This week, PubPeer filed a motion to quash a subpoena demanding that they reveal the names of some of their commenters. Here’s another installment of PubPeer Selections:
- “Papers like this one need to be judged against high standards because of the large audience of vulnerable spinal-injury patients desperately seeking hope for their paralysis.” Commenters critique a recent Nature paper.
- The discussion about a theoretical physics paper becomes heated.
- Commenters raise questions about some of the cell identification techniques in a Nature paper on lung regeneration.
- A paper claiming that “fears can be inherited through the generations,” according to a Nature News story, has come under scrutiny in various quarters.
- “Aren’t we scientists supposedly educated to deal with probabilities rather than 100% certainties?” Commenters debate whether images in a Cell Reports paper have been spliced.
I wish to propose one prototype for PPPR.
A first suggestion of how PPPR could be conducted effectively if a micro-topic is focused on. It is evident that there are different grades of knowledge among peers, but one effective way in which I have found PPPR to work best is by analyzing a micro-topic. For example, here I exemplify how PubPeer is serving the PPPR cause well, using Anthurium tissue culture as an example. This allows for the literature to be broken down into more micro-manageable sections, and analyzed by “absolute” peers. This gradation explains how (in increasing levels of specialization):
plant science > horticultural science > ornamentals > Araceae > Anthurium
The gradation could also be:
plant science > biotechnology > plant tissue culture > ornamentals > Araceae > Anthurium
Thus, this simple two-prong approach will allow true peers with ultra-specialized skills to manage certain sectors of the literature. It takes time, it is exhaustive, you are not remunerated, nor praised. If anything, this approach increases enemies, conflicts and stress. However, this is the “ugly” phase of PPPR that is essential to correct the literature. Of course, correcting the literature does not rely exclusively on the efforts of those who conduct PPPR, and report it, but it requires the desire by the authors, the editors, the journals and the publishers involved, to see – and appreciate – the errors (or worse), and then to collectively correct the literature. By doing so, future students and academics will benefit from a corrected literature that has been revised, not by those who are “inside the system”, but by experts who can independently judge aspects of the literature from “outside the system”.
In this prototype PPPR experiment, which has taken more than 2 years to achieve, PubPeer has served by providing a formal public record, with a formal number + website, equivalent to – and linked to – a DOI. This is the really important aspect, in fact, that allows for centralization of information. Errors and problems with PubPeer aside, including unhelpful comments by some commentators, or snide comments by others, or even trolls, PubPeer serves an extremely important function to solidify PPPR, at the end of the meta-analysis.
To exemplify how this prototype can then be used, I/we have divided the literature, which exceeds 100 published papers (Chinese literature excluded), into two main categories: A, which includes papers with greater potential ethical concerns, such as duplications, and B, which include errors in scientific aspects, which may or may not invalidate the conclusions of that study, but which may certainly affect its reproducibility.
A very important note here: such comments must never be recommendations for errata, expressions of concern or retractions, because that decision will ultimately (in some cases, unfortunately) not depend on the will of the PPPR reviewer. It will depend on the rules and decisions made by the editors, journals and publishers, in the presence or absence of authors’ desire to correct the literature. So, the PPPR itself is extremely complex, but the remaining process is no easy task either.
So, getting back to the Anthurium tissue culture literature, here go the PubPeer entries for category A:
https://pubpeer.com/publications/2FA8EA21F3078386552B1B4A71C6C3#fb17166 (with one retraction resulting from PPPR reports)
https://pubpeer.com/publications/773785209C305926E01A2F28265C3F#fb18158
https://pubpeer.com/publications/625423731873AA049FEAD97C824708
https://pubpeer.com/publications/AFD78E3F00791586BA5D65DDC24E2F
https://pubpeer.com/publications/7D433CBB6E3CF846EF798FA747B0F0#fb16726 (which has a separate case addended, which resulted in a stealth retraction resulting from PPPR reports)
https://pubpeer.com/publications/5740E043753A1CEC14D4494FD2EAA8#fb18313
And for category B:
https://pubpeer.com/publications/D3F78C630ED61FAAD1083C8925D90D#fb18201
https://pubpeer.com/publications/62C91250EC6834E221A129EFE40F88#fb18205
https://pubpeer.com/publications/5356319CAAB66A911B78574FF2711C#fb18264
https://pubpeer.com/publications/40F0D7CDEE1BE9C43B9008E7819877#fb18267
https://pubpeer.com/publications/40B0C68EA97A393602B9255C7C651D#fb18548
https://pubpeer.com/publications/FE35CECA0454A5577B9693F5C4B009
There is of course, the possibility of conducting PPPR on an individual paper basis, on an author basis, or on a journal-by-journal basis. The later type is now in an experimental phase and a report will be made at RW in 2015 on that.
The heated discussion in theoretical physics is generating a lot of heat, for sure, but it may be hard to separate signal from noise.
It concerns a paper which has just appeared in “International Journal of Theoretical Physics”. It is behind a paywall but here is the preprint: http://arxiv.org/abs/1211.0784 I have submitted a note http://arxiv.org/abs/1412.2677 to the same journal pointing out a major flaw in the paper (and referring to two other already known major flaws). The journal is published by Springer. Its editorial board consists almost entirely of very big names indeed in physics; four of them have Nobel prizes. (Most of the editorial board are men in their 70’s or 80’s but of course this does not mean that they are no longer active scientists).
The paper discusses a non-quantum Aspect-type experiment (Bell type experiment) which the author believes will exhibit quantum correlations. I point out that Christian’s experiment will *certainly* disprove his own claims. Those who know Bell’s inequality and read my note will know why, but clearly many readers of PubPeer and of Retraction Watch are unaware of the context. To explain, I must recap some (elementary) technicalities of Bell’s theorem (1965) and the CHSH inequality (1969). You may want to Google these terms. Here is a recent Nature article: http://www.nature.com/news/physics-bell-s-theorem-still-reverberates-1.15435 and here is the wikipedia article: https://en.wikipedia.org/wiki/Bell%27s_theorem
Christian describes an experiment which is supposed to determine correlations E(a, b) between the spin of particle 1 measured in direction “a” and the spin of particle 2 measured in direction “b”. There are “N” pairs of particles and their spin (+/- 1) is measured in many pairs of directions “a” and “b” (represented as unit vectors in R^3). Christian believes that his experiment will result in the so-called singlet correlations E(a, b) = – a . b = minus the the inner product between the two vectors “a” and “b”. Suppose we look at four correlations, based on the four combinations between two particular choices for a and two particular choices for b. I choose them all in the same plane and refer to them by giving the angles they make with some fixed reference direction in the plane. Take a_1 = 90 degrees, a_2 = 0 degrees, b_1 = 45 degrees, b_2 = 135 degrees. Define E_ij = E(a_i, b_j). You will notice that each E_{ij} = +/- 1/sqrt 2. Three of the correlations are negative, one is positive. E11 – E12 – E21 – E22 = 2 sqrt 2 > 2. However as I show in my note, Christian’s experiment will deliver |E11 – E12 – E21 – E22| less than or equal to 2.
Christian’s first “disproof of Bell’s theorem” appeared on arXiv in 2007. It initially caused some interest and some controversy, but has since been more or less forgotten. Since then he posted about 15 papers elaborating this original result on arXiv. Many of them are chapters in his book “Disproof of Bell’s Theorem: Illuminating the Illusion of Entanglement” http://www.amazon.com/Disproof-Bells-Theorem-Illuminating-Entanglement/dp/1599425645 ; for an eBook, second edition, go to http://www.brownwalker.com/book/1612337244 . This is the first time, as far as I know, that he succeeded in getting one of these papers published in a peer reviewed journal, which is why it is interesting.
Incidentally, the CHSH inequality is due to Clauser, Horne, Shimony and Holt (1969) and Abner Shimony was Christian’s PhD supervisor. Christian used to have impeccable research credentials. They brought him research fellowships at Perimeter Institute and at an Oxford college, a research grant from FQXi, and a lot of highly connected friends. (He is a charming person, by the way!)
I first heard about Christian’s disproof of Bell’s theorem at a conference in Berlin in about 2008, at which Christian gave a talk on his work. Since he obtained his theoretical results by redefining “correlation” in a non-standard way, using geometric algebra, it was clear to me that it was quite simply *irrelevant* to the ongoing discussions on Bell’s theorem, but his use of “geometric algebra” seemed amusing. I decided to learn about this, it might be useful. I met Christian in Oxford and he explained to me a little bit. Then I discovered that not only was his approach irrelevant, it also depended on an elementary sign error, carefully hidden (through notational ambiguity) in a routine calculation.
These same issues had been discovered independently by others working in the field of quantum foundations, and confirmed by anyone who took the trouble to check, and Christian’s work has since been totally ignored, as far as I know.
Later still the arXiv preprint of the paper presently under discussion appeared. I got involved in discussions at several internet fora on this paper. It seemed to me that at last it should be possible to explain to the most fervent supporter of Christian that his work depends on one elementary error after another, since the new paper contains an error which anyone can see for themselves by doing one line of elementary algebra. No need to learn about geometric algebra, no need to learn about bivectors and Clifford algebra and differential geometry, no need to learn anything subtle about Bell’s inequality. But of course, nothing will shake the cherished beliefs of the true believers.
All of this is interesting from a psychological and sociological point of view: why is Bell’s theorem so poorly understand? Why does it constantly attract well-meaning and intelligent people who come up with flawed “disproofs” again and again? Often, these disproofs catch the attention of the popular science media. For instance, I became aware of Hess and Philipp’s work, published in PNAS in about 2004, through my local newspaper. Hess and Philipp are (were) two highly regarded US scientists, stepping a little bit outside of their usual fields. Karl Hess is still not giving up … http://www.amazon.com/Einstein-Was-Right-Karl-Hess/dp/9814463698/
I sincerely hope that retractionwatch will not let this evolve into a similar thread as the one on PubPeer. If that is the type of PPPR they (=PubPeer) approve, count me out. I have to say that the nastiness comes primarily from one side, and it is the same side that argues that if a journal publishes a paper, (all) its Editorial Board members have endorsed the paper. Uhm…no.