I have to wonder, how much of an impact this paper would have made in pre-Twitter science?
Your opening premise in Science Fictions - that science is a 'Social Construct' - initially felt overstated, if not alarmist when I first read it.
Scientific discoveries should be self-evident, no? Wouldn’t all extraordinary claims naturally undergo rigorous scrutiny so that their truthiness* would be apparent? Doesn’t the process of science, as Carl Sagan noted “…with the most rigorous, skeptical scrutiny of all ideas, sifts the wheat from the chaff”? Why would scientific claims need marketing and sales departments?
Turns out, you were right.
*Fascinating that substack didn't spellcheck me for truthiness. Well done, Colbert.
How do 3 independent reviewers on a Nature manuscript all miss that the control group is the only one not CRISPR-ed? Normally I like reviewer anonymity, but this is dereliction.
The reviews are available online with the paper - one of the reviewers was unconvinced by the data, and made that clear, but the authors argued against that review without doing more experiments, and the editor accepted it.
I wonder how rare such peer review failures really are.
There's a peer reviewed paper about COVID in Nature Scientific Reports (which is, I'm told, different and not as prestigious as Nature itself), in which the very first sentence of the paper is factually wrong. It's a claim about Swedish COVID mortality and you can see it's wrong just by using OurWorldInData. It was wrong for the entire peer review period too so there's no publication date issue there.
The same team (ICL epidemiology) got another paper published in Nature that was built on what was very obviously a circular argument. They even admitted in the paper that their claims weren't about reality but that paper was hugely impactful, made the news, and even has thousands of citations. Eventually someone got a response published but only like half a year later.
When you see mistakes like those get through you've got to wonder what's going on. Laymen shouldn't be able to spot basic errors like that with almost zero effort.
I'm aware of the problems. I'd still have thought (perhaps naively) that a super surprising conclusion like this would get at least 1 of the 3 reviewers to scrutinize the method section.
I think you've got a typo in the following sentence. "...showed that synonymous mutations are far less likely to be linked to a huge range of human traits and diseases (nearly 5,000 of them) than mutations that actually alter the genetic sequence." I think you meant to say "... than mutations that actually alter the AMINO ACID sequence" (ie, that the amino acid is actually altered in the protein). All of the mutations of course are actually altering the genetic sequence.
One way that I have heard of synonymous mutations actually leading to deleterious/advantageous effects rather than to neutral ones is if some synonymous transfer RNA is much rarer than another (for example, if UUC transfer RNA was much rarer than UUU transfer RNA). If that was true, switching from a DNA sequence of UUU to a sequence of UUC would result in protein assembly being slowed down dramatically; and, of course, vice versa, switching from UUC to UUU, would lead to assembly of that protein proceeding much more rapidly.
(From a non-scientist, just a fan of learning about science.) I see a poetic parallel between 1) mutations and 2) theories that contradict established science and promise to revolutionize our understanding. Similar to how the vast majority of mutations in this study are “significantly deleterious” and a small percentage are “significantly beneficial,” I would not doubt that the vast majority of flashy “paradigm-shifting” theories are significantly wrong and only a handful are of any benefit.
The competition assays were in test tubes, not Petri dishes. (See section "En masse competitions in YPD" in the Methods.)
Quite right - now fixed! I was confusing it for the CRISPR part, which was in Petri dishes. Thank you!
*megabases, not megabytes!
D'oh! Brain switched off while typing. Thanks for this!
I have to wonder, how much of an impact this paper would have made in pre-Twitter science?
Your opening premise in Science Fictions - that science is a 'Social Construct' - initially felt overstated, if not alarmist when I first read it.
Scientific discoveries should be self-evident, no? Wouldn’t all extraordinary claims naturally undergo rigorous scrutiny so that their truthiness* would be apparent? Doesn’t the process of science, as Carl Sagan noted “…with the most rigorous, skeptical scrutiny of all ideas, sifts the wheat from the chaff”? Why would scientific claims need marketing and sales departments?
Turns out, you were right.
*Fascinating that substack didn't spellcheck me for truthiness. Well done, Colbert.
How do 3 independent reviewers on a Nature manuscript all miss that the control group is the only one not CRISPR-ed? Normally I like reviewer anonymity, but this is dereliction.
The reviews are available online with the paper - one of the reviewers was unconvinced by the data, and made that clear, but the authors argued against that review without doing more experiments, and the editor accepted it.
I wonder how rare such peer review failures really are.
There's a peer reviewed paper about COVID in Nature Scientific Reports (which is, I'm told, different and not as prestigious as Nature itself), in which the very first sentence of the paper is factually wrong. It's a claim about Swedish COVID mortality and you can see it's wrong just by using OurWorldInData. It was wrong for the entire peer review period too so there's no publication date issue there.
The same team (ICL epidemiology) got another paper published in Nature that was built on what was very obviously a circular argument. They even admitted in the paper that their claims weren't about reality but that paper was hugely impactful, made the news, and even has thousands of citations. Eventually someone got a response published but only like half a year later.
When you see mistakes like those get through you've got to wonder what's going on. Laymen shouldn't be able to spot basic errors like that with almost zero effort.
I'm aware of the problems. I'd still have thought (perhaps naively) that a super surprising conclusion like this would get at least 1 of the 3 reviewers to scrutinize the method section.
I think you've got a typo in the following sentence. "...showed that synonymous mutations are far less likely to be linked to a huge range of human traits and diseases (nearly 5,000 of them) than mutations that actually alter the genetic sequence." I think you meant to say "... than mutations that actually alter the AMINO ACID sequence" (ie, that the amino acid is actually altered in the protein). All of the mutations of course are actually altering the genetic sequence.
One way that I have heard of synonymous mutations actually leading to deleterious/advantageous effects rather than to neutral ones is if some synonymous transfer RNA is much rarer than another (for example, if UUC transfer RNA was much rarer than UUU transfer RNA). If that was true, switching from a DNA sequence of UUU to a sequence of UUC would result in protein assembly being slowed down dramatically; and, of course, vice versa, switching from UUC to UUU, would lead to assembly of that protein proceeding much more rapidly.
And you haven't even touched on the ludicrous "simulations" performed in the last figure... So amateur
(From a non-scientist, just a fan of learning about science.) I see a poetic parallel between 1) mutations and 2) theories that contradict established science and promise to revolutionize our understanding. Similar to how the vast majority of mutations in this study are “significantly deleterious” and a small percentage are “significantly beneficial,” I would not doubt that the vast majority of flashy “paradigm-shifting” theories are significantly wrong and only a handful are of any benefit.
Thanks for the piece. FYI UK copyright law permits the quotation of a work for the purposes of criticism and review. https://www.copyrightuser.org/understand/exceptions/quotation/ is a good guide to this.
So far as I'm aware this even applies to material published in nature!