Wednesday, March 20, 2019

Science has a problem. Here is how you can help.

[I have gotten numerous requests by people who want to share Appendix C of my book. The content is copyrighted, of course, but my publisher kindly agreed that I can make it publicly available. You may use this text for non-commercial purposes, so long as you add the copyright disclaimer, see bottom of post.]

Both bottom-up and top-down measures are necessary to improve the current situation. This is an interdisciplinary problem whose solution requires input from the sociology of science, philosophy, psychology, and – most importantly – the practicing scientists themselves. Details differ by research area. One size does not fit all. Here is what you can do to help.

As a scientist:
  • Learn about social and cognitive biases: Become aware of what they are and under which circumstances they are likely to occur. Tell your colleagues.
  • Prevent social and cognitive biases: If you organize conferences, encourage speakers to not only list motivations but also shortcomings. Don’t forget to discuss “known problems.” Invite researchers from competing programs. If you review papers, make sure open questions are adequately mentioned and discussed. Flag marketing as scientifically inadequate. Don’t discount research just because it’s not presented excitingly enough or because few people work on it.
  • Beware the influence of media and social networks: What you read and what your friends talk about affects your interests. Be careful what you let into your head. If you consider a topic for future research, factor in that you might have been influenced by how often you have heard others speak about it positively.
  • Build a culture of criticism: Ignoring bad ideas doesn’t make them go away, they will still eat up funding. Read other researchers’ work and make your criticism publicly available. Don’t chide colleagues for criticizing others or think of them as unproductive or aggressive. Killing ideas is a necessary part of science. Think of it as community service.
  • Say no: If a policy affects your objectivity, for example because it makes continued funding dependent on the popularity of your research results, point out that it interferes with good scientific conduct and should be amended. If your university praises its productivity by paper counts and you feel that this promotes quantity over quality, say that you disapprove of such statements.
As a higher ed administrator, science policy maker, journal editor, representative of funding body:
  • Do your own thing: Don’t export decisions to others. Don’t judge scientists by how many grants they won or how popular their research is – these are judgements by others who themselves relied on others. Make up your own mind, carry responsibility. If you must use measures, create your own. Better still, ask scientists to come up with their own measures.
  • Use clear guidelines: If you have to rely on external reviewers, formulate recommendations for how to counteract biases to the extent possible. Reviewers should not base their judgment on the popularity of a research area or the person. If a reviewer’s continued funding depends on the well-being of a certain research area, they have a conflict of interest and should not review papers in their own area. That will be a problem because this conflict of interest is presently everywhere. See next 3 points to alleviate it.
  • Make commitments: You have to get over the idea that all science can be done by postdocs on 2-year fellowships. Tenure was institutionalized for a reason and that reason is still valid. If that means fewer people, then so be it. You can either produce loads of papers that nobody will care about 10 years from now, or you can be the seed of ideas that will still be talked about in 1000 years. Take your pick. Short-term funding means short-term thinking.
  • Encourage a change of field: Scientists have a natural tendency to stick to what they know already. If the promise of a research area declines, they need a way to get out, otherwise you’ll end up investing money into dying fields. Therefore, offer reeducation support, 1-2 year grants that allow scientists to learn the basics of a new field and to establish contacts. During that period they should not be expected to produce papers or give conference talks.
  • Hire full-time reviewers: Create safe positions for scientists specialized in providing objective reviews in certain fields. These reviewers should not themselves work in the field and have no personal incentive to take sides. Try to reach agreements with other institutions on the number of such positions.
  • Support the publication of criticism and negative results: Criticism of other people’s work or negative results are presently underappreciated. But these contributions are absolutely essential for the scientific method to work. Find ways to encourage the publication of such communication, for example by dedicated special issues.
  • Offer courses on social and cognitive biases: This should be mandatory for anybody who works in academic research. We are part of communities and we have to learn about the associated pitfalls. Sit together with people from the social sciences, psychology, and the philosophy of science, and come up with proposals for lectures on the topic.
  • Allow a division of labor by specialization in task: Nobody is good at everything, so don’t expect scientists to be. Some are good reviewers, some are good mentors, some are good leaders, and some are skilled at science communication. Allow them to shine in what they’re good at and make best use of it, but don’t require the person who spends their evenings in student Q&A to also bring in loads of grant money. Offer them specific titles, degrees, or honors.
As a science writer or member of the public, ask questions:
  • You’re used to asking about conflicts of interest due to funding from industry. But you should also ask about conflicts of interest due to short-term grants or employment. Does the scientists’ future funding depend on producing the results they just told you about?
  • Likewise, you should ask if the scientists’ chance of continuing their research depends on their work being popular among their colleagues. Does their present position offer adequate protection from peer pressure?
  • And finally, like you are used to scrutinize statistics you should also ask whether the scientists have taken means to address their cognitive biases. Have they provided a balanced account of pros and cons or have they just advertised their own research?
You will find that for almost all research in the foundations of physics the answer to at least one of these questions is no. This means you can’t trust these scientists’ conclusions. Sad but true.


Reprinted from Lost In Math by Sabine Hossenfelder. Copyright © 2018. Available from Basic Books, an imprint of Perseus Books, a division of PBG Publishing, LLC, a subsidiary of Hachette Book Group, Inc

36 comments:

  1. Very good to have this available, but I think the first of the last three question should be re-phrased. As it is, "no" would be the "good" answer with doesn't resonate with the final question. (The third, being an or-question, also is not easy to answer yes/no...)

    ReplyDelete
  2. I think paying reviewers a fair wage is actually the best idea. Thou shalt not muzzle the ox who treadeth out the corn.

    ReplyDelete
  3. I am doing my own thing for more than twelve years now. Conclusion: you are right.

    ReplyDelete
  4. > Don't chime colleagues...

    Like your book, you mean "chide".

    > Nobody is good at everything, so don’t expect scientists to.

    Like the book, "to be".

    > You’re used to ask for conflict of interests due to

    This is incorrectly copied from the book. "You're used to asking about conflicts of interest due to funding from industry. But you should also ask about conflicts of interest due to short-term grants or employment."

    The plural is on "conflict", and the verb is "asking".



    ReplyDelete
    Replies
    1. Thanks for pointing out. I had forgotten that this was the version prior to copy-editing. I've made those changes now.

      Delete
  5. You deserve a lot of credit for having the courage to provide such an important public service while you endure relentless attacks. Bravissima!

    ReplyDelete
  6. Nice initiative ...

    ... In the first years in University, I remember some Teachers talking about Science is Self-correcting but It ends as some sort of Keynesian Market's Invisible Hand as the "mechanism" for Self-correction ... Not really Self-correcting, just another Discipline subjugated by Market's Trends ...

    If There is an Organized and General Agreement between Scientists to collectively confront "The Mythology about, Science is by Market's Grace a Self-correcting Discipline" ... That would be very remarkable ... Like a Second Shot to Vienna's Circle endeavor that deeply influenced The Splendor of Physics at the beginning of the XX Century ... but Now, "On Steroids" by the Current Information Technological Resources ...

    A Noble Goal for A Noble Discipline ...

    The questions, Had Scientists the commitment for such kind of "Philosophical Enterprise" towards Objectivity ??

    Are Policy Makers interested on Social Mechanisms for Displaying Dynamics of Interdisciplinary Objectification towards the Production of Accurate and "Objective" Information ??

    Is The Human Specie ready and/or adequate to realize such complex Collective Endeavor ??

    ReplyDelete
  7. I don't think the point is about getting paid for reviewing papers; I review papers frequently (two so far this year) and count it as hours worked. I am paid for it. As I was when I was a grad student; I reviewed papers (and my advisor reviewed my reviews!).

    I think the point isn't money, but removing the conflicts of interest (to some degree) by making the job full-time, or even a part-time position for retired professors, or non-research "teaching" professors (a new career track I have seen in some colleges).

    I'm not sure what the incentives for this would be, to convince some organization to pay for it. My personal incentive is generally altruistic; I perform good detailed reviews as a duty to the field. It's anonymous, so I don't get anything out of it except either reading something interesting, or knowing I made a paper better, or that I preventing some nonsense pollution.

    One thing I WOULD mandate if ever appointed the King of Science is double-blind reviews; no author info on the review copy of the paper. Yeah, sometimes you can still tell, some authors have quirks of writing, or are overly fond of citing themselves, but I think it helps anyway.

    It is a little too easy to give famous authors deference and presume they know what they are talking about.



    ReplyDelete
  8. I recently re-read Roger Fouts' book Next of Kin, that is about chimpanzees and his research on their language abilities. I was rather godsmacked by reading this in a way I was not when I read it first 20 years ago.

    I will say that I think it is a good idea to read about subjects removed from what you study or research. Physicists should occasionally read about psychology, molecular biologist should read about planetary geology and medical doctors should read about quantum physics on occasion.

    It is a rather extreme view of academic departments, where in Fouts' experience with this sort of animal behavior work he ran into appalling instances of abuse and arrogance. It mirrored some of my experiences as well and one reason I was perfectly happy to work in applied science and industry.

    The thing that gripped me the most is what Fouts challenged was the hierarchical notion of form and function. He challenged the idea of Noam Chomsky who said that language occurred by a genetic determinism that made humans absolutely unique. He taught chimps sign language and later found the progeny of these chimps learned the same from their mothers and the other chimps in the troupe. This broke the idea of there being a sort of executive process that drives everything. Darwin rejected this idea, and the theory of evolution broke the idea of there being some grand designer, mentioned a lot in a book called the Bible, that imposed everything. The language abilities of chimps is based on gestures, where if you think about we humans do a lot of as well when we talk, and the process of linguistic development was more incremental or quantitative and not of some qualitative jump.

    I have been musing over whether in physics we are doing much the same. We tend to have this idea in physics, which in some ways is almost Aristotelian, of there being these formal causes or “potentia of forms” that drive what is physically real. Things may not be that way at all, but rather structure is something that occurs because it is selected for. While we do not have Darwinian survival of the fittest in physics, we do have the idea of quantum interference, where in some ways the structure that define the universe, or local physical symmetries and conservation principles, may not be due to some grand scheme or imperium, but rather are what operate under general constructive and destructive interference of general quantum waves. The search for an optimal quantum error correction code may be had through some selection “sieve” that means the optimal code is constructive interference.

    Maybe we have to return to Socrates who said that in the end he and everyone else knew nothing.

    ReplyDelete
  9. These are interesting times. On the news this morning (in the US), Google and Facebook are being asked to censor misinformation about the risks of vaccines for children. Anti-vaccine activists are invoking their free speech rights to criticize and question the status quo. They view their activism as a "community service."

    Even people with training in critical thinking skills and cognitive bias can suffer from cognitive bias. Also, people generally don't appreciate tough questions and constructive criticism.

    It's a complex issue that affects every facet of human civilization, not just scientific research. I've even seen conflicts of interest and cognitive bias cause serious problems in families. It's an issue from the bottom to the top.

    ReplyDelete
    Replies
    1. "It's a complex issue that affects every facet of human civilization, not just scientific research."

      True. I would like to include learned bias as well. Berzelius was a name in Chemistry. He said organic compounds cannot be synthezised in the laboratory. A learned bias such as this prevents observation because you don't look or refuse to look in learnedly sanctified forbidden places. If you don't look, how can you see? To refuse to look is to ignore. The act of ignoring is ignorance. If Wohler gave in to the learned bias instilled in the psyche of the chemists of his day, that is, if he allowed the bias to interfere in his observation, would he have synthesized urea in the laboratory? No. He would have shut the doors of enquiry, which is to ignore or be cloistered in the dingy, claustrophobic corridors of ignorance. Other examples of learned biases are hypothetical ether and Newtonian absolutes. A Prejudice for or a prejudice against is a bias. The prejudice or the bias is what makes the observer. The observer is the learned bias, the prejudice or any other bias for that matter. How does bias or prejudice arise? It is a product of our conditioning. Conditioning means programming. A bias is a program that runs and influences observation. Such an observation is no observation at all because you cannot see anything new.

      Delete
    2. . . . Then you can see the fact as it is; you are going to color the fact, skew it, distort it according to the bias.

      Delete
  10. The collapse of education is not the result of the pathologies that you diagnose. It is the precondition.

    ReplyDelete
  11. Personally, I think the collapse of education is due to the greed and corruption brought on by the conversion of universities, research centers, the grant process and even academic publishers into explicitly for-profit businesses.

    In my lifetime (and IMO), it has increasingly become about patents and commercial or military (Department of Defense in the USA) partnerships. It has become about "fame" that gets big grants for big projects, like $10 million "centers" for projects, every one of them over-promised in order to get the $. The marketing sounds great! You need a big hammer to break a big problem, and the ramifications of breaking it will be endless sunshine and red roses, so put us to work!

    As money and profits and patents and the bottom line become paramount, then just like other multi-billion dollar enterprises, honesty and morality become luxuries. Nothing matters but the bank account and assets and growth.

    Then we see what is observable today, lies and misleading narratives about the bottom line and "impact" for just about everything. And just like big business, we find people to lead the charge that just don't care what they have to say to close the deal. While people that do care about the truth find themselves at a fundamental disadvantage for gaining influence or power, because wonderful lies are wildly more successful at raising money.

    ReplyDelete
    Replies
    1. @ Dr. A.M. Castaldo,

      That seems to be an accurate sociological generalization ...

      The Challenge is How to confront that in the day by day circumstances ...

      Culture is like a river, and individuals are like tiny corpuscles subjugated in the river's flow ...

      Delete
    2. @ Dr. A.M. Castaldo

      I think you have hit the mark exactly! Indeed almost everything that Sabine has been discussing boils down to the simple question - are scientists overwhelmingly focused on discovering the truth, or are they primarily driven by other goals?

      IMHO this problem has spread through huge swathes of science, with the exception of areas where bad science = no profits.

      Delete
    3. @First: If I knew ... :-)

      The only way I know that works is collective sacrifice of some sort; finding enough people below the top (i.e. not in charge) that are willing to give something up by not going along with the people that are in charge. Basically a non-violent protest, a lose-lose game that, nevertheless, effects a positive change.

      For example, read on the 13-month long Montgomery Bus Boycott, against racial segregation. It worked. Obviously the blacks boycotting sacrificed a lot of time and energy by walking miles every day for over a year, but in the end they prevailed.

      We have the same problem in science. We go along with the corrupting influence of the mantra "it's all about the money" because it seems there is no other way. Not with causing ourselves pain and hardship.

      But that is the point; causing ourselves pain and hardship in a way that our sacrifice will cost the existing leaders of the field enough to force a change. All in the legal realm, of course.

      In my view, Dr. Hossenfelder has a start on that kind of sacrifice, she stopped working in the field due to scientific principle. She is clearly insulted and lied about frequently, for talking about those principles. I consider that a sacrifice too. I am not in the community of physicists, but I hope she builds a following, with other people willing to follow her example and insist upon the truth, and calling out the falsehoods and rampant bad logic, and adhering to the principles of actual science. Even if that constitutes a personal sacrifice, for the good of the future of science.

      But that is the part I don't know how accomplish, I'm not Martin Luther King.

      I truly think the future of science is at stake; throwing $10 or $20 billion dollars down a hole every decade and never getting the (falsely promised) litany of results -- That creates distrust, and being the premier project of all of science, it creates distrust in all of science.

      We see a similar problem in medicine; with anti-vaxxers that don't trust the medical establishment, or their statistics, or their studies, or anything else. Can you blame them? More than half of medical studies can't be replicated. These same doctors okayed the oxytocin for long term use and caused an opioid epidemic. Doctors once okay'd smoking as safe. People go years on end with misdiagnoses, spoken with full confidence.

      We see the same thing with climate change. I believe in it, but why should anybody without my education believe in it? Climate change sounds like overblown hyperbole. When the rest of science is full of hype and lies and self-interested claims, laymen are actually being scientifically correct to suspect the seemingly hyperbolic claims of climate scientists may be fabricated for some kind of selfish interest.

      Just like the hyperbolic life-changing claims of medicine, and physics, and economics, and AI, that never seem to quite pan out.

      Delete
    4. @Dr. A.M. Castaldo. Medicine isn't all overblown hyperbole. The 5-year survival rate for breast cancer in women increased from 75% in 1975 to 90% in 2000. And similar improvements have happened for many other types of cancer. Maybe this isn't the "cure for cancer" we were promised, but it's a remarkable improvement.

      Delete
    5. @Peter Shor: I agree with you; I did not mean to imply all of it, just, specifically, all the hyperbolic life-changing claims.

      Physics is not all overblown hyperbole either. Neither is artificial intelligence; tangible advances have been made there too. Self-driving cars are quite a feat already; Watson is a marvel at Jeopardy.

      I am talking specifically about what seems to me rampant promotional hyperbole in all these fields, generally about "what is just around the corner," if not by practitioners then not corrected by them when given an opportunity.

      We have a lot of success stories in AI, but life-like, human looking, conversational helper robots isn't one of them. Watson or other AI might be able to improve medical diagnoses, but I haven't seen AI diagnosticians in any doctor's offices yet, and no telling if and when a diagnostic Watson will be common clinic equipment.

      Science does make progress. But IMO hyperbole infects all these fields, in large part to secure funding (or fame which indirectly aids funding).

      And then of course the hype is what makes the headlines, because reporters like something shiny. Because consumers like something shiny. And then the hype bubble pops and splatters, so all scientists seem like hustlers, even when talking about real progress, or trying to get an actually rational project funded.

      You're right, there have been some remarkable improvements. But it also isn't the "cure for cancer" the public has been promised a few hundred times, with half a million people still dying of cancer every year (2018, 590K). So often that "the next cure for cancer" has become synonymous with "hype".

      Delete
    6. I think the danger of saying "scientists in all fields make hyperbolic claims" is that it obscures the fact that there something really has gone wrong with high-energy physics that hasn't in most other fields of science. While medicine promises lots of cures that don't pan out, some of them really do work, and save millions of lives. And medicine often drops the techniques that don't work (and when they don't, it's the pharmaceutical companies and not the researchers who are at fault). Similarly, machine learning in AI is generating an enormous amount of hyperbole right now. But deep learning really does do some remarkable things, and many of the researchers are trying to damp down the hyperbole, although it's not working that well. (And it's not just two or three lone voices, like it is in high energy physics, but a reasonable number of people.) The NAS has recently come out with a prediction for the timeline in quantum computing that is more pessimistic (and probably more realistic) than much of the hyperbole right now, and some people are paying attention to it.

      Delete
    7. Those are fair points. I guess I'm thinking they are all cut from the same cloth; but there aren't many $10B projects in the world (even our biggest supercomputers are in the half billion range), so perhaps that level of money is what makes HEP the worst in the realm.

      HEP may just be out of testable new ideas after the LHC, where other fields are not. It seems that way to me: The big idea totally failed to find what we wanted, but it has to work so let's try it again, harder.

      Delete
  12. Thank you!

    I wanted to express my gratitude for the book of yours.

    About 15 years ago I read Weinberg's "Dreams of Final theory", and it defined my views on science, in particular my distrust for philosophy. Now your book shifted my attitudes a lot.

    ReplyDelete
  13. This is good advice, thanks for posting it!

    ReplyDelete
  14. Just finished reading your book. You are brave for publishing it; thank you and keep up the good work that you do.

    ReplyDelete
  15. Excellent list. I'd only add accrediting agencies as targets of the list for administrators. Most admins don't WANT to change, but an admin who did try to follow your suggestions would risk a loss of accreditation.

    ReplyDelete
  16. A cage containing a banana with a hole large enough for a monkey's hand to fit in, but not large enough for a monkey's fist (clutching a banana) to come out; anecdotally used to catch monkeys that lack the intellect to let go of the banana and run away.

    Human behavior retains this need to aquire, maintain, and profit from an idea, invention, discovery, in general, any status increasing event.

    I have seen the same idea or invention discovered dozens of times and this potential increase in the store of human knowledge never sticks.

    How does Bee's rules for proper behavior delete this need for status enhancement. Could "Prevent social and cognitive biases" be synonymous stated as "beware the monkey trap"?

    ReplyDelete
  17. The most important cognitive bias was aptly summarized by Uptain Sinclair:

    It is difficult to get a man to understand something when his salary depends upon his not understanding it.

    ReplyDelete
  18. Polistra wrote: an admin who did try to follow your suggestions would risk a loss of accreditation

    Could you elaborate on that a little please, with one or two examples?

    ReplyDelete
  19. I also enjoyed your book. My take-away from it is that, in a field that is not advancing much, fads take over and change rapidly. Keeping up with the fads is important for grant funding and researchers' advancement.

    ReplyDelete
  20. Build a culture of criticism
    Today it seems scientists seek consensus. IMO this is bad for science. It leads to groupthink. One observation or one scientist is enough to overturn the consensus. Debate is good. Scientists should try to falsify theories. Strictly speaking, as Karl Popper put it, you cannot prove a scientific theory, you can only falsify it.

    Hire full-time reviewers
    Scientists should post their papers online and let everybody review it. IMO this is the best practice. An excellent example is Perelman’s proof of Poincare Conjecture. Perelman did not publish his papers in any mathematical journal. He only posted it online at arxiv. Mathematicians all over the world could read them. They organized their own study groups to discuss and peer-review the papers. One of the greatest revolutions in mathematics was done without the participation of the gatekeepers of journals.

    ReplyDelete
  21. To say about any scientific theory "Ultimately, it's probably wrong" is probably right.

    ReplyDelete
  22. Scientists are the ones who go daily, or sometimes 24/7, into the lab to collect data. And then analyze it. It's a grind. Those who can't do that just publish articles about scientists.

    ReplyDelete
    Replies
    1. Since you just published a claim about scientists, I presume you fall into your second category of 'Those who can't do that'. :-)

      Delete
  23. Science no problem except that generated by those who want to sell books about science problems.

    ReplyDelete
  24. Comment moderation is on. Because I want to censor comments I don't agree with. Take Ken Abbott for example, the guy is an idiot.

    ReplyDelete

COMMENTS ON THIS BLOG ARE PERMANENTLY CLOSED. You can join the discussion on Patreon.

Note: Only a member of this blog may post a comment.