Comment by estearum
6 days ago
The evidence presented:
The State killed a lot of people between the 14th and 20th centuries and also the homicide rate went down.
Wow!
QED
Good thing there weren't other major confounding changes between errmmm... the longbow and the atomic bomb. Or Dante's Divine Comedy and jazz.
I'm convinced. Why'd you even put the note "if not conclusive" with evidence this strong?
I'd be more shocked if culling a full per cent of men yearly did nothing. Plus a lot died at the scene of crimes or in prison awaiting trial. The question is how much and precisely what. Doing reliable social science is hard enough on current data or interventions. It's very hard with historical data over that sort of time period. However, we get better knowledge by discussing interesting hypotheses and how to study them better. This one is interesting and there may be something to it. It's also at least quasi-testable; someone could fund a study on examining alleles associated with aggression in historical remains.
Note that Frost and Harpending are pretty conservative in their estimates; they figure only ballpark half the decline could be explained by this.
If you were to approach this question with intellectual honesty, you would identify pretty quickly that there are far better ways to try to answer it.
Case-control methods, natural experiments, surveys of criminals, and meta-analyses of the prior.
Literally any method other than "pick 600 year period and say 'vibes shifted generally across a continent and then homicide went down'"
Of course this question has been studied extensively for decades and the current conclusion is: completely inconclusive!
There's some evidence it increases violent crime, some that it decreases it, most evidence doesn't clearly show any effect at all.
So whatever effect it may have, it almost certainly isn't very strong, or is countervailed by opposing effects.
I think that if we're proposing the State, which we know to be fallible in so many cases, should make irreversible decisions like "executing suspected bad guys" more frequently, then we should have extremely strong evidence that it would actually achieve the desired result.
> It's also at least quasi-testable; someone could fund a study on examining alleles associated with aggression in historical remains.
Good luck establishing how "alleles associated with aggression" contributes to violence. I'm pretty sure most of the people who adopt your position would argue that their "aggressiveness" is a virtue in whatever competitive landscape they choose to occupy.
You are talking about the kind of research we can do today. You can't really do case-control for medieval populations easily, nor surveys of criminals, nor of the broader population since everyone is several centuries dead. Natural experiments might work and are exactly one of the things we should see further researched in this area. Meta-analyses can't happen until there's other research to meta-analyze.
I think we're in violent agreement here; yes, this obviously bears further investigation. The way good science gets done is "We have some preliminary evidence that could support a certain hypothesis. We think people should do further investigation." Then you go do that further investigation to see if you can reject the null.
The alleles point, though, is weaker. You're not just looking at stuff like MAO-A activity, also CDH13, COMT, other variants. We actually have a pretty good set worth analyzing that are pretty well-characterized in research, so we don't have to depend on any one particular allele. We have a pretty good set of those that aren't associated with, I don't know, aggression in boardrooms.