The Death of Science

The institution of science has been a failure for a long time.

Has science just gone off the rails, and all we need to do is find our way back to real science?

Or should we accept that science is inherently limited for deeper reasons, and move away entirely from putting science as we know it on a pedestal? In other words, change our priors, change our presuppositions?

I suppose it’s both.

But before doing anything, we need to understand the problem as it presents itself. I have touched upon different aspects of the issue, such as in

• When Common Sense Trumps Science,

• We Are at a Metaphysical Nexus, or

• The Libet Experiments and How to Maximize Free Will.

One of the best books to read on what’s wrong not only with science-worship, but with science as practiced today and the mindset it spawned, is Iain McGilchrist’s The Matter With Things.

In chapter 13, he gives a jaw-dropping account of just how bad it is on a purely institutional level, even without getting into the fundamental philosophical issues. This chapter quite shocked me, to be honest, even though I already knew a fair bit of it and had already thought about some of these things.

In this post, I just want to give you some of the highlights for your convenience:¹

1. Due to specialization, every scientist takes almost every scientific “result” except the tiniest area of his expertise purely on authority, without having looked into it in any way. This includes results from his own field, and even his own subfield. In other words, to a large degree we have to just trust that most science is not built on falsehoods upon falsehoods, but we often can’t really know.

2. The replication crisis: there are many studies on this. For example, according to a survey published in Nature, a whopping 70% of researchers have tried and failed to reproduce someone else’s experiment. And yet, less than 20% said they had ever been contacted by another researcher who failed to reproduce their results. Ergo: the replication crisis is hidden, because failed replications are simply not reported, not even privately. And the replication crisis not only affects “soft” sciences like psychology, but even such fundamental things as the measurement of physical constants!

3. The most prestigious scientific journals are, in a way, the most affected: for example, so-called “scientific breakthroughs” reported in these journals are often nothing of the sort, and later get debunked in smaller journals, without anybody noticing it. The result is, as some researchers have shown, that studies in the high-impact journals are less likely to give an accurate estimate of effect size than those in the low-impact journals.

4. In his famous paper, “Why most published research findings are false,” John Ioannidis observes that the hotter a scientific field (the more scientific teams are involved), the less likely the research findings are to be true. Researchers tend to tweak their statistical voodoo until they get the desired results.

5. According to a survey published in BMJ, “one in seven UK-based scientists or doctors has witnessed colleagues intentionally altering or fabricating data during their research for the purposes of publication.” Junior researchers are encouraged to keep their suspicions and observations of fraud to themselves.

6. It is estimated that only 20% of cited papers have actually been read!

7. A whole lot of papers cited in studies are entirely irrelevant and just cited for the sake of it—and if not, they are often used to make a point that the cited paper actually contradicts!

8. The profit-oriented nature of open-access journals and even scientific conferences means that they often accept anything if you pay the bill. Hoaxes are legion. One guy used the autocomplete function of his iPhone to compose a completely nonsensical article, using a fake name, and submitted it to the International Conference on Atomic and Nuclear Physics in Atlanta. It was accepted within three hours, with a request for payment of a $1,099 registration fee!

9. Fabrication, plagiarism etc. are wide-spread. Just how bad the safeguards are can be illustrated by one spectacular case, where renowned science publisher Springer Verlag had to withdraw 120 papers from its publications because they were found to be gibberish created by computer programs that simulate science writing. (Perhaps it worked because today’s “science writing” is gibberish itself?)

10. Peer review I: peer review, often used as a magic wand to proclaim gospel truths by the High Priests of TheScience, is a joke. Just one revealing example is a hoax where someone submitted to 304 journals a paper whose obviously flawed method “should have been rejected by a high-school chemistry student” and whose conclusions were meaningless. Only 36 of these 304 generated review comments that at least detected some of the flaws. But even of those 36, 16 were then accepted, despite a damning review comment!

11. Peer review II: PR is, frankly, stupid even on principle, and there’s hardly any evidence it works. According to Richard Smith, former editor of the BMJ, peer review is ”a flawed process, full of easily identified defects with little evidence that it works. Nevertheless … scientists and editors have a continuing belief in peer review. How odd that science should be rooted in belief.” Smith also notes that “Robbie Fox, the great twentieth-century editor of the Lancet, joked that the Lancet had a system of throwing a pile of papers down the stairs and publishing those that reached the bottom.” Obviously, peer review also creates echo chambers, fashions, and power games, where an established cartel punishes new and fresh ideas.

12. Peer review III: in one study, 12 papers with high-recognition authors that had already been published were simply copied and resubmitted to the same journals but with no-name authors from made-up institutions. Only 3 journals detected the resubmission, one was accepted, and the remaining 8 were rejected! Sometimes the peer reviewers pointed out serious methodological flaws, and “none of the twenty reviewers who recommended rejection even hinted at the possibility that a manuscript might be acceptable for publication pending revision or rewriting.” Mind you, these papers had already been published by the same journals! What a difference an established name makes.

13. Peer review IV: people also forget that peer review is a new phenomenon, and wasn’t around during what many consider the golden age of science. Einstein apparently was furious when he was to be subjected to peer review for the first time, and went on to publish elsewhere.

14. The sheer size of science and research teams makes innovation almost impossible. One analysis of 65 million papers, patents, and software products from 1954-2014 concludes that “small teams disrupt science and technology by exploring and amplifying promising ideas from older and less popular work,” as opposed to individuals in large groups.

Note that some of these numbers come from surveys among scientists where they are likely to severely underreport issues, or from the very few brave insiders who cared enough to share their observations: it is likely that we are just seeing the tip of the iceberg here.

Also note that we aren’t even talking yet about ruthless people using scientific institutions to gain money and power, the corruption of science by corporate and state interests, the entrenched dogmas, ideologies, etc., or the persecution of “heretics” who dare questioning orthodoxy or—gasp!—writing their own blogs outside the institutions. And still, even without all that, the picture looks mind-bogglingly bad.

To close, here’s a great conversation between Iain McGilchrist and Rupert Sheldrake, two of the very few real “scientists” (the word feels almost like an insult at this point) around these days. I particularly liked how they went their own ways, and Sheldrake’s story about his year at Harvard is hilarious and speaks to his character.

Can science be saved? What about the “scientific method” and the hopes of mankind connected to science?

Frankly, I’m pessimistic. A lot of dead wood has to be burned—institutionally, intellectually, philosophically—before we will get a new Enlightenment, and it probably won’t involve existing institutions, or at least not in any way resembling their current form.

We will likely have to go back to the basics: to individuals and small, close groups of academics who actually think. They will likely be independent and with little funding for a while, but this could be a good thing: it forces you to make every experiment count, to look for inexpensive ways of doing it, to emphasize deep reflection (which costs nothing but time), and to move away from big machines, expensive toys, and the hyper-bureaucratic organization and corporate corruption that often go along with those. Plus, if not much money is to be made, this might also weed out those who are just in it for the three P’s™: Pension, Power, Prestige.

That being said, science sometimes does need lots of funding. Maybe one day we will get there, hopefully in a different world, under different paradigms.

1. For a complete list of references, please do yourself a favor and get a copy of Iain McGilchrist’s The Matter With Things; you’ll find them in chapter 13. Please read the other chapters as well :)

That being said, here’s some further reading:

Ioannidis JP, “Why most published research findings are false,” PLoS Medicine, 2005, 2 (8), e124

Smith R, “Peer review: a flawed process at the heart of science and journals,” Journal of the Royal Society of Medicine, 2006, 99(4)

And here’s a great article which shows the many absurdities of the peer review process:

The rise and fall of peer review