This rubs me wrong for the same reason that “no evidence for...” claims rub me wrong.
We have a probably-correct model, the hygiene hypothesis broadly understood. We have a plausible corollary of that model, which is that kids eating dirt helps their immune system (I had never heard this particular claim before, but since you mention it, it seems like a plausible corollary). We should have a low-but-not-ridiculously-low prior on this.
(probably some people would say a high prior, since it follows naturally from a probably-true thing, but I don’t trust any multi-step chain of reasoning in medicine)
When I read the title, I thought “Oh! I guess someone showed the specific behavior of eating dirt doesn’t help, so I should update against the hygiene hypothesis!” But the post presents no evidence this is wrong. It’s just saying there are no studies of it.
This seems kind of like framing the proverbial parachute point as “‘Parachutes prevent falling injuries’” Is Basically Made Up”. It’s not made up! It was assigned a high prior based on other things we know! Nobody has given us any evidence for or against that prior, so we should stick to it.
I think “don’t let kids eat dirt” originally had a much lower prior than “parachutes prevent falling injuries”, that was specifically overcome by the impression of evidence that doesn’t exist. There are lots of things in dirt we know are dangerous- pesticides, car exhaust, lead, animal waste… Maybe the benefits of dirt outweigh that, maybe they don’t, we don’t know because no one has checked. I also expect us to notice that parachutes fail without rigorous evaluation, whereas the effects of marginal dirt will be harder to notice.
I will be sad if people walk away with the impression that HH/OF are wrong, or that geophagy has concrete evidence against it rather than merely an absence of evidence and some very concrete reasons to think city dirt in particular is dangerous. That’s why I went out of my way to specify “no evidence” and “only applies to dirt, not hygiene/old friends hypothesis in general”, even though it was more annoying to write and I imagine makes the piece less fun to read. I suppose I could have leaned into that trade-off harder, but I don’t expect it to make any difference in practice. I’m curious what you think I could have done differently that would leave people with more accurate beliefs.
[One possibility is “don’t publish at all”, but I think “hey, this common wisdom has much less support than you think” is in general useful to say when true and that banning the genre would be net negative].
This rubs me wrong for the same reason that “no evidence for...” claims rub me wrong.
We have a probably-correct model, the hygiene hypothesis broadly understood. We have a plausible corollary of that model, which is that kids eating dirt helps their immune system (I had never heard this particular claim before, but since you mention it, it seems like a plausible corollary). We should have a low-but-not-ridiculously-low prior on this.
(probably some people would say a high prior, since it follows naturally from a probably-true thing, but I don’t trust any multi-step chain of reasoning in medicine)
When I read the title, I thought “Oh! I guess someone showed the specific behavior of eating dirt doesn’t help, so I should update against the hygiene hypothesis!” But the post presents no evidence this is wrong. It’s just saying there are no studies of it.
This seems kind of like framing the proverbial parachute point as “‘Parachutes prevent falling injuries’” Is Basically Made Up”. It’s not made up! It was assigned a high prior based on other things we know! Nobody has given us any evidence for or against that prior, so we should stick to it.
I think “don’t let kids eat dirt” originally had a much lower prior than “parachutes prevent falling injuries”, that was specifically overcome by the impression of evidence that doesn’t exist. There are lots of things in dirt we know are dangerous- pesticides, car exhaust, lead, animal waste… Maybe the benefits of dirt outweigh that, maybe they don’t, we don’t know because no one has checked. I also expect us to notice that parachutes fail without rigorous evaluation, whereas the effects of marginal dirt will be harder to notice.
I will be sad if people walk away with the impression that HH/OF are wrong, or that geophagy has concrete evidence against it rather than merely an absence of evidence and some very concrete reasons to think city dirt in particular is dangerous. That’s why I went out of my way to specify “no evidence” and “only applies to dirt, not hygiene/old friends hypothesis in general”, even though it was more annoying to write and I imagine makes the piece less fun to read. I suppose I could have leaned into that trade-off harder, but I don’t expect it to make any difference in practice. I’m curious what you think I could have done differently that would leave people with more accurate beliefs.
[One possibility is “don’t publish at all”, but I think “hey, this common wisdom has much less support than you think” is in general useful to say when true and that banning the genre would be net negative].
It sounds like the problem might actually be mainly the title? Instead
No one has actually checked whether (Literally) Eating Dirt Benefits Kids
or something.