producing more at once, or increasing profits that makes things cleaner or safer, and the obvious argument is the opposite. If the regulations that force cleanliness and safety get peeled back too far, you could be left with all of the conditions for worst case scenarios because of how much more efficient the thing has become.

Is it just a question of variance in the size of the issue? Is having 10 incidents of 10 people each the same as one with 100?

I’ve argued in systems reliability that once you are already high reliability you are better off lowering incident size vs total unavailability.

What data might that be? I've looked and there doesn't appear to be any authoritative source of long-term data for either # of reported foodborne illness outbreaks or # of individuals sickened in this way per year, either of which would clearly demonstrate a dip in sickness rates if industrial-scale food production does, in fact, result in cleaner production.