Okay so, it all started with a children’s cartoon.

In December, the BBC released on YouTube an old animated video about life in Roman Britain, which featured a family with a dark-skinned father. This depiction recently caught the ire of an Infowars editor, who tweeted, “Thank God the BBC is portraying Roman Britain as ethnically diverse. I mean, who cares about historical accuracy, right?”

To which Mary Beard—best known as a classicist at Cambridge, and more recently known for taking on internet trolls—replied, “this is indeed pretty accurate, there's plenty of firm evidence for ethnic diversity in Roman Britain.” To which Nassim Nicholas Taleb—best-known for railing about epistemic arrogance in The Black Swan, and recently known for arguing on Twitter—replied:

Taleb went on to tweet several charts of DNA variation among modern Europeans that he presented as “data” as opposed to Beard’s “anecdotal reasoning.” And so Taleb and Beard went back and forth, back and forth.

Oh how quickly the conversation jumped from children’s cartoon to Infowars rant to genetics. Having completed a close reading of the entire thread—you’re welcome—I think the most charitable interpretation is a classic Twitter case of arguing past one another. Beard is saying there were indeed dark-skinned people in Roman Britain. Taleb cries BS: A mixed family was not typical of the time. Those positions are not inconsistent. We each have hills to die on, I suppose.

That genetics even came up at all in a debate about ancient Roman history is indicative of science’s stature in these fractious times. Genetics gets invoked as neutral, as having none of the squishiness of historical interpretation.

But that is simply not true—as applied to Roman Britain or any other time or place in the ancient world. Geneticists, anthropologists, and historians who rely on DNA to study human migrations are well aware of the limitations of DNA analysis. At the same time, ancestry DNA tests are becoming ever cheaper and more popular, and misconceptions abound.

“We have written sources. We have archaeological sources. Now we have genetic sources, but no source speaks for itself.” says Patrick Geary, a historian at Princeton’s Institute for Advanced Study, who is using DNA to track barbarian invasions during the fall of the Roman empire. “Every kind of source must be interpreted. We are only at the beginning of how to properly interpret the genetic data.”

To start, much of the research into genetic variation (including much of what powers commercial DNA tests) is based on where people live now. One might consider a certain genetic signature typical of Italians based on who lives in Italy today. But how common a gene is in modern Italians may not reflect how common that gene was among Romans who invaded Britain 2,000 years ago. Over millennia, populations frequently moved and sometimes completely replaced the existing population in a given area.

Past migrations and invasions aren’t always evident in the DNA of modern people. In fact, the modern British population shows little genetic evidence of Roman, Viking, and Norman invasions—but this one data point is hardly enough to overturn the preponderance of historical evidence that shows these invasions did happen.

In recent years, researchers have turned to ancient DNA from burial sites to better understand ancient populations. Last year, a study of nine ancient Roman skeletons in Britain found a lot of similarity with British Celtic populations. One skeleton, though, showed much more affinity with modern Middle East populations.

This is a more direct picture of the past but it’s still an incomplete one. First, the number of bodies available to sample is often small. Second, the number of samples that yield DNA after hundreds or thousands of years are even smaller. And lastly, the amount of DNA you can get is usually a tiny portion of the genome. “You have to be very careful about what assumptions you bring into your study,” says Jennifer Raff, an anthropologist who studies ancient DNA at the University of Kansas. For example, a recent intriguing study of 90 Egyptian mummies showed they were more genetically similar to modern Middle Easterners than central Africans. But of course only the wealthy were mummified, so it’s not a complete picture of ancient Egypt.

Geary, the historian at the Institute for Advanced Study, is studying ancient DNA from cemeteries around present-day Lombardy in Italy. He is very careful about how he presents his work and avoids speculation in his talks. While his research has turned up two distinct groups of people, he told me he resists giving them names that identify one or the other as the “real Lombards”:

I was talking to one of our board of trustees at the Institute—a billionaire who has an interest in what we’re doing—and I said, “Well, we have this central northern population and this southern.” He said, “No, no, you can’t call them that. You’ve got to give them names. That’s how you’re going to get attention and funding.” But of course that’s exactly what we mustn’t do because then one falls into this ethnic discourse that we are trying to avoid.

Applying these labels—and maybe even the act of resisting labels—is a matter of historical interpretation. Genetic data is subject to interpretation like any kind of data. When something as trivial as a five-minute children’s video can inflame the culture wars, so will any genetics study that even touches on notions of race and ethnicity.