John “not Jaws” Williams writes:
For some time, I’ve noticed postmodernist views cropping up in the magazine Science, mostly in the book reviews. In the 23 Feb 2024 issue, however, they arrived full blown in a set of opinion pieces about genetics education (e.g., Duncan et al., Vol. 383, Issue 6685, pp. 826-828, excerpt below):
“The methods of conducting genetics research and its outcomes are steeped in, and influenced by, power and privilege dynamics in broader society. The kinds of questions asked, biological differences sought, and how populations are defined and examined are all informed by the respective dominant culture (often Eurocentric, white, economically privileged, masculine, and heteronormative) and its predominant ways of knowing and being (3). Findings from human genetics and genomics research subsequently play into existing sociopolitical dynamics by providing support for claims about putative differences between groups and the prevalence of particular traits in particular groups (3). Historically, such research has been used in support of eugenic movements to legitimize forced sterilization and genocides. Yet it would be a mistake to assume that such research is merely a discredited past relic, a stain on the otherwise objective and rational track record of genetic research. Rather, it was mainstream work conducted by prominent researchers and supported by major professional societies. The reality is that some modern human genetics is still informed by the same racist logic (4).” (Duncan et al., The sociopolitical in human genetics education; (3.) J. A. Hamilton, B. Subramaniam, A. Willey, Fem. Stud. 43, 612 (2017). (4.) A. C. F. Lewis et al., Science 376, 250 (2022).
Well, yes, science is done by people, and is subject to all the silly and sometimes pernicious stuff that goes with that, but this seems overblown, to say the least, and not that well founded. For example, ref. 4 argues against using any geographically based categories for genetic variation among people, by showing that the clustering evident in geographically based genetic samples is not apparent in BioMe data from New York, where almost everyone is from somewhere else – therefore, the geographical clusters are a “by-product of sampling strategy” [see their Fig. 1].
As an old progressive, I more or less agree with a lot of the points Duncan et al. make, but I still find their essay alarming. Duncan et al. do acknowledge that science can produce “knowledge that is credible and valuable,” but postmodernists by and large assume that since scientific knowledge is socially created, like other kinds of knowledge, it should not be “privileged” over other ways of knowing. I see this where I live in NW California, where some involved with environmental restoration want to put “indigenous science” on a par with regular science. However, I have the sneaking suspicion that postmodernists also want to claim that their “way of knowing” is best – this fairly oozes out of the language quoted above.
Perhaps the main thing that bothers me about Duncan et al. is that they say nothing about the role of genetics in undermining the ideas they complain about, such as the notion that races are “real.” There is nothing, for example, about Lewontin’s celebrated 1972 paper showing that, averaged over loci, there is much more genetic diversity within rather than among groups. This gets to the question: given that genetics has been misused in the past, what should people teaching genetics now do about it? My sense is that the most important thing is to teach genetics well; that showing students why the concept of races as biological categories doesn’t make sense is more persuasive than telling them that it doesn’t.
Yeah, I know that the above is about genetics, not statistics, but I expect that statistics have been used to support the “respective dominant culture,” too.
I’m not sure how to think about all this. Postmodernism doesn’t come up in the sort of political science that I do. Or maybe I should say that it’s all postmodern, in the sense that even basic questions such as the value of democracy are questioned; there are no foundations. And postmodernism doesn’t come up in statistics at all. Political issues do arise in statistics, most notoriously with biological measurements, but these involve specific application areas, not the core of statistics. So when I hear about stories such as above, I’m coming from the outside.
P.S. Several commenters say that Duncan et al. make some good points, even if their writing style is annoying and they also have some major lapses. I think that’s kinda Williams’s point! If he thought the arguments had no merits at all, maybe he wouldn’t have thought it even noteworthy that it has problems.
Re: “This gets to the question: given that genetics has been misused in the past, what should people teaching genetics now do about it? My sense is that the most important thing is to teach genetics well; that showing students why the concept of races as biological categories doesn’t make sense is more persuasive than telling them that it doesn’t. ”
I heard a talk from someone who was cited by the Buffalo shooter (Robbee Wedow) who addresses this head on.
The answer they shared was going into highschool classrooms and emphasizing a genetics curriculum centered around the exact quote above “More variation within groups than between”. They developed and implemented a whole curricula centered around this. It was reportedly very effective in educational outcomes.
https://link.springer.com/article/10.1007/s10519-026-10251-7
I normally consider pointing out typos very annoying, but this one is important: “In the 23 Feb 2004 issue” should be *2024*. (I was wondering at first: who cares about a 22 year old article?) Incidentally, it’s 22 Feb.
Raghu:
The typo was my bad. In John’s note to me, he just wrote 23 Feb, then I added 2024 to be clear. But I mistyped it! I just went in and fixed it.
If you took a more postmodern perspective on surveys, you’d be asking what “happiness” means on a 1-5 scale and what “science for good” really means. If you took a more postmodern view on politics, you’d be asking who the two-party system serves. I’d be shocked if this kind of work isn’t going on somewhere in the science-adjacent humanities literature.
I did love this quote from Williams, “However, I have the sneaking suspicion that postmodernists also want to claim that their “way of knowing” is best – this fairly oozes out of the language quoted above.” And while that may be true, their way of knowing resolves to accepting the fact that you can’t “know,” because all the concepts on which our knowledge is based are social constructs. We really need the other Gelman for the cognitive science take on this. I thought about this kind of thing a lot when I worked on semantics and taught philosophy of language. The problem is that like statistics, it’s a deep dive to get to the forefront of people’s thinking on these topics, and I’m no longer up to date. I can highly recommend the first article in Richard Rorty‘s book, Contingency, Irony, and Solidarity, though I worry it may still be too technical for someone who hasn’t taken a year or so of 20th century analytical philosophy classes—ironically, Rorty’s being hoisted by his own petard in his writings.
I also enjoyed that quote, but my interpretation was that by saying their way of “knowing” is best and implicitly criticising any other (read: the ones usually considered mainstream) as being based on privilege rather than merit, they’ve tried to preclude any room for good faith disagreements. Can’t really debate a theory when one party says you shouldn’t be allowed to participate in the discourse.
I don’t think John Williams has actually engaged even superficially with any of the thinkers considered postmodern, and has just piled all left-leaning academic thought together. Postmodern, postcolonial, neo-marxist, third-wave feminist, surely all those are secretly the same thing
It honestly sounds like a parody of so-called “grievance studies”, except the grievance about “Eurocentric, white, economically privileged, masculine, and heteronormative” is too on the nose.
The reasoning in the quotation from Duncan et al. isn’t, on its face, “postmodern” or “anti-science.” Every claim they make is straightforwardly empirical and in line with decades of work in the history, philosophy, and sociology of science. (If anyone’s interested in a quick overview of what philosophers of said about this, I’d recommend the readings from unit 3 of my philosophy of science syllabus: https://philsci-ucm.netlify.app/)
Williams makes four specific critiques of Duncan et al. First, that they’re overinterpreting their evidence. That critique might or might not be accurate, and we could debate about the appropriate standards of evidence for a book review or opinion piece vs. a research article. But, in any case, it’s a completely ordinary scientific critique. If Duncan et al. responded by, say, dismissing the very idea of empirical evidence, that would be one thing. But we have no evidence that’s how they’d respond.
Second, that postmodernists think that (a) science cannot produce “knowledge that is credible and valuable” and/or (b) “scientific knowledge … should not be “privileged” over other ways of knowing.” Williams acknowledges that Duncan et al. explicitly reject (a), so that critique is irrelevant. As for (b), because there’s no evidence that Duncan et al. are postmodernists, this critique is not (yet) relevant. In the quote, Duncan et al. do not appear to make any claims at all about any other extra- or non-scientific ways of knowing.
Williams’ third critique is that Duncan et al. do not cite one particular paper published 52 years before their piece, which made a relatively narrow point about genetic variation using the tools available at the time. Again, we could argue about how thorough the bibliography should be for a short opinion piece; but in any case this is a perfectly ordinary intra-scientific critique.
The fourth critique (which I’ll acknowledge is reading a bit into Williams) is that Duncan et al. do not explain what genetics teachers should do to avoid racist misappropriations of genetic research. Here I would note that the final section of Duncan et al. is on “Recommendations for Education.” Williams might also be interested in this recent profile in Stat of Brian Donovan’s work, to empirically address exactly this question: https://www.statnews.com/2026/04/07/brian-donovan-fighting-racism-with-genetics-education/
That quotation is just an excerpt from the larger piece. Williams doesn’t say he finds the excerpt “alarming”, he says he finds the essay alarming: “… I more or less agree with a lot of the points Duncan et al. make, but I still find their essay alarming.”
All I’ve read is the excerpt, and I am still dizzy from my eyes rolling involuntarily so many times… in spite of the fact that I agree with the factual content of every statement in it!
Agreed! Williams seems to largely agree with the article but imagines things he can critique them for.
I posted below, but this is really well stated. The article is pro-science! It’s asking how to counter the anti-science approaches of people who use socially constructed categories to do pop culture interpretations of “heredity” in some kind of 9th grade fruit fly experiment way.
Williams makes a lot of good points, and the Duncan et al article is pretty terrible, but I think people might be misled by his sentence, “ref. 4 argues against using any geographically based categories for genetic variation among people, by showing that the clustering evident in geographically based genetic samples is not apparent in BioMe data from New York, where almost everyone is from somewhere else – therefore, the geographical clusters are a “by-product of sampling strategy” [see their Fig. 1]. ” To clarify, ref 4 is *not* saying something as dumb as “hey look, all these people are from New York and they’re genetically diverse, so geography doesn’t carry information about genetics”. They’re saying that if you sample people from all over the world, you find a continuum of genetic variation, whereas discrete clusters are produced by sampling schemes that instead concentrate sampling in a few well-separated populations and don’t sample people living between them.
To put it another way, ref 4’s primary problem with “geographically based categories” is not the “geographically based” part, it’s the discrete categories part. You can agree or disagree with that—I just wanted to clarify their argument.
“Well, yes, science is done by people, and is subject to all the silly and sometimes pernicious stuff that goes with that,”
Uh, yes. You’re framing it in sort of a handwavey way, but what it means is that not terribly long ago in American history, science told people that there was a hierarchy of races, and that science was used to drive public policy like eugenics. That’s a really serious problem, especially because science is still “done by people.”
“Duncan et al. do acknowledge that science can produce “knowledge that is credible and valuable,” but postmodernists by and large assume that since scientific knowledge is socially created, like other kinds of knowledge, it should not be “privileged” over other ways of knowing.”
That’s sleight of hand on your part. You start by talking about Duncan (who say something you agree with) but then switch in mid-stream to unnamed “postmodernists” and criticize them. If Duncan is not actually saying this, why are you concerned about the article? Because you have a “sneaking suspicion” about stuff that’s oozing out of the language?
I’m not sure who you’re arguing with but it doesn’t seem to be Duncan.
I think you may be trying to respond to Andrew, but most of this post is a quote from someone else. Andrew just says at the end, “I’m not sure how to think about all this.”
No, I know. I was thinking of Williams as “you”
Total,
You are right that Duncan et al. did not explicitly say that science is just another socially created way of thinking, but I think a fair working definition of a postmodernist is: anybody who writes language like the excerpt I sent to Andrew.
They *did not say* what you are criticizing. “You look like a post-modernist” does not change that.
This whole thing amounts to “I don’t like the writing style.” Ok then. Good for you
If you are a scientist you should be doing the reading before using a specific specific term that has a real definition. Just say you don’t like to read paragraphs that make you uncomfortable. It’s fine to say that. No one is asking you to do serious study of the sociology of science.
I got a PhD in genetics, and I’ve always found the geographically-based population genetics nonsensical. People are always moving all over the place, not just in NYC. The rate of movement may be faster or slower, but it’s never zero anywhere. It gets worse when they assign names to the populations and talk of, e.g. a “French” genome, when being of French nationality is a political construct that is decoupled from genetics (e.g. imagine an African immigrating to France and becoming a French citizen, or a Frenchman immigrating to America and becoming an American citizen). They kind of justify their approach by saying they sample people who have “traditionally” lived in an area, but 1) that is kind of subjective and 2) excludes the non-“traditional” people. Suppose we found a group of families who have traditionally lived in the NYC area (since when? 1776? the Dutch colony? or prior to that?) and extensively sequenced their genomes. The resulting information would tell us nothing about the genomes of 99% who actually live in NYC right now! That is perhaps the worst-case scenario for population genetics, but I still feel the approach would be uninformative for all but the most isolated, inbred villages. So, I actually think the articles under discussion have a good point and are not anti-science at all. In fact, by questioning dogma and highlighting the limitations of our knowledge, they are helping to advance the scientific process.
Our understanding of things like sickle cell anemia is greatly enriched by geographically-based population genetics, and that understanding has reduced the suffering of a marginalized population.
In the end, geographically-based population genetics is a tool in a researcher’s toolbox. Where did Native Americans come from? How long have the North Sentinelese been isolated? Are Okinawans genetically advantaged in reaching old age? These are questions for geographically-based population genetics. It doesn’t make sense to criticize a tool because someone has, might, or will misuse it.
Yes, it’s a good point. The massive gnomAD* database for example, contains nucleotide and amino acid sequence variants from getting on for a million individuals so far (from full genomes or exomes). Not surprisingly it is valuable to drill down into discrete populations and so the data are also compiled as separate entries for population groups (European, African American, Ashkenazi Jews, South Asian etc. – Finnish and Amish populations are also grouped if I remember correctly). This allows for identification of group-specific variation with just the sort of disease relevance of the sort you describe (for sickle-cell disease).
Two other huge benefits are the identification of highly-represented specific genetic variants which can then be removed from clinical assessment pipelines (if lots of people carry a specific variant then it’s unlikely to be a pathogenic variant) and identification of population groups where more sequence info would be useful (e.g. under the “All of US” program).
* Large highly cooperative reseach efforts like gnomAD (there are lots of examples) rather gives the lie to the overwrought doomongering represented on the recent “I have seen the future of science. It is ruled by bitter competition instead of collaboration….”
Here is a non-paywalled link to the entire Duncan et al article:
https://www.science.org/doi/pdf/10.1126/science.adi8227
I’m not recommending that you read this, but if you do, there is a sort of buzzword bingo that can help to make it a bit more entertaining. Count the straw men that get slain like orcs in a Tolkien movie!
I sympathise with the Duncan et al. opinion piece despite some of the odd framings described above. It does little harm to attempt to become more inclusive and cognisant of potential slights to others.
On the “strawman” point, this one caught my eye:
“The current instructional overemphasis on Mendelian inheritance and the central dogma contribute to problematic views that are deterministic (genes exclusively dictate phenotypes) and essentialist (groups are homogeneous and inherently different from each other).”
Really? Learning about Mendelian inheritance is fundamental to understanding human disease especially the large number of diseases associated with genetic variants. That should certainly be a part of any curriculum in genetics. To suggest that studying Mendelian inheritance (it’s not clear what’s meant by “overemphasis”) might tend towards a view that “genes exclusively dictate phenotype” is certainly an odd one! Don’t think anyone thinks that.
Likewise, what has “the central dogma” got to do with this? If by “central dogma” they mean Francis Crick’s “central dogma” then haven’t they got the wrong end of the stick? The “central dogma” simply states that sequence information doesn’t flow from protein to nucleic acid or protein. That’s irrelevant in the context of deterministic or essentialist views.
I don’t actually see anything postmodern about this. Sure maybe it would benefit from some positive examples but this
“We build on existing efforts in genetics education to shift instruction to more complex and accurate models of gene–environment interactions (multifactorial inheritance) and human trait variation within and across populations (5). Such complex knowledge of genetics, and the explicit countering of race as biological, can lessen genetic deterministic and essentialist belief”
really gets at the crux. We see it on the comments all the time. People with no training in modern genetics–using “Mendelian inheritance” models, talking vaguely about “demographics” when they mean “Black people” using observational data with no genetic information to make claims about which groups are inherently smarter than which other groups. Even though, genetically speaking, there are no such groups. Yes, socially speaking, there are such groups. And those groups are “real because people believe they are real.” And they are real in their consequences. But talking about the social and political consequences of these constructed groups — all too real the day after the wiping out of the little that remained of the voting Rights Act — is something very different than talking about the consequences of genetic variation. Genetics did not lead 6 justices to write an opinion that in a political sense will harm Black voters.