Contact Me
Menu

Chasing Shadows: What virology never controlled for

Scroll to top
Scroll to top

Chasing Shadows

What virology never controlled for

Originally published November 18, 2024.

Start with the boring part, because the boring part is where this whole thing lives.

A control experiment is what tells you your result came from the thing you think it came from. You run the procedure twice. Once with the variable, once without. If both runs give you the same outcome, the variable was never doing the work. Every science student learns this in the first month. Most of them find it tedious. It is the single most important idea in experimental method.

Hold that thought.

In 1954, John Enders and Thomas Peebles published a paper on measles that became the template for how viruses get found. They took throat washings from sick patients, added them to cultures of human and monkey kidney cells, and watched the cells swell, distort, and break apart. Virologists call this the cytopathic effect. Enders concluded the effect was associated with the measles agent.

To their credit, they flagged the problem themselves. They noted in the paper that similar changes might come from other agents present in the monkey kidney tissue, or from factors nobody had identified. Then they concluded they were looking at measles anyway.

Here is what they did not do. They did not run the procedure with no patient sample added. They did not run it with samples from healthy people. They did not run it with samples from people who were sick with something else. Any one of those three would have answered the obvious question, which is whether the cells were dying from something in the specimen or from the treatment they were receiving. And the treatment was not gentle. Cells in culture get starved of nutrients, passaged repeatedly, and dosed with antibiotics and antifungals, some of which are toxic to kidney tissue specifically.

Seventy years later, that gap has not been filled. It has been inherited.

Look at what happened in 2020. The teams claiming to isolate SARS-CoV-2 used monkey kidney cells, added unpurified material from patient swabs, stressed the cultures, and reported cytopathic effects. When people asked what the controls looked like, the answers were thin. The CDC responded to a records request about the control groups in its own isolation paper with thirty-seven pages that included internal emails about which images were pretty enough to publish, and nothing describing how the mock experiment was set up. New Zealand’s ESR described its negative control as a flask run under the same conditions using infection media only. That is a check for contamination in the pipeline. It is not a comparison against another human specimen.

One group was more forthcoming, probably by accident. Peng Zhou’s team in Wuhan was asked by email about their methods and disclosed that the experimental cultures received double the antibiotic concentration of the controls. They also disclosed that cytopathic effects appeared in one well out of twenty-four. So the experimental and control arms differed in a known cell toxin, and the positive result showed up at a rate you would want to look at twice before naming a pandemic after it.

I want to be careful here, because this is where these arguments usually overreach. None of that proves the cells were dying from the antibiotics. It proves nobody checked. Those are different claims, and the second one is the one I can defend.

Then there is the word itself.

In ordinary English, to isolate something is to separate it from everything else. In virology, isolation means something else entirely. It means you took a sample, put it on cells, and saw the cells change. Nothing has been separated from anything. The field knows this. Vincent Racaniello, a working virologist with no sympathy for any of the above, has described how the terminology gets absorbed by training rather than defined, and how a researcher may pick up a usage without ever checking whether it is right. He is candid about it. He does not seem troubled by it.

I am troubled by it, and not for technical reasons. When a health ministry tells the public that a virus has been isolated, ordinary people hear the ordinary meaning. They picture a particle, separated, examined. That is not what happened, and the gap between what was said and what was heard is not the public’s fault. This is the part of the story I keep coming back to, because it is not really a dispute about cell culture. It is a dispute about whether words in public health communication should mean what people think they mean.

The same assumption runs down through the genomes.

The SARS-CoV-2 sequence was assembled by software from millions of short fragments found in one patient’s lung washings. To assemble them, the program needed a reference to align against. Fan Wu’s team used two: a SARS sequence called Tor2 and a bat sequence called SL-CoVZC45. Follow those back and you find the same method again. SL-CoVZC45 was built in 2018 by designing PCR primers from sequences already deposited in GenBank. The SARS genome from 2003 was assembled against a mouse hepatitis sequence used as a backbone. Keep going and you reach the first published coronavirus genome, avian infectious bronchitis, in 1987, which traces to filtration experiments on sick chickens in the 1930s.

At no point along that chain did anyone start over from scratch. Each sequence was built by reference to the one before it. That is not a scandal by itself. That is how reference genomics works, and it works well when the provenance of the starting material has been established independently. The question worth asking is whether it ever was here, and whether anyone has gone back to check.

You may have seen the freedom of information campaign that came out of this, where institutions in dozens of countries were asked to produce records of purifying the virus directly from a patient. None of them produced such a record. I think this result is real and I also think it is routinely oversold, including by people I agree with. A records response tells you what an institution holds. It is not a finding, and treating it as one hands critics the easiest possible rebuttal. What it does show is narrower and still worth something: the specific procedure most people assume was performed, was not performed, at least not by the bodies that were asked.

None of this is new to the history of medicine. We have misread patterns before, and the misreadings look exactly like this from the inside.

Sailors on long voyages fell ill one after another, and physicians saw contagion. It was scurvy, and the shared factor was the diet, not the proximity. Pellagra spread through the American South in a pattern so convincing that the public health establishment spent years hunting the germ. Joseph Goldberger showed it was niacin deficiency, fed infected material to volunteers who did not get sick, and was largely disbelieved for the rest of his career. Beriberi ran the same course in Asia and turned out to be thiamine. In each case the pattern was real, the observation was competent, and the interpretation was wrong, because the mind supplies an agent before it supplies a deficiency.

Thomas Kuhn described how this persists. A paradigm does not fall when the evidence against it accumulates. It falls when the people invested in it retire. In the meantime the framework absorbs anomalies, and the absorbing feels like rigor to everyone doing it.

I don’t think the people in this field are lying. That story is too easy and it lets everyone off the hook. What I think is that a generation trained inside a method inherits the method’s blind spots along with its techniques, and that the blind spot is protected by everything downstream of it. Careers, grants, diagnostic industries, and public health institutions were all built on top of a set of experiments that were never controlled the way an undergraduate is taught to control them. Nobody has to conspire for that to hold. It holds because opening it would cost more than any individual can afford.

So I am not going to tell you what causes illness. I don’t know, and anyone who tells you they have it fully worked out is selling something, including the people selling it from my side of the fence.

What I am telling you is narrower and I think harder to dismiss. The control experiments that would settle a foundational question in this field are cheap, obvious, and have not been published. They were flagged as necessary by the researcher who set the template in 1954. Asking for them is not fringe. It is the most ordinary request in science, and the fact that it now reads as radical tells you more about the state of the conversation than about the request.

Share This Post

More To Explore

Chasing Shadows: What virology never controlled for

Chasing Shadows

What virology never controlled for

Originally published November 18, 2024.

Start with the boring part, because the boring part is where this whole thing lives.

A control experiment is what tells you your result came from the thing you think it came from. You run the procedure twice. Once with the variable, once without. If both runs give you the same outcome, the variable was never doing the work. Every science student learns this in the first month. Most of them find it tedious. It is the single most important idea in experimental method.

Hold that thought.

In 1954, John Enders and Thomas Peebles published a paper on measles that became the template for how viruses get found. They took throat washings from sick patients, added them to cultures of human and monkey kidney cells, and watched the cells swell, distort, and break apart. Virologists call this the cytopathic effect. Enders concluded the effect was associated with the measles agent.

To their credit, they flagged the problem themselves. They noted in the paper that similar changes might come from other agents present in the monkey kidney tissue, or from factors nobody had identified. Then they concluded they were looking at measles anyway.

Here is what they did not do. They did not run the procedure with no patient sample added. They did not run it with samples from healthy people. They did not run it with samples from people who were sick with something else. Any one of those three would have answered the obvious question, which is whether the cells were dying from something in the specimen or from the treatment they were receiving. And the treatment was not gentle. Cells in culture get starved of nutrients, passaged repeatedly, and dosed with antibiotics and antifungals, some of which are toxic to kidney tissue specifically.

Seventy years later, that gap has not been filled. It has been inherited.

Look at what happened in 2020. The teams claiming to isolate SARS-CoV-2 used monkey kidney cells, added unpurified material from patient swabs, stressed the cultures, and reported cytopathic effects. When people asked what the controls looked like, the answers were thin. The CDC responded to a records request about the control groups in its own isolation paper with thirty-seven pages that included internal emails about which images were pretty enough to publish, and nothing describing how the mock experiment was set up. New Zealand’s ESR described its negative control as a flask run under the same conditions using infection media only. That is a check for contamination in the pipeline. It is not a comparison against another human specimen.

One group was more forthcoming, probably by accident. Peng Zhou’s team in Wuhan was asked by email about their methods and disclosed that the experimental cultures received double the antibiotic concentration of the controls. They also disclosed that cytopathic effects appeared in one well out of twenty-four. So the experimental and control arms differed in a known cell toxin, and the positive result showed up at a rate you would want to look at twice before naming a pandemic after it.

I want to be careful here, because this is where these arguments usually overreach. None of that proves the cells were dying from the antibiotics. It proves nobody checked. Those are different claims, and the second one is the one I can defend.

Then there is the word itself.

In ordinary English, to isolate something is to separate it from everything else. In virology, isolation means something else entirely. It means you took a sample, put it on cells, and saw the cells change. Nothing has been separated from anything. The field knows this. Vincent Racaniello, a working virologist with no sympathy for any of the above, has described how the terminology gets absorbed by training rather than defined, and how a researcher may pick up a usage without ever checking whether it is right. He is candid about it. He does not seem troubled by it.

I am troubled by it, and not for technical reasons. When a health ministry tells the public that a virus has been isolated, ordinary people hear the ordinary meaning. They picture a particle, separated, examined. That is not what happened, and the gap between what was said and what was heard is not the public’s fault. This is the part of the story I keep coming back to, because it is not really a dispute about cell culture. It is a dispute about whether words in public health communication should mean what people think they mean.

The same assumption runs down through the genomes.

The SARS-CoV-2 sequence was assembled by software from millions of short fragments found in one patient’s lung washings. To assemble them, the program needed a reference to align against. Fan Wu’s team used two: a SARS sequence called Tor2 and a bat sequence called SL-CoVZC45. Follow those back and you find the same method again. SL-CoVZC45 was built in 2018 by designing PCR primers from sequences already deposited in GenBank. The SARS genome from 2003 was assembled against a mouse hepatitis sequence used as a backbone. Keep going and you reach the first published coronavirus genome, avian infectious bronchitis, in 1987, which traces to filtration experiments on sick chickens in the 1930s.

At no point along that chain did anyone start over from scratch. Each sequence was built by reference to the one before it. That is not a scandal by itself. That is how reference genomics works, and it works well when the provenance of the starting material has been established independently. The question worth asking is whether it ever was here, and whether anyone has gone back to check.

You may have seen the freedom of information campaign that came out of this, where institutions in dozens of countries were asked to produce records of purifying the virus directly from a patient. None of them produced such a record. I think this result is real and I also think it is routinely oversold, including by people I agree with. A records response tells you what an institution holds. It is not a finding, and treating it as one hands critics the easiest possible rebuttal. What it does show is narrower and still worth something: the specific procedure most people assume was performed, was not performed, at least not by the bodies that were asked.

None of this is new to the history of medicine. We have misread patterns before, and the misreadings look exactly like this from the inside.

Sailors on long voyages fell ill one after another, and physicians saw contagion. It was scurvy, and the shared factor was the diet, not the proximity. Pellagra spread through the American South in a pattern so convincing that the public health establishment spent years hunting the germ. Joseph Goldberger showed it was niacin deficiency, fed infected material to volunteers who did not get sick, and was largely disbelieved for the rest of his career. Beriberi ran the same course in Asia and turned out to be thiamine. In each case the pattern was real, the observation was competent, and the interpretation was wrong, because the mind supplies an agent before it supplies a deficiency.

Thomas Kuhn described how this persists. A paradigm does not fall when the evidence against it accumulates. It falls when the people invested in it retire. In the meantime the framework absorbs anomalies, and the absorbing feels like rigor to everyone doing it.

I don’t think the people in this field are lying. That story is too easy and it lets everyone off the hook. What I think is that a generation trained inside a method inherits the method’s blind spots along with its techniques, and that the blind spot is protected by everything downstream of it. Careers, grants, diagnostic industries, and public health institutions were all built on top of a set of experiments that were never controlled the way an undergraduate is taught to control them. Nobody has to conspire for that to hold. It holds because opening it would cost more than any individual can afford.

So I am not going to tell you what causes illness. I don’t know, and anyone who tells you they have it fully worked out is selling something, including the people selling it from my side of the fence.

What I am telling you is narrower and I think harder to dismiss. The control experiments that would settle a foundational question in this field are cheap, obvious, and have not been published. They were flagged as necessary by the researcher who set the template in 1954. Asking for them is not fringe. It is the most ordinary request in science, and the fact that it now reads as radical tells you more about the state of the conversation than about the request.

Share This Post

More To Explore

Contact Me