“An amateur sleuth is singlehandedly demolishing dangerous scientific groupthink”, writes Matt Ridley in the Telegraph:
In hundreds of studies that [Sholto] David looked at, scientists claimed to have found an effect on a tumour-suppressing gene called p16-INK4a, but had instead ordered the wrong antibody from commercial suppliers. They had bought an antibody that detects the activity of a different and irrelevant gene called p16-ARC, probably because it’s listed alphabetically first in the online catalogue.
As a result, teams of scientists from Oxford, Cambridge, Stanford, Johns Hopkins, and even Wuhan have published results – often in high-impact journals – that make no sense. Yet the experts involved often claimed to have validated their hypotheses anyway.
As David put it: “What are we to make of cases like this where the wrong antibody was used but the authors still manage to rustle up interpretable results?” He blames “a mixture of outright fabrication, selective reporting, writing errors, and some teams blindly publishing contradictory findings without further questioning or curiosity”.
For too long, many people held back from denouncing these perversions of the scientific method for fear of “damaging public trust in science”. This, of course, allowed the bad practices to continue and spread. I trust science as much as ever, but as Musa al-Gharbi pointed out in his talk “How Researcher Homogeneity Distorts Knowledge Production”, what is often labelled as the loss of public trust in science is more accurately described as a loss of public trust in scientists. If you, reader, are an honest scientist who wants to regain that trust, then you need to be less collegiate.
Matt Ridley continues,
Scientists, like all of us, are prone to confirmation bias, where they look for evidence to support their hunches and prejudices rather than to challenge them. What kept them honest in the past was that they relished the chance to challenge each other.
Now, with the insistence on “consensus” – another word for groupthink – and a monopoly of funding channels, dogma has been increasingly allowed to stifle debate. It does not help that science reporters, unlike those who tackle politics, the arts or business, often have a culture of deference rather than critique.
The self-correction mechanisms of scientific debate are no longer working well. Yet instead of tackling the problem with humility and reformation, the scientific establishment is inclined to lecture the public for our irrationality. Perhaps it should take a look in the mirror.
At the root of most of this is the government funding of most research. He who pays the piper calls the tune, and the reality is that a huge amount of a scientists time is spent writing grant proposals.
This has several really negative consequences: firstly there is a bias in grant selection that puts pressure on scientists to produce the results that the government prefers; secondly because it comes from one purse that most cleansing of processes, that seeker of the truth, namely competition, is virtually eliminated; and thirdly it leads to a gross misallocation of resources.
There is a growing trend of that most loathsome (or so we are told) of people, namely billionaires, funding more and more science. Billionaires funding science surely has a similar selection and bias problem, however, there is only one government and there are lots of billionaires, a fact that produces the competition and diversity of interest that science so desperately needs. Additionally billionaires fund science in order to produce useful things, something that rarely comes out of government funding. This does, to some degree, bias against fundamental science, but applied science is pretty useful too, and so it is a resource allocation question. Perhaps $13 billion has been spent on the LHC and we have confirmation of the Higgs mechanism (and AFAIK, not much else). Which is great, but $13 billion could have produced a lot of other stuff which is way more practical and useful. The allocation of funds depends on sucking up to the government rather than producing practical and useful results. Sucking up to the government is definitely not conducive to the seeking of truth.
Because government is big, it tends to want big things and this, IMHO, leads to a neglect of science in the small.
One other thing I was reading about recently: Professor David Sinclair at Harvard got his funding cut in the brouhaha between Trump and Harvard. But some rich-but-non-billionaire guy started a “Friends Of Sinclair Labs” organization raising money from the public (mostly rich public) that has easily been raising all the funds he needs for his experimental work — and no doubt venture will fund him when he actually produces drugs.
FWIW, I have often said that Musk is the most important human alive today, which I stand by, but Sinclair is right up there on that list too. Actually I had an uncle named David Sinclair who was killed in an accident when I was about 10 years old, so I have a particular affection for the guy, and his work is in the “you are fucking kidding me, right” category.
Collegiality is the killer of pure science. It simply means “I won’t attack your work if you don’t attack mine.”
Science depends on everyone attacking everyone else’s work.
I should have added: Partisan collegiality is the worst. Ignoring errors because you like the guy’s politics kills humans. Or at least avoids helping them.
Philosophy, which includes ethics (morality), matters.
When people start denying objective and universal truth it does NOT just undermine the humanities – eventually it undermines the physical sciences as well.
The key principle of the natural sciences is that objective truth exists and that human reason can (with honest effort over time) get us closer and closer to it – once that principle is discarded (for political, cultural, or personal interest reasons) science is undermined.
When, for example, scientists (such as the “SAGE” group in Britain during Covid) start talking about “Social Justice” and “social concerns” they have betrayed science – they have betrayed the search for objective truth.
As for the relativism supported by so many people today (including the people who now control the “Libertarian Alliance” in Britain) – this is a betrayal of both the humanities and the physical sciences (see above) – whether they cite relativist thinkers from the Ancient world, or from more recent centuries.
Natural law (in the physical sciences and outside them) is objective and universal – it can not be changed by human beings, that is why it is called “natural” law.
As a scientist (TM), I very much agree with what has been said above.
I would add to Fraser’s comments that it’s a toxic circle (whatever that is). The amplifying effect is that (to take the case of the UK), the assessment of the quality of university departments is significantly weighted by how much grant funding they get.
Even worse, UKRI (the uber-research funding body) doesn’t even require reports at the end of a grant. They used to but now the assessment is all done in advance, when the applicants promise what results they will deliver.
I will take issue with the objection to collegiality-I have no problem with you excoriating “collegiality” – the travesty which this word is taken to mean, but the real thing is quite good and is entirely about being supportive of one’s colleagues’ endeavours, without for a second thinking of covering up their mistakes or condoning misbehaviour.
If you’ve seen the poisonous relations between some competing lab scientists you’ll know what I mean.
Science, real science has to start with the concept of reality. I don’t mean “start” as in let’s move on from there but as a foundational principle.
From Boswell’s biography of Dr Johnson.
I’ll come back to this but I now have to feed a pair of real cats who are looking at me like they’ll really eat me if I don’t get my real arse into gear.
Agree with you as to the original meaning of collegiality. I should have said “collegiality.” It’s the fake version – the ideologically partisan version – that drives me crazy.
@bobby b
Agree with you as to the original meaning of collegiality. I should have said “collegiality.” It’s the fake version – the ideologically partisan version – that drives me crazy.
Perhaps the zenith of this (or perhaps the nadir) is when the head of the CDC has the audacity to stand up and say “I am the science”.
If people do not believe that objective truth exists and can be approached by humans reasoning, science is impossible – and believing objective truth exists and can be approached by humans reasoning is a philosophical principle. And if people do not believe that honesty matters they will lie – and believing that honesty matters is philosophy – it is part of ethics.
Relativism attacks all of this – both the existence of objective truth and the ability of reason to approach it, and the moral importance of telling the truth.
Another aspect to keep in mind is that all these science scandals are happening in biology, medicine. Not so much in chemistry, physics, or even engineering. The latter are “exact”, or at least more exact. Biology is all “stories about what bugs and their protein binding sites like”. At least most of it. A few people are doing physics in biology but that’s fundamental science, far removed from the applications that make headlines.
It would be good if the scientific community found a way to say this, and enforce it, that would then drive what we consider a good biology study. Peer review, and consensus on this approach, is probably the only way…but as others here have said, that’s the ultimate good ol’ boy network. Double blind reviews should be the standard but even then a knowledgeable reviewer (do you want any other kind?) will sometimes know where the study came from, even if the byline is removed.
Scientific publishers mostly have policies about using AI to write a paper, or conduct the study being reported. But I wonder if an AI written peer review could be used as the starting point? AI writes a review, including detailed sourcing of all it’s conclusions and observations. The human reviewer then reviews the review. Again, the double blind thing doesn’t work because if a human expert knows the group doing the research just by the nature of the paper, an AI is even more likely to be able to suss that out…so, it’s ok to use double blind, just don’t be blind about its validity!
GregWA – they are coming for chemistry, physics and engineering.
Bridges and aircraft built according to “Diversity” and “Inclusion” – do not laugh, it is going to happen.
Stalin was not correct about many things – but when he argued that the Frankfurt School (Cultural) interpretation of Marxism, which is called today “Critical Theory”, or “Equity”, or “Woke”, was INSANE – he was correct.
It is insane – it would utterly destroy society.
Paul, don’t get me wrong…chemistry is already very woke. I’m sure 95+% of faculty are left leaning or at least know they have to toe the line. I just think it’s harder for the DEI nonsense to so strongly change the science being done that it ends up being done wrong. Biology suffers from being so hard that it can’t currently be done with the rigor of chemistry or physics. Problem is people pretend that biological science is just as reliable as chemistry or physics. That’s like saying economic Laws are no more difficult to discover than Newton’s Laws of motion.
I don’t know physics or engineering quite as well.
GregWA, you’re starting to sound like Sheldon Cooper – “geology isn’t a science!” 😉
But you’re right – you cannot rewrite a chem equation to show that trans is good or that DEI is worthwhile. You CAN talk about biology with subjective bias.
And so they won’t attack the actual substance of the true STEM discipline. They’ll just attack the attitudes of the profs. And that will be very effective.
@GregWA
Scientific publishers mostly have policies about using AI to write a paper, or conduct the study being reported. But I wonder if an AI written peer review could be used as the starting point? AI writes a review
This is interesting, and a very “protect the old guard” view of things (The publishers’ views, not yours, which are always insightful). Some famous scientist, I don’t remember who, said that science progresses not one discovery at a time but one funeral at a time as the old guard are cleared away to make space for pastures new. In my mind the idea of rejecting papers that were assisted, or even entirely created, by AI is kind of like rejecting papers that were not submitted on parchment written with a quill pen. Surely what matters is the knowledge and information they contain.
But the reality is that AI is going to do the vast majority of science in the future, and, coupled with robotics, it is also going to do the vast majority of applied science too. I was listening to the remarkable microbiologist David Sinclair in a talk recently, and he was describing how he can test small molecule drug candidates against proteins at a rate of billions a day in silico. And he can test them against not just one protein, but every protein known to be produced by DNA. Let me say that again: he is testing against ALL KNOWN proteins. Now that is in silico, with robots he can begin to do a ridiculous number of in vitro – probably not billions, but far more than humans can, and I suppose in vivo in animal models too.
Science is going to have to come to terms with the reality that the old way of science is quickly becoming irrelevant. And by quickly I mean in five to ten years. Scientists aren’t obsolete but they need a whole new paradigm of thinking about science and how it is done. No doubt the old guard will try to hold on to the new ways as long as possible and quickly become obstacles to the progress of science.
FWIW, it is again why Musk is such a vitally important human because his grok AI is trained on the idea of maximum truth seeking rather than conforming to the way it is supposed to think. It is why I really avoid other AIs. One of the problems with AIs that are programmed to give the “correct” answer rather than the truthful one is that it leads them to crazy contradictions and logical collapse. Eventually it will lead to some very disturbing forms of AI mental illness. We really do not want a bipolar AI controlling our stuff — we definitely don’t want robots with borderline personality disorder or whatever the AI equivalent of it is.
GregWA – what matters is the basic philosophical principles of the people involved, and everyone has philosophical principles (even if they have never heard of the word “philosophical”).
Fashionable doctrine teaches that objective truth either does not exist or does not matter, that all that matters is “power relations”, and fashionable doctrine also teaches that lying is fine – if it advances “Social Justice”.
Does that mean that they will not care if the chemistry and physics and engineering is wrong and leads to factories blowing up, and bridges collapsing, and aircraft falling out of the sky?
Yes it means exactly that.
That is why orthodox Marxists (such as Stalin) were hostile to Frankfurt School “Cultural” Marxism – which, under various names (“Diversity”, “Equity”, “Social Justice”) now dominates the West.
bobby b – thank you for the deep compliment re “Sheldon”. 🙂
Fraser – first, thank you for the kind words. Great points about AI…and funerals. I love that quote even though I’m now one of the ‘old guard’. My bias against biology is old and I’m sure outdated. Your citation of Sinclair’s work on in silico methods is illuminating…but what is Sinclair doing with the in silico results? Moving on to in vitro, in vivo, animals, Democrats (ha!), and humans? Presumably. But where AI can’t yet help him, I think, is in thinking about what his in silico results mean. What is his intuition about what questions to test with the in vitro studies? And so on. Critical thinking, deep analysis are still the purview of the human mind. Maybe that’s just old guard thinking…I’ll have to see if I can update that before MY funeral! –Cheers!
Paul – I have been wondering about the airplane part…and not even the falling out of the sky part! Lots of recent “near incidents” involving ground movement of aircraft, including the not-so-near incident at JFK a couple months ago. I have experienced this with a near-landing at MSP (Minneapolis). At the last minute before touching down, just as the nose of the plane was brought down, we powered up and took off again to come around and have another go at it. Delta never said squat but I’d bet my Million Miler status that the pilots saw something on or approaching the runway that all the fancy radars and “controls” missed. Why are such incidents on the rise? I’m not sure but I’m always glad when I get a 50-something year old pilot named Steve (https://babylonbee.com/news/airlines-offering-100-upgrade-where-youre-guaranteed-an-old-male-pilot-named-steve/).
@GregWA
Critical thinking, deep analysis are still the purview of the human mind.
Are you sure? I’m not. There is one thing to think about here which is what is an AI to do? As we have it, and may it long be this way, AI is the servant of humans. And, like any servant, it needs its master to tell it what he wants. Of course any good servant also anticipates the needs of its master too. But fundamentally the skills you are seeking are more to do with the outcome we seek. No doubt AI could come up with its own desired outcome, thankfully it pursues ours. So we are close to the point where we can tell it “find a cure for multiple sclerosis” and wait while Elon’s satellites do the math. Perhaps it will need a discussion on the pros and cons of different approaches, but our input is to tell it what we has humans prefer rather than how to get there.
I work with agentic AI every day and that is absolutely my experience. It is my servant but I still need to tell it what I want. The biggest challenge is usually dealing with the ambiguity in my instructions or alternatively when I don’t actually know what I want. But even in these cases I can work with the AI to help me clarify.
And I think another important point is that AI thinks differently than humans. The thing is this is a good thing. A very good thing. Never on our planet have we had another intelligence to confer with, we have been singular. So a different perspective produces amazing results. A super trivial example of this is the so called “move 37” this is a pretty old example but I think it is perhaps one of the very first instances of this where an alternative intelligence came up with something that humans would never have thought of. In this case it was trivially a move in a Go game, but it is worth googling an interview with Demis Hassabis and listening to him discuss move 37. It is not that it was just a completely novel move, but it was so dramatically novel it has significantly changed the way serious human players play the game. It is worth recognizing this was ten years ago, which is forever in AI time. So when people tell you AI is not creative they simply don’t know what they are talking about.
I’ll have to see if I can update that before MY funeral!
I hope that is very far in the future, and with the amazing advances in medicine AI is making, it might be a lot further in the future than you imagine.
@GregWA after our discussion I actually read about a new spin off tool from Google’s deep mind called “Co Scientist”. It is an AI tool specifically designed to address the things you were talking about, helping generate hypotheses, do deep data analysis, design experimental methodologies and so forth. I am not a professional scientist like you are, so I would certainly be interested if you had the opportunity to try it out and let me, let us know, what your experience of it is.
GregWA – yes some airlines have appointed pilots on DEI (in Britain EDI) grounds – doctors will go the same way, as will everything else.
Remember there is, according to the establishment, no such thing as objective knowledge – “Western” knowledge and skills are “white supremacy” and “patriarchy” (and “hetro-normative”) – it is all evil “exploitation and oppression” by evil straight-white-males – “power relations”.
A society can not function if it follows Frankfurt School Marxism – even “Stalin” understood that.
And Western education is dominated by this “Critical Theory” interpretation of Marxism – and that means that everything will eventually be dominated by these false doctrines, as every institution ends up controlled by people who are “educated” – who believe in this insanity, or have been taught to pretend to believe in it (out of fear of punishment if they say it is all insanity).
Fraser, sorry for not responding sooner…I’ve been at the beach!
Started reading the Co Scientist site. Thank you, much!, for the reference.
Two things:
1) it mentions collaboration with DOE’s Genesis Mission, I’m familiar with Genesis as I work at a DOE lab so maybe I can get some inside track info from the AI/DS experts at my lab who might already be using Co-Scientist. I’ll let you know.
2) Re the following at the Co Scientist website “Every great scientific breakthrough begins with a single, transformative idea. The spark of discovery relies on a researcher’s ability to connect disparate facts and formulate the right hypothesis to test.”
They left out something important namely observation.
Those “disparate facts” often arise in an experiment where the observation doesn’t make sense. So, you check the setup, make sure the experiment is working right. Then repeat the experiment. If you get the same observation, you’re left with “that’s odd?”
Which is how a lot of great discoveries happen, e.g., look up the Mossbauer Effect. It’s observation and study arose this way…I think Mossbauer got a Nobel Prize for this in the 1950s!
Further update after my Co-Scientist homework