Ok, I just took a look at the sample result set that Eby linked to, and I, like you, am not impressed. This guy is so much a part of the noise, he isn’t even the noise that could be reasonably mistaken for a real expert.
So… here’s the thing. You and wedrifid are doing something that has me concerned.
Specifically, you’re putting me into a position where, for consistency, I feel compelled to argue a case for something that I myself don’t currently have a hugely high degree of confidence in… simply because you’re not actually providing in your arguments, any information which I could either specifically agree or disagree with.
IOW, comments like, “quack”, “laughable”, and “noise” do not give me any information about your epistemology, and therefore light up on my board “[citation needed]” just as much as it did for what I’ve been reading from D’Adamo.
So, it would be nice if you could identify specific concerns instead… who knows, I might agree with them!
OTOH, my consistent experience is that just because somebody has a stupid-sounding theory, doesn’t mean their advice doesn’t actually work. (Likewise, people who have good theories are often lousy at giving usable advice.)
Heck, take Seth Roberts as an example here: the entire idea of drinking oil or sugar water to lose weight is also “laughable”, “noise” and “quack”… and yet it still seems to work for plenty of people.
Heck, there are elements of Roberts’ theory that don’t make sense to me, from a “fewest elements to make the circuit” point of view. (For example, I don’t think “set point” is a real thing; I think it’s more likely an epiphenomenon of something else.)
But that doesn’t mean I wouldn’t recommend it… just that I’d avoid recommending it to people who jump to conclusions first and ask questions later. ;-)
Anyway… specifics, please. Otherwise, I’ll simply bow out of this discussion on the assumption that you don’t actually have any new evidence to present.
IOW, a knee-jerk dismissal on grounds of ridiculousness isn’t an independent data point for Bayesian analysis. Citing papers disproving D’Adamo’s hypotheses, on the other hand, would be quite welcome.
(I’ve actually googled around for criticisms of both his blood type and genotype work, and have yet to find a single cite that doesn’t have a more-recent countercite; the vast majority of criticism, however, appears to be of the, “that’s so silly I won’t even bother to argue the idea seriously” variety. Maybe this is evidence that he’s a quack, but at least his responses to the critiques include some citations in his favor.)
A big red flag is that where I expected to see an analysis of what mixture of the various ancestral types a person is, I see a declaration that the person fits into one of 6 little boxes. This despite the fact these types describe multiple features, controlled by different genes, in a sexually reproducing population that represents all types. Now, not all genes are selected for independantly, genes nearby on the same chromosone can be correletated. But a model of which dietary traits correspond to which easily measurable traits should be more complicated than assigning a cluster of dietary traits to a cluster of easily measured traits.
A big red flag is that where I expected to see an analysis of what mixture of the various ancestral types a person is, I see a declaration that the person fits into one of 6 little boxes.
FYI, my wife’s actual profile showed this; more precisely, IIRC it rated her as 44% Explorer, based on the traits given. It did not show what percentages the other 5 boxes broke down to, and I don’t know whether those factors were also taken into account in the analysis. (I also don’t know what precisely the percentage represents; i.e. is it a probability, a “percentage of your traits”...?)
The sample profile I linked appears to date from 2008; so perhaps the percentage report was added to the software later. But in both cases, if I understand correctly, the software simply presents the highest-scoring of the six boxes, rather than saying, “this is you”.
Still, compared to most ways of nutritionally grouping people, six is actually a LOT of boxes.
But a model of which dietary traits correspond to which easily measurable traits should be more complicated than assigning a cluster of dietary traits to a cluster of easily measured traits.
From what I read in his book, he describes the types in terms of basic strategies for responding to the environment, where there are only a few good choices to make. IOW, the stereotypes are supposed to represent stable strategies for responding to infections, shortages, and other stressors. That is, there are not an unlimited number of ways to do things in those areas, so you end up with large clusters.
I have not studied this in any detail, mind you; I confess my primary interest in the book was more to look at the food lists for my type, to compare against my personal dietary history.
As I said, I’m less interested in the plausibility or sensibility of a theory per se, than with the correlation of its advice with the obtaining of results… especially results for myself in particular. (And I remain cautiously optimistic on that front where his advice is concerned.)
FYI, my wife’s actual profile showed this; more precisely, IIRC it rated her as 44% Explorer, based on the traits given. It did not show what percentages the other 5 boxes broke down to, and I don’t know whether those factors were also taken into account in the analysis. (I also don’t know what precisely the percentage represents; i.e. is it a probability, a “percentage of your traits”...?)
Better, but still not good enough. If it is a mixture, what about the other 56%? Which 44% of the explorer traits? If it is a confidence level, the model doesn’t seem to rate itself very highly, so why should I be impressed with it?
From what I read in his book, he describes the types in terms of basic strategies for responding to the environment, where there are only a few good choices to make. IOW, the stereotypes are supposed to represent stable strategies for responding to infections, shortages, and other stressors. That is, there are not an unlimited number of ways to do things in those areas, so you end up with large clusters.
Suppose there are 2 stable strategies that each say how to deal with several specific problems, such that either of the 2 strategies work but a mixture somehow fails. These strategies, being complicated, are coded for by multiple genes. Suppose a man with one strategy and a women with the other strategy mate and have offspring. Those offspring are going to inherit some mixture of the two strategies, even discounting complications such as being heterzygous where the parents are homozygous, and therefore will employ an unfavorable mixture. You can not have multiple non mixable complicated traits in a sexually reproducing population, without tricks like having each member have the complete code for all possible traits, and a varying gene that switches on one of them, which we observe in sexual dimorphism at not much otherwise. Eliezer has written of this.
You can not have multiple non mixable complicated traits in a sexually reproducing population, without tricks like having each member have the complete code for all possible traits, and a varying gene that switches on one of them, which we observe in sexual dimorphism at not much otherwise.
But we do observe epigenetic traits—variations in gene expression based on environmental conditions, such as genes that act differently depending on how much exercise you get, or the level of testosterone in the fetal environment, or various other things.. D’Adamo’s claim here is that his typing groups are a combination of gene inheritance and gene expression, and his notion of “strategies” isn’t really the same as say, having a completely different way of digesting foods.
It’s more like identifying which places to store fat in first—something that (IIUC) we already know is heritable. The fact that fat gets stored isn’t changed, just how much, where, and how quickly. Something like that can make a big difference on a practical level to a person’s life, without being a particularly complex adaptation in itself.
Ok, I just took a look at the sample result set that Eby linked to, and I, like you, am not impressed. This guy is so much a part of the noise, he isn’t even the noise that could be reasonably mistaken for a real expert.
So… here’s the thing. You and wedrifid are doing something that has me concerned.
Specifically, you’re putting me into a position where, for consistency, I feel compelled to argue a case for something that I myself don’t currently have a hugely high degree of confidence in… simply because you’re not actually providing in your arguments, any information which I could either specifically agree or disagree with.
IOW, comments like, “quack”, “laughable”, and “noise” do not give me any information about your epistemology, and therefore light up on my board “[citation needed]” just as much as it did for what I’ve been reading from D’Adamo.
So, it would be nice if you could identify specific concerns instead… who knows, I might agree with them!
OTOH, my consistent experience is that just because somebody has a stupid-sounding theory, doesn’t mean their advice doesn’t actually work. (Likewise, people who have good theories are often lousy at giving usable advice.)
Heck, take Seth Roberts as an example here: the entire idea of drinking oil or sugar water to lose weight is also “laughable”, “noise” and “quack”… and yet it still seems to work for plenty of people.
Heck, there are elements of Roberts’ theory that don’t make sense to me, from a “fewest elements to make the circuit” point of view. (For example, I don’t think “set point” is a real thing; I think it’s more likely an epiphenomenon of something else.)
But that doesn’t mean I wouldn’t recommend it… just that I’d avoid recommending it to people who jump to conclusions first and ask questions later. ;-)
Anyway… specifics, please. Otherwise, I’ll simply bow out of this discussion on the assumption that you don’t actually have any new evidence to present.
IOW, a knee-jerk dismissal on grounds of ridiculousness isn’t an independent data point for Bayesian analysis. Citing papers disproving D’Adamo’s hypotheses, on the other hand, would be quite welcome.
(I’ve actually googled around for criticisms of both his blood type and genotype work, and have yet to find a single cite that doesn’t have a more-recent countercite; the vast majority of criticism, however, appears to be of the, “that’s so silly I won’t even bother to argue the idea seriously” variety. Maybe this is evidence that he’s a quack, but at least his responses to the critiques include some citations in his favor.)
A big red flag is that where I expected to see an analysis of what mixture of the various ancestral types a person is, I see a declaration that the person fits into one of 6 little boxes. This despite the fact these types describe multiple features, controlled by different genes, in a sexually reproducing population that represents all types. Now, not all genes are selected for independantly, genes nearby on the same chromosone can be correletated. But a model of which dietary traits correspond to which easily measurable traits should be more complicated than assigning a cluster of dietary traits to a cluster of easily measured traits.
FYI, my wife’s actual profile showed this; more precisely, IIRC it rated her as 44% Explorer, based on the traits given. It did not show what percentages the other 5 boxes broke down to, and I don’t know whether those factors were also taken into account in the analysis. (I also don’t know what precisely the percentage represents; i.e. is it a probability, a “percentage of your traits”...?)
The sample profile I linked appears to date from 2008; so perhaps the percentage report was added to the software later. But in both cases, if I understand correctly, the software simply presents the highest-scoring of the six boxes, rather than saying, “this is you”.
Still, compared to most ways of nutritionally grouping people, six is actually a LOT of boxes.
From what I read in his book, he describes the types in terms of basic strategies for responding to the environment, where there are only a few good choices to make. IOW, the stereotypes are supposed to represent stable strategies for responding to infections, shortages, and other stressors. That is, there are not an unlimited number of ways to do things in those areas, so you end up with large clusters.
I have not studied this in any detail, mind you; I confess my primary interest in the book was more to look at the food lists for my type, to compare against my personal dietary history.
As I said, I’m less interested in the plausibility or sensibility of a theory per se, than with the correlation of its advice with the obtaining of results… especially results for myself in particular. (And I remain cautiously optimistic on that front where his advice is concerned.)
Better, but still not good enough. If it is a mixture, what about the other 56%? Which 44% of the explorer traits? If it is a confidence level, the model doesn’t seem to rate itself very highly, so why should I be impressed with it?
Suppose there are 2 stable strategies that each say how to deal with several specific problems, such that either of the 2 strategies work but a mixture somehow fails. These strategies, being complicated, are coded for by multiple genes. Suppose a man with one strategy and a women with the other strategy mate and have offspring. Those offspring are going to inherit some mixture of the two strategies, even discounting complications such as being heterzygous where the parents are homozygous, and therefore will employ an unfavorable mixture. You can not have multiple non mixable complicated traits in a sexually reproducing population, without tricks like having each member have the complete code for all possible traits, and a varying gene that switches on one of them, which we observe in sexual dimorphism at not much otherwise. Eliezer has written of this.
But we do observe epigenetic traits—variations in gene expression based on environmental conditions, such as genes that act differently depending on how much exercise you get, or the level of testosterone in the fetal environment, or various other things.. D’Adamo’s claim here is that his typing groups are a combination of gene inheritance and gene expression, and his notion of “strategies” isn’t really the same as say, having a completely different way of digesting foods.
It’s more like identifying which places to store fat in first—something that (IIUC) we already know is heritable. The fact that fat gets stored isn’t changed, just how much, where, and how quickly. Something like that can make a big difference on a practical level to a person’s life, without being a particularly complex adaptation in itself.