[mikeash]

In theory, all regulations need to be carefully considered and walk a fine line in order to do more good than harm.

But I was replying to a pair of comments talking about the practical history of regulations over the past few decades. That history diverges wildly for the two types of regulation. Economic regulation was massively lightened, resulting in huge benefits. At the same time, safety regulations for airplanes and cars were massively increased, also resulting in huge benefits.

So, yes, the whole point is that “regulations good/bad/anything” is nonsense, as the recent history of economic and safety regulations illustrates.

[AnthonyMouse]

What I'm getting at is that the types of regulation don't really have anything to do with it. It's not that all safety regulation is good and all economic regulation is bad (though if you wanted someone to disprove "all economic regulation is bad" you might be in for a wait). You can screw up safety regulations just as bad if you're not careful.

It's also difficult to evaluate these things based on statistical trends, because it's easy to pass a safety regulation that e.g. mandates a specific safety method, and then you get an immediate safety improvement by making that state of the art method universal, but as time passes that ossified requirement becomes stale and prevents newer, better methods from replacing it. And all you see in the numbers is the improvement from the original mandate spreading through the installed base over time, not the comparison to what would have happened when something even better was invented a year later but was prohibited by the rule from being deployed. (As evidence for this, the original 737 was introduced prior to the moon landing.)

It's also problematic to look only at the one target. If we improve aircraft safety per mile but only by making aircraft more expensive (or destroying competition, which does the same thing), that makes flying less competitive against alternatives that are much more dangerous, like driving. So you can get a safety improvement on paper even though the result is more people are killed.

That doesn't mean you can't make a rule that actually improves safety, but it's harder than it seems to get it right.

[mikeash]

Sure. I’m just saying that the actual history of the past few decades in the US shows a stark contrast with these two types of regulations in terms of how they worked and the consequences they had.

Regarding measures that appear to improve safety but are actually a net loss because it encourages driving instead, the FAA does evaluate regulations this way. For example, this is why infants aren’t required to be in a child seat and are allowed to just be held by a parent. They determined that this requirement would cause more families to drive due to the extra cost, and the result would be a net loss of life.

[jaclaz]

>In theory, all regulations need to be carefully considered and walk a fine line in order to do more good than harm.

Yes, and one of the possible definitions of a (technical, safety related) regulation is "the added cost that the society is reasonably willing (or capable) to pay for a reasonable increase in safety"

The issues always revolve around how much the "reasonably willing to pay" is and what is the corresponding "reasonable" increase in safety (and how exactly to measure the latter).