“It is not enough to wear the mantle of Galileo…”

Alessandro Strumia is back in the press again. Earlier this month CERN decided to sever all ties with him due to the fallout from that presentation. I’ve written about the Strumia case previously, both here at the blog (at some length — see this and this) and in the pages of Physics World, so won’t rehash the many arguments against his thoroughly biased and pseudoscientific claims about women in physics. Prof. Strumia also got in touch with me in January, following my criticism in Physics World, and an e-mail exchange ensued. I’d have liked to have made that exchange public here at Symptoms… but Alessandro preferred not to have our debate in the open.

What’s clear from today’s article in The Sunday Times is that Strumia isn’t going to let counter-evidence or counter-arguments affect his ideology. Once again, and to quote from the piece I wrote for Physics World (if you’ll excuse the self-plagiarism), he’s presenting himself as the “ever-so-courageous rational scientist speaking “The Truth”, when, of course, he’s entirely wedded to a glaringly obvious ideology and unscientifically cherry-picks his data accordingly.” Cue a Letter To The Editor…

“Dear Editor,

Alessandro Strumia (Alessandro Strumia: the data doesn’t lie
— women don’t like physics, Sunday Times, March 24) claims that his views on women in physics have been censured due to “excessive political correctness”. Many years ago, the physicist Robert L Park highlighted a key proviso for those who opine that their “radical” theories are being stifled, viz.

Alas, to wear the mantle of Galileo it is not enough that you be persecuted by an unkind establishment, you must also be right

Your article on CERN’s severance of ties with Strumia, while refreshingly even-handed, didn’t quite capture the deeply pseudoscientific tenor of his “analysis” (and I use that term advisedly). Prof. Strumia asserts, on the basis of a fundamentally flawed and credulous set of suppositions, that the IQ of the authors of a scientific paper scales directly with the number of citations accrued. En route, he confused correlation with causation, cherry-picked his cited sources to a remarkable extent, and indulged in overwrought rhetoric more akin to an amateur YouTube pundit than a professional, established scientist speaking to his peers at a conference.

Strumia’s presentation was a masterclass in what Richard Feynman, the physicist’s physicist, described as cargo cult science: to the untrained eye it looks scientific, but the essential ingredients of objectivity, rigorous self-criticism, and lack of ideological bias are sorely missing. Although I don’t agree with Strumia being dismissed for his reactionary views, his time would be better spent on informing himself about the complexity of the underlying science than crying victimhood at the hands of “The Establishment”. Might I recommend Angela Saini’s “Inferior: How Science Got Women Wrong“?

Yours sincerely,

Philip Moriarty”

 

 

Author: Philip Moriarty

Physicist. Metal fan. Father of three. Step-dad to be. Substantially worse half to my fiancée Lori, whose patience with my Spinal Tap obsession goes to far beyond 11...

31 thoughts on ““It is not enough to wear the mantle of Galileo…””

    1. Alessandro,

      As noted in our lengthy email exchange (and in the blog post above), it would be helpful if you could put aside the overwrought and over-emotional “identity politics/cultural Marxism/PC orthodoxy” buzzwords and focus on the science. (I see lots of victimhood here but not from the source you’re suggesting.)

      The crux of your argument for the talk at CERN was that IQ scales directly with number of citations. In our lengthy e-mail exchange you have yet to come close to providing a semblance of a credible justification of that pseudoscientific claim.

      Note that correlation and causation are two very different things, despite appearances to the contrary at times: http://www.tylervigen.com/spurious-correlations .

      I look forward to your justification that citations — a measure of impact of a paper with a wide variety of confounding variables — scale directly with the IQ of *each individual author of a paper*. I’ve been waiting a number of months and have yet to get a credible response from you. See https://muircheartblog.wordpress.com/2018/10/21/the-natural-order-of-things-part-iii-the-song-remains-the-same/ for more.

      Thanks and best wishes,

      Philip

      Like

      1. Dear Philip,

        let me summarize what I really claim: citations allow to devise bibliometric indices strongly correlated with human evaluations of scientific merit. When applied to fundamental physics, the statistical distributions of such indices show a gender difference well fitted by a difference in variability. This difference persists after accounting for confounders. Its size is compatible with the difference in variability claimed by other literature in abilities reasonably relevant for research in physics.

        The relevance of a gender difference in variabilities for physics and STEM was recently discussed, among others, by the former Harvard theorist Lubos Motl, by the former Harvard president Lawrence Summers and by the former Google engineer James Damore. In the polemics that followed, experts weighted the non-univocal scientific literature, producing meta-reviews with these conclusions: «Males are more variable on most measures of quantitative and visuospatial ability, which necessarily results in more males at both high- and low-ability extremes». «Substantial evidence suggests that the male advantage in mathematics is largest at the upper end of the ability distribution». «On average, male variability is greater than female variability on a variety of measures of cognitive ability, personality traits, and interests. This means men are more likely to be found at both the low and high end of these distributions». «Based on the meta-analyses we reviewed and the research on the Greater Male Variability Hypothesis, Damore is correct that there are population level differences in distributions of traits that are likely to be relevant for understanding gender gaps at Google». What happened to Motl, Summers, Damore etc shows that the problem goes beyond science.

        best,
        Alessandro

        Like

        1. Dear Alessandro,

          I’ve addressed the points in your comment above time and time again in the e-mail exchange we had, as you know full well. Why don’t we just publish that exchange instead of tediously going around the houses yet again? I’ve had to endure enough Gish gallops in the past to know how this plays out. I spent a great deal of time addressing your comments in good faith via e-mail. Why don’t we save ourselves a lot of time and effort and just publish the e-mail trail? Remember that you are supposedly the proponent of open debate.

          But, in a nutshell, how do you credibly deconvolve the (immutable?) biological/genetic effects you claim from the environmental influence? Where is your compelling, watertight evidence to show that the genetic/biological signal is stronger than the environmental (i.e. sociocultural) noise?

          With regard to this point, I recommended Saini’s and Fine’s books to you. Did you get round to reading them?

          Best wishes,

          Philip

          Like

          1. Alessandro,

            Frther to my comment above, the following is from my most recent response to you (24 Jan 2019), to which I did not receive a reply. I have stripped out direct quotes from your preceding e-mail as you did not give permission to make the exchange public.

            Best wishes,

            Philip

            —-

            Anyway, back to the science. Thank you for finally leaving the nonsense about “PC orthodoxy” to one side and focussing on addressing my points in your most recent e-mail. I will in turn address your points but, first, let’s remind ourselves of your central, and fundamentally flawed (at best), premise: citations are a direct measure of scientific quality and therefore of IQ. Before I tease apart your other (secondary) arguments, let’s just focus on the crux of your ‘thesis’, such as it is.

            There’s a glaring lack of self-consistency in your position, Alessandro. You yourself have admitted that citations are a “proxy” for scientific quality/merit that relies on the perception of the scientific community. In other words, you are using a metric that is fundamentally derived from the social dynamics underpinning science and then arguing that this (magically) tells you about innate (and immutable?) IQ differences.

            There are very many papers in the literature that have attracted high numbers of citations that are just plain wrong (or, indeed, fraudulent) and/or lacking in scientific creativity, insight, or intelligence. Even for papers that are as correct as they can be at the time of publication (because science is always evolving), the number of citations is not a direct indicator of scientific quality. Are you really of the opinion that a paper that has attracted 100 citations is definitively of twice the scientific quality as another that has accrued 50 citations, Alessandro? I hope not. And yet this is the entire basis of your reasoning.

            Here’s one good example of where citations are most definitely not a measure of scientific quality: https://muircheartblog.wordpress.com/2013/04/19/philip-moriarty-not-everything-that-counts-can-be-counted/ (An update on the citation counts for the papers discussed in that post: As of the time of writing (24/01/19), Manoharan et al.’s elegant and highly impressive tour de force paper: 48 citations. Our paper on nanofluid dewetting (which clearly lacks the scientific insight, elegance, and quality of Manoharan et al.’s work): 78 citations).

            Another example? Try https://www.nature.com/articles/nmat1116 This paper has attracted 429 citations to date. And, as we discuss at length in this — https://journals.plos.org/plosone/article?id=10.1371/journal.pone.0108482 — it’s fundamentally flawed, involving extremely poor scanning probe microscope operation and analysis. Yet this paper, and other papers based fundamentally on the flawed reasoning in that original work (including Nat Materials 7 588 (2008); 794 citations), attract large citation counts.

            Or, as an example which is rather closer to home for you, how about the 500 or so papers published on a statistical blip? https://en.wikipedia.org/wiki/750_GeV_diphoton_excess Are the citations to those papers meant to tell us about scientific quality when each and every one was based on an analysis of noise? You, and others, ‘discovered’ a non-existent signal in noise. (The parallels with your credulous extraction of an IQ signal from citation data are intriguing, to say the least.)

            You will, of course, argue that it’s the trends that are important, not individual examples. But you’re flat-out wrong. Correlation is not causation. And those individual papers are the exceptions that effortlessly counter your simplistic “rule”. (Think of it as a Popperian approach, if you like.) If an analysis is based on faulty reasoning, viz. your premise that scientific quality scales directly with citations, it doesn’t matter how you massage the numbers. Similarly, if a signal turns out to be a statistical blip it doesn’t matter how persuasive the arguments associated with that blip may or may not be, or how compelling the correlations happen to be; it’s junk science. Some of us just place a higher bar on scientific evidence than others, Alessandro. Your willingness to jump on board an over-enthusiastic analysis of noise is indicative of the low bar you are willing to set yourself for robust, watertight evidence. Best not to chase ambulances (https://telescoper.wordpress.com/2018/10/03/the-strumia-affair/#comment-336480 )

            Like

            1. sorry, your email got blocked by the new anti-spam filter at my University.

              Citations are a useful tool despite counter-examples of papers cited for wrong reasons. Counter-examples mean that the correlation between bibliometric indices based on citations and scientific merit is imperfect. Still, a positive correlation is enough for the gender analysis (the signal is big enough that it can be seen looking an imperfect variable). Tables 6, 7, 8, 9, 10 of https://www.sciencedirect.com/science/article/pii/S1751157718301512?dgcid=author compare various bibliometric indices with human evaluations of scientific merit: some indices show a high positive correlation, others don’t. The gender analysis is based on good indices. Furthermore, one can see the same gender difference looking at the number of fractionally-counted publications, without using any information on citations.

              As I wrote in my slides, the situation is analogous to Higgs discovery: zero single events can be unambiguously attributed to the Higgs; still a statistical analysis of enough events allows to infer that the Higgs exists.

              PS my papers about the 750 GeV resonance made clear that “new data will decide if it􏰀 will reach the Higgs in the Particle Data Group or if 􏰀it will instead reach N-rays in the cemetery of anomalies”. Research needs exploring dangerous topics that might sink. Crossing the Pillars of Hercules is more dangerous than circumnavigating Elba island, but it’s needed for discovering “India”.

              Like

          2. I read the book where Saini systematically tries to cast doubts about results finding gender cognitive differences. Some criticisms are interesting, but seem not enough to challenge the evidence.
            One can criticise theories that try to explain why autism is 3 times more prevalent among men, but insisting that all such differences must be socially constructed degenerates in a crusade against common-sense.

            Concerning my results, I only claim that gender differences observed in bibliometric data about fundamental physics can be interpreted in terms of two main ingredients well documented in the scientific literature: gender differences in interests and higher male variability. Needless to say, “for every complex natural phenomenon there is a simple, elegant, compelling, wrong explanation”: dealing with complex systems any simple interpretation can easily be incomplete, including a hypothetical gender discrimination. In any case, it is interesting that data can be explained without invoking such hypothesis and focusing instead on the two ingredients above.

            Like

            1. Alessandro,

              I’ll get back to you in detail later today — I’ve got undergrad lab demonstrating and a lecture to do this morning. But I’ve addressed your points in detail previously via e-mail. I don’t quite understand why we don’t just publish that e-mail trail? It would save us both so much time. Thank you, however, for so helpfully backing up the point I made about cargo cult science in the letter to The Sunday Times.

              For now,

              1. Correlation is not causation. This is Stats 101. See http://www.tylervigen.com/spurious-correlations

              One of my favourites is this…

              2. The Twitter thread linked below will be of interest. We’ve covered a number of the points in our email correspondence, as you may recall, but Celeste Labedz makes them much more pithily and compellingly than I ever could:

              3. https://twitter.com/AmandaGefter/status/1109898272632655872

              Like

            2. As you know, journals like provocative inaccurate titles. What I really talked about is gender differences in interests. A meta-review of the literature summarizes the scientific status of this phenomenon as: «Gender differences in interest and enjoyment of math, coding, and highly systemizing activities are large. The difference on traits related to preferences for “people vs. things” is found consistently and is very large». From R. Su, J. Rounds, P.I. Armstrong, “Men and things, women and people: A meta-analysis of sex differences in interests”, Psychol. Bull. 135 (2009) 859. Such difference in interests is consistent with gender differences in preferred faculties, which are big and similar in different countries. Female students are majority in many faculties: do you claim that it’s because these faculties discriminate against male students? If not, why attacking physics and STEM for the opposite reason? Physics departments are equally open to everybody: men are not invited with limousines and women don’t need to wear moustaches to pass fences. On the contrary, many departments deliberately worked to attract more female students. However, less women choose physics in countries with more freedom of choice, and gender differences in interest are growing in Scandinavian countries that worked to remove socially constructed gender differences.

              PS I see that you and others repeat the generic sentence “correlation is not causation”. Sure, but can you tell which specific criticism do you have in mind? Do you want to tell that bibliometric indices should not be trusted as proxies of scientific merit, despite that they can achieve large correlation with human evaluations of scientific merit?

              Like

  1. Alessandro,

    Apologies for the delay in responding — undergrad teaching and admissions duties obviously take priority.

    I’m going to start by addressing a comment that you relegated to a PS. It’s a wonderful insight into how we differ in our approach to science and the analysis of data. (Your comments are in bold in the following.)

    ” my papers about the 750 GeV resonance made clear that “new data will decide if it􏰀 will reach the Higgs in the Particle Data Group or if 􏰀it will instead reach N-rays in the cemetery of anomalies”. Research needs exploring dangerous topics that might sink. Crossing the Pillars of Hercules is more dangerous than circumnavigating Elba island, but it’s needed for discovering “India”.”

    This is such a telling comment, Alessandro. It speaks volumes that you argue it’s intellectually courageous to interpret noise as signal; the grandiose reference to the “Pillars of Hercules” only adds to the comedy value. (As I’ve noted previously, the parallels with your credulous (but oh-so-brave) “citations are a direct measure of IQ” interpretation are striking.)

    Here’s another “brave” interpretation of a signal: https://muircheartblog.wordpress.com/2018/10/24/how-not-to-do-spectral-analysis-101/

    …and the next time one of our first year undergrads mistakes that “peak” of 15 counts above the background in their X-ray diffraction data as a signal, I’ll follow your advice and tell them they’re a true scientific explorer “crossing the Pillars of Hercules”.

    Or maybe I won’t.

    Let me remind you yet again of what Feynman said:

    I’m talking about a specific, extra type of integrity that is not lying, but bending over backwards to show how you’re maybe wrong, that you ought to do when acting as a scientist. And this is our responsibility as scientists, certainly to other scientists, and I think to laymen….

    That quote is taken from Feynman’s famous cargo cult science speech, which I’ve referenced just a few times now. If you don’t mind, I think I’ll continue passing on this advice to the students and researchers I work with rather than encouraging them to cross the “Pillars of Hercules”, i.e. cut corners and be less self-critical.

    Here’s another quote from that timeless speech:

    I was shocked to hear of an experiment done at the big accelerator at the National Accelerator Laboratory, where a person used deuterium. In order to compare his heavy hydrogen results to what might happen to light hydrogen he had to use data from someone else’s experiment on light hydrogen, which was done on different apparatus. When asked he said it was because he couldn’t get time on the program (because there’s so little time and it’s such expensive apparatus) to do the experiment with light hydrogen on this apparatus because there wouldn’t be any new result. And so the men in charge of programs at NAL are so anxious for new results, in order to get more money to keep the thing going for public relations purposes, they are destroying—possibly—the value of the experiments themselves, which is the whole purpose of the thing. It is often hard for the experimenters there to complete their work as their scientific integrity demands.

    “….so anxious for new results….” I wonder what Feynman would think of your “ambulance chasing” approach to science, Alessandro?

    PS I see that you and others repeat the generic sentence “correlation is not causation”. Sure, but can you tell which specific criticism do you have in mind? Do you want to tell that bibliometric indices should not be trusted as proxies of scientific merit, despite that they can achieve large correlation with human evaluations of scientific merit?

    It’s a generic sentence for very good reason, Alessandro! It’s exceptionally good advice that, much like Feynman’s wonderful guidance about bending over backwards to prove ourselves wrong, you have ignored when it comes to your simplistic and flawed approach to attempting to analyse gender differences.

    I have told you, time and again, what “specific criticisms” I have in mind. As I noted yesterday, it is frustrating to have to repeat all of this when I spent a considerable amount of time detailing my criticisms in our e-mail exchange.

    Of course bibliometric indices should not be used in isolation as a quantitative proxy of scientific merit! This naive and gullible assertion is a cornerstone of your “citations scale with IQ” thesis. Are you really of the opinion that a scientist with 5,000 citations is definitively of twice the quality of a scientist with 2,500 citations? Or that a paper with 100 citations is twice the quality of that with 50 citations? (I’ve now asked that latter question of you three times, Alessandro. Perhaps this time you could give me an answer?)

    Your following sentence is circular in its ‘reasoning’: “Do you want to tell that bibliometric indices should not be trusted as proxies of scientific merit, despite that they can achieve large correlation with human evaluations of scientific merit?”

    Bibliometric indices are “human evaluations of scientific merit”, not some type of independent quantitative measure! Citations are a function of very many social dynamics within the scientific `ecosystem’. That Manoharan et al.’s elegant and exceptionally high quality paper received many fewer citations than our rather more pedestrian work on nanofluids (see above and https://muircheartblog.wordpress.com/2013/04/19/philip-moriarty-not-everything-that-counts-can-be-counted/) is very much not an isolated example.

    What’s remarkable — and, yes, I’ve made this point before — is you then go that extra step in unblinking credulity and assume that citations are directly related to IQ. Instead of stretching and over-reaching, Alessandro, try following Feynman’s advice once in a while and bend over backwards.

    I’ll address your other comments in a different reply.

    Like

    1. hi, I too have other duties. Let me try to quickly answer to your two issues above.

      First: yes, bibliometric indices indeed are “human evaluations of scientific merit”. I used bibliometrics because it differs from other methods like textile machinery differs from manual working: you get the same kind of product in a much more efficient way. This is why bibliometrics gets adopted, despite Luddite resistance. Bibliometrics is a method of quantifying the average opinion of the community. It might be wrong and evolve with time, as more data arrive, but still it’s the best we have. I agree that one needs to be careful in interpreting O(2) differences when looking at single cases. But my results are based on 70,000 authors, and their bibliometric outputs span five orders of magnitude. This is why the final result does not change when using different metrics that give different O(2) factors: fluctuations average out. I think we agree on the philosophical aspect of the issue, but the quantitative aspect is relevant. Just to make an example: would you argue that an author that wrote 200 publications that received 130000 citations might be as good as an author who wrote 9 publications that received 219 citations?

      Coming to your other point, the relevant issue is the statistical significance of the gender differences found in bibliometric data, discussed in the preprint that I cannot post on arXiv. Criticising it on the basis of the statistical significance of the 750 GeV anomaly is an irrelevant point, in the style of those books that try to negate gender differences by raising many irrelevant points because they don’t have one good point. Criticising whatever contradicts an ideological agenda by raising clouds of dust is a method of politics, not a method of science. Studying anomalies is part of good science, because anomalies indicate specific topics with a higher probability of becoming future real discoveries. As in many other fields there is no warranty of success, but trying is needed to make progress.

      Like

      1. I know, like me, that you’re busy so I’ll give you the benefit of the doubt but those aren’t counter-arguments, Alessandro.

        would you argue that an author that wrote 200 publications that received 130000 citations might be as good as an author who wrote 9 publications that received 219 citations?

        Without reading their papers or considering the other aspects of their career, I cannot answer that question. Brian Wansink springs to mind. Again. (See our e-mail exchange.)

        Studying anomalies is part of good science,

        It was a statistical blip, not an anomaly. You just have a much lower threshold for scientific evidence than I (and many others) do, Alessandro. This is why you are willing to place your faith in simplistic and reductionist approaches to gauging scientific quality and why you interpret a correlation as a direct causal relationship.

        Like

        1. Philip, you can read 200+9 papers to check the correctness of the opinions of your colleagues who cited them. But you cannot read 1.3 million of papers. That’s why we need trusting the opinions of colleagues who read these papers, and use bibliometrics.

          About anomalies, check hep-ph on arXiv today: 7 papers out of 26 are about anomalies less significant that the deceased 750 GeV anomaly (formally due to a 4σ local statistical fluctuation, ignoring other aspects of the story). Studying anomalies does not mean having “a much lower threshold for scientific evidence”. Studying and clarifying is one approach to science. What is not science is your ParticleForJustice colleagues who worked on the 750 GeV anomaly and (in the absence of better arguments) attack me for having worked on the 750 GeV anomaly.

          Like

          1. Alessandro,

            Good. We’re making progress.

            That’s why we need trusting the opinions of colleagues who read these papers, and use bibliometrics.

            Bibliometrics are indeed a measure of the opinions of different scientists and scientific communities. I agree entirely. This has been my point all along. It’s a matter of trust and opinions. Bibliometrics are not a quantitative, incontrovertible measure of scientific quality, and certainly not, as you naively suggest, a measure of IQ. That trust, and those opinions, are just as subject to bias as in any other sphere of human activity. I hate to cite “Cargo Cult Science” yet again but here’s Feynman on the role of social dynamics in physics…

            One example: Millikan measured the charge on an electron by an experiment with falling oil drops and got an answer which we now know not to be quite right. It’s a little bit off, because he had the incorrect value for the viscosity of air. It’s interesting to look at the history of measurements of the charge of the electron, after Millikan. If you plot them as a function of time, you find that one is a little bigger than Millikan’s, and the next one’s a little bit bigger than that, and the next one’s a little bit bigger than that, until finally they settle down to a number which is higher.

            Why didn’t they discover that the new number was higher right away? It’s a thing that scientists are ashamed of—this history—because it’s apparent that people did things like this: When they got a number that was too high above Millikan’s, they thought something must be wrong—and they would look for and find a reason why something might be wrong. When they got a number closer to Millikan’s value they didn’t look so hard. And so they eliminated the numbers that were too far off, and did other things like that.

            If you want a slightly more recent example, see the evolution of the value of Hubble’s constant in this post: https://muircheartblog.wordpress.com/2018/07/08/the-truth-the-whole-truth-and-nothing-but/

            Here’s David Colquhoun on the corrosive influence of your type of blind faith in the validity of citation metrics: https://nfais.memberclicks.net/assets/docs/Workshops/david%20-%20metrics-nfais-webinar-011218.pdf

            As regards anomaly vs statistical blip, I’ll get back to you on that soon. But I’ll note for now that “They did it too” is never a particularly compelling response…

            Like

            1. As I wrote at the beginning, a perfect bibliometric index is not needed: bibliometric indices positively correlated to scientific merit are enough. The issue is analogous to measuring an energy spectrum with an imperfect reconstruction of energy: it’s enough to see features on scales larger than the energy resolution.

              Bibliometrics can be criticised, but to put things in perspective one should also criticise a workshop that made strong claims based on anecdotal evidence and avoiding inviting experts. Different methods are used in the literature, and the review published in (https://www.psychologicalscience.org/pdf/Women-Academic-Science.pdf) writes: “the overall picture is one of gender neutrality in GEEMP fields [P is physics], notwithstanding frequent claims to the contrary”. This is what I found. It’s a big problem that scientific institutions push so heavily on politically-biased victimisation narratives. In the US this problem is being addressed by a new law that cuts funds to institutions that fail to protect freedom of speech.

              [Edit PM 27/03/2019 08:36 — Please see further down this thread (March 27 2019, 08:18 am) for my response to this comment.]

              Like

            2. Alessandro,

              This “anomaly” vs “statistical blip” issue is intriguing. You’re right, it’s a side issue in terms of our wider discussion but it’s still an interesting insight into the behaviour of different communities of physicists. (There’s those social dynamics again…)

              I’m not a particle physicist, as you know, but we certainly use particle accelerators in our work. Here’s an example of one experiment we did a couple of years back: https://www.youtube.com/watch?v=tWWyAklkYqM (which has yet to be published because we cannot get good agreement between experiment and theory and we need more beamtime to resolve the discrepancies.)

              If, as in the case of the 750 GeV “blip”, the experimental signal turns out to be noise, do you, and others in your community, retract the papers that claimed that the spurious peak was a true signal and interpreted it accordingly? Or do you leave those papers in the literature, knowing full well that the experimental data were just noise?

              Retracting a paper is never easy — I speak from experience: https://pubpeer.com/publications/DBC836E90EFB5926817D2B2A98B55C — but don’t you think we have a duty to remove work from the literature/scientific record if it’s subsequently shown to be based on incorrect assumptions/noise?

              Liked by 1 person

            3. You ask when papers should be retracted. In my community papers about disappeared anomalies are considered irrelevant, but not incorrect, so they are not retracted. They just stop being cited. In my solo papers I use to add a note “it’s dead” on the arXiv version, but dealing with journals and collaborations is more complicated.

              This is part of the bigger problem. Studies of physics beyond the Standard Model degenerated in speculations, because the SM works so well. Should thousands of papers about supersymmetry or extra dimensions at the weak scale be retracted given that such speculative physics has not been found? Or maybe other theoretical speculations that cannot be tested (for example about physics at the Planck scale) are even more irrelevant than testable excluded speculations? In my opinion it’s reasonable to try proceeding in this way, as long as we recognise that it’s not the “normal” way of making progress in physics. It’s just what can be done in the present difficult situation. In the past it gave something intersting.

              Like

  2. Alessandro,

    From your comment above,

    “Female students are majority in many faculties: do you claim that it’s because these faculties discriminate against male students? If not, why attacking physics and STEM for the opposite reason? Physics departments are equally open to everybody: men are not invited with limousines and women don’t need to wear moustaches to pass fences. On the contrary, many departments deliberately worked to attract more female students. However, less women choose physics in countries with more freedom of choice, and gender differences in interest are growing in Scandinavian countries that worked to remove socially constructed gender differences.”

    You have an exceptionally simplistic/”black and white” understanding of (gender) bias, Alessandro. Do you read any of the links I provide? Or is it yet again a case of, as Philip Ball put it so very well (see https://muircheartblog.wordpress.com/2018/07/08/the-truth-the-whole-truth-and-nothing-but/) “they’re damned if they are going to let your actual words deprive them of their right to air their preconceived notions“?

    I sent you the Twitter thread below yesterday. Did you read any of the articles linked therein?

    As regards that so-called Norwegian paradox (that is raised with tedious regularity in certain circles [1]), see, for example, https://capx.co/what-jordan-peterson-gets-wrong-about-the-nordic-gender-paradox/

    [1] It’s like a Godwin’s law for gender bias debates: “As an online discussion about gender differences grows longer, the probability of mention of the “Norwegian paradox” approaches 1″.

    Like

  3. @Alessandro Strumia (March 26 2019, 3:02 pm)

    (The WordPress theme I’m using only allows a certain number of “nested” replies, hence the “@” to identify the reply to which I’m responding.)

    As I wrote at the beginning, a perfect bibliometric index is not needed: bibliometric indices positively correlated to scientific merit are enough. The issue is analogous to measuring an energy spectrum with an imperfect reconstruction of energy: it’s enough to see features on scales larger than the energy resolution.

    …and as I wrote at the beginning of our email exchange, Alessandro, just how imperfect is that bibliometric index? Can you quantify the uncertainty? Of course you can’t. And if you’ll excuse my lifting Pauli’s words ever so slightly out of context, if you can’t quote your uncertainty values, you’re not even wrong.

    Thank you, however, for the wonderful analogy of the energy spectrum. This is perfect for our purposes. I’ve done a lot of photoelectron spectroscopy in my time and even when I was in the early days of my PhD what would wind me up a great deal is the following…

    Spectra comprising nothing more than a broad featureless Gaussian peak (say ~ 2 eV FWHM) would often be published. And yet the authors would claim that they could “fit” that peak with any number of much narrower Gaussian (or, worse, Lorentzian-Gaussian convolution, i.e. Voight) peaks (arising from different chemical/oxidation/electronic states at the surface of the sample). And of course they could. Indeed, they’d get an excellent fit. A completely physically meaningless fit, but an excellent fit nonetheless…

    This is analogous to your exceptionally naive “fitting” of the citation data to extract what you credulously believe is a relationship between IQ, gender, and citations. You claim good agreement between your “model” and the data. Anyone can get “good agreement” on the basis of the fitting parameters they choose/fudge. It’s whether those fitting parameters have any physical significance that’s the key thing.

    And, once again, here’s that generic piece of exceptionally important advice: correlation is not causation. I hope you enjoyed the “cheese consumption vs # of engineering degrees awarded” example from http://www.tylervigen.com/spurious-correlations in one of my earlier comments. Here’s another good one: http://tylervigen.com/view_correlation?id=3857 (A correlation coefficient of 0.9938. Wow. There must be something in it, right?)

    It’s a big problem that scientific institutions push so heavily on politically-biased victimisation narratives. In the US this problem is being addressed by a new law that cuts funds to institutions that fail to protect freedom of speech.

    Alessandro, please focus on the science. Although it’s amusing that you keep whining about victimisation narratives while all the while you complain about being the victim of the nefarious “PC orthodoxy/cultural Marxist” forces, it gets us nowhere. As I’ve said elsewhere:

    “… what I have always found striking throughout my time online is that those who claim vociferously to be solely driven by reason, logic, science, rational debate, and/or individualism — aka the Fuck Your Feelings (FYF) brigade — are often among the most hypersensitive, overwrought, tribal, and emotionally driven out there. The type of over-emotional response that the FYF tribe attribute to the big, bad bogeyman of “The Left” is equally, and often more, prevalent within their own ranks. (There are key parallels here with the deeply intolerant patriotic correctness of the right.)”

    Like

  4. @Alessandro (March 27 2019 at 1:40 pm)

    Should thousands of papers about supersymmetry or extra dimensions at the weak scale be retracted given that such speculative physics has not been found?

    That is not at all the point I was making, Alessandro, as I believe is entirely clear from my comment. My point was that the 750 GeV `peak’ is now known to be nothing other than a statistical blip. It was noise that went away when more data were acquired. If our group here at Nottingham were to acquire X-ray absorption, (resonant) photoemission, Auger electron spectroscopy, or X-ray standing wave data at a synchrotron which we subsequently could not reproduce — or, more embarrassingly, became aware that we had misinterpreted noise as signal — we would retract the paper. Why shouldn’t this be the case for the particle physics community?

    It’s not at all about speculative physics not being found. It’s that noise was misinterpreted as signal. There’s a huge difference in those scenarios.

    Like

    1. I would not say “noise was misinterpreted as signal”. I would say: an anomaly (later turned out to be noise) was tentatively interpreted as signal, writing “more data is needed”. From a scientific point of view, papers about a γγ resonance at 750 GeV are now as irrelevant as papers about a gluino at 750 GeV. Both papers have been written because we lack more promising topics.

      Like

      1. “…later turned out to be noise…”

        No, it was noise from the start, Alessandro. Let’s not play wholly transparent semantic games here. You didn’t know it was noise from the start but that’s irrelevant — you know it’s noise now. So why aren’t those papers based on nothing more than a misinterpretation of noise retracted from the literature?

        If a PhD researcher based a thesis on the analysis of noise should they be awarded the PhD?

        All I know is that if our group embarrassingly misinterpreted noise as a real signal we would retract that paper as soon as we knew. Why should it be any different in particle physics? Noise is noise.

        Like

        1. Some theoretical papers about anomalies that later disappear propose ways to clarify the issue, and ideas that can be applied to other contexts. I don’t think that retracting such papers is a useful symbolic action. I think it’s more useful what I did: while giving a seminar about the 750 GeV anomaly I mentioned rumours about it, rather than pretending that the anomaly was still alive until the official announcement. This concretely helped colleagues who avoided wasting one more month on a statistical fluctuation.

          Like

          1. “I don’t think that retracting such papers is a useful symbolic action”

            Well, as I’ve said before, it all depends on where you set the bar for scientific evidence, Alessandro. I would strongly argue that papers that involve an analysis of irreproducible data should be retracted. (That low frequency hum you can hear right now is from Feynman turning in his grave at the idea that irreproducible measurements should be published.) Indeed, about the only place they should appear is here: http://www.jir.com/

            But then, of course, those oh-so-important citations to an analysis of irreproducible data would be null and void…

            Like

            1. I think that your “null and void” conclusion is exaggerated, because dropping papers about anomalies (a small fraction of papers) would not change conclusions about gender . Furthermore, an improved bibliometric index based on the Page Rank algorithm (used by google) has a built-in mechanism that reduces the weight attributed to papers about topics that have been later abandoned: the same gender differences are found using this index.

              PS Max Born said “science advances one funeral at a time”, a sentence appropriate for the grave of the 750 GeV anomaly. Let’s stop exhuming it. It’s not useful .

              Like

  5. Robert L Park
    Alas, to wear the mantle of Galileo it is not enough that you be persecuted by an unkind establishment, you must also be right.

    But Galileo was right. And those that rejected his thinking were wrong. But they could not see there own model as wrong despite many anomalies. They could not see the flaws in their own model despite the fact that, in hindsight, these anomalies seem obvious to us now. Those who invoke Galileo don’t do so because they wish to be treated kindly. They do so to point out how easy it is to believe you understand when really you just believe.

    It’s not what you know that is the problem.. It’s what you think you know that just ain’t so.

    Are You Confused About Hydrogen Bonding In Water?

    (Read the comment stream between myself, Professor Dave and Robin Bedford)

    James McGinn / Self Declared Genius

    Like

    1. James,

      The point is that those who whine about The Big, Bad (PC) Establishment shutting them down fail often to recognise that there’s a much simpler explanation.

      “They do so to point out how easy it is to believe you understand when really you just believe.”

      Precisely. And that’s exactly the point that Park was making. “It’s not enough to wear the mantle of Galileo…”

      As regards YouTube comment streams, I hope you’ll excuse my scepticism about the quality of discourse therein: https://imgs.xkcd.com/comics/youtube.png

      There are some occasional exceptions, admittedly: https://muircheartblog.wordpress.com/2018/07/08/the-truth-the-whole-truth-and-nothing-but/

      Philip

      Like

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s