Wednesday, April 25, 2012

Retractions, decreasing funding, and other crises in science

In recent days, many people wrote about various crises in science. Some of these complaints may sound similar to what I have been saying for many years but there are substantial differences, too. In The New York Review of Books, Steven Weinberg wrote about
The Crisis of Big Science
My hat tip goes to David Simmons-Duffin (twitter). Weinberg begins with Rutherford who received a £70 grant from the Royal Society in London. He hired one postdoc, one undergraduate, and they just discovered the atomic nucleus. That was a rather financially efficient setup for science. ;-)

Science got bigger, however, and Weinberg explains that it's needed. His article, somewhat incoherently, switches from the social and historical questions to properties of the Standard Model, issues in astrophysics and cosmology, and other things. Weinberg tries to be as pessimistic as possible, acknowledging that the P.R. failures surrounding the Superconducting Supercollider have influenced his mood for decades.

While my idea about the percentage of GDP that people should spend on science is probably very close to Weinberg's estimates, I ended up with the impression that his article is another big P.R. bummer. The attacks on the manned spaceflights were still OK – we know that Weinberg considers this activity (which still induces a significant part of the people's curiosity and excitement) to be a complete waste of time and money.

But what I was really stunned by was the final paragraph. Weinberg's conclusion seems to be that there should be higher and more progressive taxes, "especially on investment income". Holy cow. What does this discussion on science funding have to do with musings about the tax rate or even the relative composition of the IRS revenue?

Science funding is just a small portion of the budget. Most of the public expenses have nothing to do with the scientific research which is why the scientific research and its funding shouldn't be saying anything about the size or sources of the bulk of the public budgets. The composition of the IRS revenue is a political and economic question and of course that when you analyze these issues rationally, you will find out that it's better for a country to have a flatter tax rate – regardless of the products and services that are being paid from the revenue. In democratic countries, voters still have the right to choose politicians according to their opinions about the right overall spending and some rough distribution of the revenue. They also have the right to overrule the conclusion of economics that the tax rates should be as flat as possible.

I know that Steven Weinberg and others believe that it's OK for them to be militant leftists and they may even believe that this extreme left-wing ideology is correlated with science. But it's not. And by showing his true thinking, Weinberg helps to reinforce the impression of the public – and people like me – that what he really wants is the Big Government while science is just one of the justifications for such a Big Government. It's too bad and be sure that if the likes of Weinberg won't avoid similar arrogant excursions into politics – demagogical political arguments justified by would-be scientific arguments that are completely bogus (you can't prove that the progressive taxation is better by listing the particles of the Standard Model) – be sure that the will of the public to pay money to the science will keep on decreasing and rightfully so.

While certain things have to be expensive, the amount of money wasted in the Big Science has become comparable to the money wasted by the Big Government as such. And that's too bad. I am pretty sure Rutherford and many others would be surprised what kind of bureaucracy and how big number of researchers is needed today to achieve a relatively modest progress. I don't think we should return to the era when science was researched by a dozen of people on the globe. But we should still carefully avoid the breathtakingly stupid propaganda that every dollar invested to science and its overhead, however low-quality and unimportant one, is a good investment. It is surely not.

This brings me to the second topic.


In The New York Times, Carl Zimmer wrote that
A Sharp Rise in Retractions Prompts Calls for Reform.
The article focuses on medical and biological research but it's meant to be much more general. Two years ago, Ferric Fang, an editor the journal Infection and Immunity, noticed the skyrocketing number of retracted papers since 2000. In the PubMed database of papers alone, the number increased from 3 in 2000 to 180 in 2009 despite the fact that the increase of the number of papers grew just by 44 percent or so.

Is that a catastrophe? Well, I am not sure whether this increase is proving a deterioration of the published papers. Don't get me wrong, I do believe that the average quality is dropping but the increase of the number of retractions – by orders of magnitude – is overstating the drop in quality. Most of the increase of the retractions is due to the fact that people began to like the concept of a retraction (of their own papers or of papers by someone else who may be urged to retract).

The article says that much of the increase occurs because of the heavy competition. Young scientists try to get a job and older ones try to remain relevant so they're sometimes fabricating or "improving" results or they're publishing things that are insufficiently verified, in order to increase the quantity of results, and so on. Competition is another thing that the extreme left-wing types love to assault. The Trouble With Physics written by a notorious left-wing crank dedicates several chapters to the "evil" competitive pressures that are suppressing "seers" (cranks) like the author himself.

Let me tell you something. Much like in other human activities, competition is vitally important in science as well. The system simply should try to allocate the resources efficiently. It should support the best researchers and encourage the worst ones to do something else. What does it mean to be a top researcher? The total value of the papers must matter. It is usually easier to obtain the same "total value" by writing a higher number of papers. But for the right geniuses, it's easier to publish a smaller number of groundbreaking papers and be sure that almost all competent "senior scientists" are certainly able to appreciate this kind of concentrated contribution. One doesn't need to "improve the life for geniuses" by simplifying the life for people who are not geniuses at all. These are different things; they're really the nearly opposite things.

As long as the value remains visibly nonzero, there's nothing wrong about a high number of papers. In some cases, people write too many papers which don't bring anything really new. In other cases, people don't do almost any research. Sometimes, however, people just write a reasonable number of papers that make sense. I think that one shouldn't care whether a scientist brings some total contributions in 5 papers or 500 papers during his career. But what's important is that the overall contributions – with the importance and validity being properly taken into account – should matter.

In fact, I don't find the number 180 of retracted PubMed papers per year to be excessively high. I am totally convinced that the number of wrong or at least worthless papers that are published in the discipline – or any major discipline – is vastly higher than 180 papers per year. And I even think that it was higher even well before the retractions became popular. For this reason, I think it is a healthy development that people are led to retract bad papers. I just hope that they're papers that are at least worse than the average ones. Otherwise the retractions make no sense – or they may even be counterproductive (e.g. a form of censorship) if the retracted papers are above (or near) the average according to their quality and if the real reasons leading to the retractions have nothing to do with the quality or validity.

What is needed to have a more solid body of published scientific research? We surely shouldn't try to "turn off" the mechanisms of competition because competition is a very good and creative mechanism. We should be just more sensitive about bad work, especially misconduct. When someone writes something really bad or something that's even been fabricated in order for the researchers to dramatically increase their status, they should be punished. Their prestige should go down, to say the least. In many cases, such a worrisome finding should be the end of their careers.

In some cases, the worst fraudsters should get at least a prison sentence. We're of course very far from this point. Instead, many of them receive The Hans Oeschger Medal from the European Geosciences Union. Holy cow. I am ashamed of being European – but frankly speaking, the Americans should be ashamed of being American once they look at the winner.

So the real problem I am seeing is that many people say that valid and important results are good and positively counted while the wrong and even dishonest research "does not matter". I couldn't disagree more. Wrong and especially fraudulent research (but even vacuous research) must reduce a scientist's status at least as much as the correct and honest research (and important research) improves the scientist's status. If such a punishment didn't exist, the percentage of the wrong and worthless papers – in which people just "try" to become important – would inevitably increase and that would be too bad because scientists need to stand on solid shoulders of the previous research. Lottery tickets can't be free, either. If they were free, everyone would be collecting these tickets all the time (writing mostly garbage papers), no one would do anything else, and the lottery company (=grant agencies, sponsors, governments) would be losing money.

Many left-wing people love to say that science has to be allowed to do experiments and it's important for scientists to err, and so on. Well, you may err as many times as you want while you're doing the research and it's indeed very important that you know that you can do so but once you publish a paper, you are correlating your good (or not-so-good) name as a scientist with the quality of the paper. People must consider such things when they evaluate you as a scientist. And I am actually totally convinced that the top scientists are doing such things and most of them have a perfectly reasonable viewpoint on other scientists that reflects both good and bad things that the other scientists have published (or said).

However, what I am really worried about is that while the top scientists are doing these things well, they are losing their political influence. I think that it's increasingly the case that the funding agencies – or even powers such as the media that have a lot of influence on funding agencíes - are increasingly often controlled by incompetent, superficial, or downright dishonest or corrupt folks. That's also why the scientific establishment is becoming increasingly inefficient. For it to prosper, the scientific establishment simply has to stay in the hands of the right people. In my opinion, the top people should have much more power in "dictating the work" to others who are just "workers of science". Too many people are trying to act as "independent geniuses" or "principal investigators" even though they're really not up to that job and they need a leadership of someone else.

Whether the number of retractions will be high or low is yet to be seen. Maybe the optimum amount of time one should spend with verifying his or her claims should be shorter than it is today and it doesn't hurt if some mistakes are ultimately found by the peers (who may do it more quickly than the author). But what's important is that the folks who publish valid and important research statistically end up better than those who are publishing mostly rubbish.

And that's the memo.

1 comment:

  1. We need more government funding of science, since government funding has been demonstrated to be a decent way of going about it. This should be paid for by whatever system of taxation society chooses, and given directly to scientific committees without political interference.

    Also, social science isn't really science, and if it deserves public funding at all, it certainly shouldn't be hidden in the science budget.

    Scientists shouldn't use their positions as scientists, as opposed to their positions as citizens, to tell everyone else how taxes should be raised. At the same time, politicians shouldn't tell scientists how scientific funding should be spent.