Friday, September 30, 2016

The Real Sins of Scientists – More Pervasive than Fraud.

Fraudulent scientists seem to be everywhere these days. In recent years we have been regaled with cheats trying to foist upon us that rabbit blood can be turned into an AIDS vaccine, eating meat makes people more selfish and that transistors can be made out of virtually anything. Normal folk don’t know where to turn, so they doubt everything: evolution, climate change, vaccination. What is happening? Has science become propaganda from PhDs perverted by the search for research money and prizes?

The popular ideal image of us scientists is as disciples of the pursuit of knowledge.  With scant regard for the trivialities of life, we refuse bodily pleasure, food and sleep in our endless search for the ultimate truths of life, the universe and everything, scrupulously obeying the doctrine of the scientific method as we solemnly unroll the red carpet of transformational discovery. Little wonder then, that, whereas the uncovering of lawyerly or political fraud is met by  knowing snickering, each new revelation of science misconduct is considered tantamount to apostasy.

Serious scientific fraud - the fabrication of some high-impact but plausible new phenomenon - propels the perpetrator ephemerally into the academic stratosphere, while misleading large numbers of fellow researchers and misdirecting precious resources. But serious fraud in science is relatively rare, if only because the perpetrators, if not delusional, know they are likely to eventually be exposed by curious colleagues.

Serious fraud must, of course, be unearthed and punished, but my contention here is that is the least of science’s problems. You see, we scientists commit many sins, all of which lead to some sizeable proportion of our published work being at least partially misleading or wrong. These sins, which do not involve fraudulent deception, are far more widespread and more damaging to scientific progress. 

Let’s delight with a troll through seven deadly sins of scientists. We start with  incompetence and   ignorance. Our hypotheses may be balderdash, logically inconsistent. Our work may be ‘shoddy’ or ‘sloppy’; we may not perform  experiments that actually test our hypotheses, failing to test alternative explanations, and not knowing to perform elementary “control“ experiments. Our computer programs may contain critical errors. We may not estimate the statistical errors in our data. We may look at data and draw completely the wrong conclusions because we don’t know the underlying principles that govern the phenomenon under scrutiny. We may write our papers as if a logical sequence of experiments had been done when in fact we randomly tried things then assembled them into something that makes a pretty story.
We can also be lazy.  We may only do one or two quickie experiments, nowhere near enough to justify the grandiose conclusions we then draw, and hope the reviewers and editor of our papers are themselves too lazy or busy to read our manuscript properly. We may not even bother to properly search the literature to find out who has done anything related to our study. We may take the path of least resistance, that of expediency, to spin a story aligned with our vision.
Then there is illiteracy. We may be unable to describe our findings in a way that anybody else can possibly follow; we may omit steps in our argument, and our writing may be grammatically awful, leaving even qualified readers flailing.
True to our nerdy stereotype, we are often myopic. Our publications may deliberately ignore closely related but highly pertinent findings of others, concentrating only on our own past achievements, such that we do not put our work into context.  Citations made to others that we do include are to papers we have not even read.
We are also self-aggrandizing. In print and in person, and especially in grant proposals, we puff up the importance of our work and castigate other, legitimate studies. I may have been cited 20,000 times, but of course it should have been 200,000!
Great scientists can be highly intuitive, but this intuition also blinds us all. Many Nobel laureates have suffered from this. Take, for example, my ‘academic grandfather’*, Linus Pauling, arguably the greatest chemist of the last century. He spent his last decades misleading humanity by trumpeting unsubstantiated ideas about the health benefits of mega-doses of Vitamin C.

You see, Pauling, in his later years, fell victim to that ubiquitous scientist’s plague:  that of  wishful thinking. This arises naturally from the initiating, creative act in science, in which various half-formed ideas shape into a concept to which we cling and may base our careers, fomenting long-held desires, and prejudices. We believe in something, a beautiful process or an imagined principle. So, blinded by our belief, we may see a trend in our data that is not really there, or a small peak in a spectrum that is really just noise. We may remove that lone, recalcitrant data point that doesn't fit our model – that’s not fraud because we really believe the data point can’t be right. The temptation to airbrush data is irresistible.  Lets add a calibration factor, fudge factor, cosmological constant. We smooth, filter and transform data onto scales that make them look more accurate. Some run an experiment ten times until they get the result they want then publish only that one. We may refuse to give access to our raw data to others – after all, we haven’t finished analyzing them ourselves and, anyway, others would misuse them.
So, we scientists are ignorant, incompetent, illiterate, lazy, myopic, self-aggrandizing wishful thinkers. Each of these seven sins has the same effect as outright fraud – wrong results, erroneous interpretations, false conclusions. So shoddy, dubious science is everywhere, leading to a large proportion of submitted papers being rejected after anonymous peer review, and a fair proportion of manuscripts that do manage to sidle past peer review being still wrong. In my, of course unbiased, opinion, about half the interesting papers in my field published in the top journals such as Nature and Science, are basically wrong – they may be brilliant,  thought-provoking, beautiful and even inspirational, but they are still wrong.

Every Wednesday my lab holds a Journal Club, in which we select one or two papers to read, and we try to understand what was done, its significance, and its validity. Sometimes we leave the room exalted by a timely and impactful piece of research. But often, when we try to ascertain whether the main conclusions of the authors are justified by the data presented, we regretfully must conclude that the answer is “no”, and occasionally we go ballistic, especially me.  A while ago I had one of those ballistic days. We read a published paper on the computational design of drugs to overcome antibacterial resistance, and concluded that every one of the main conclusions was wrong. The paper was total pigswill. If anyone reads this paper and starts a program of drug design based on it they will have been sadly misled. [Naturally, though, that it is inconceivable that anyone would hold such a subversive meeting criticizing our own work. Inconceivable (ahem).]

Why then, is there so much fluff and junk out there? Well, unlike other professional pursuits, scientific research tackles the unknown.  This makes it inherently very difficult to know which questions to ask and how to go about things. Also, we scientists are condemned to membership of a certain species of animal endowed with primitive, instinctual, jealous and lustful traits. So each new problem will have each of us looking at it with our own biases, framed by our own imperfect training and experience. So it’s hardly surprising that there can be a lot of trash to wade through before an advance can be solidified. 

So, what to do about it all? Well, the world could try a science detox, doing without science completely, but then there will be no cures for cancer, no saving the environment, no endless supply of energy, no technological terrorist foiling. Another option is to keep doing what the authorities are concentrating on now; fraud detection, witch hunting, setting up Offices of Scientific Integrity and Research Integrity and the like.  But that is no panacea. You see, the seven scientific sins are juxtaposed by seven virtues: curiosity, intelligence, vision, drive, rigour, integrity and insight, virtues propelled by appreciation of the beauty of truth. The virtues win out in the end.

*academic grandfather: the adviser of my adviser, Martin Karplus. Pauling has hundreds of such grandkids..