Dubious physics articles cast doubt on new publishers...

In my last couple of posts I've been sharing some observations on the state of science communication these days, particularly problems with predatory publishers, overzealous publicity for ambiguous research, and the difficulty for an outsider in understanding what to trust in what's published in science these days. Bottom line: it's become much easier to communicate, but harder to know whether what's being communicated is worth paying attention to.

In this post I present two more cases of dubious publication - as these are in physics I'm rather more certain they are wrong. These have received a rather mixed collection of "post publication peer review" but in a fashion that I believe would leave the non-expert quite unaware they are not useful contributions to science.

Our first case study here is from the mega-journal PLOS ONE - "Implications of an absolute simultaneity theory for cosmology and universe acceleration" by Edward T. Kipreos, of the University of Georgia. PLOS ONE is an open access journal started in 2006 (2012 impact factor: 3.73); it is the world's largest journal by number of papers published, publishing original research across many fields. Published articles are reviewed for technical validity but as long as they pass that, it doesn't matter how important the reviewers or editors feel the article is. PLOS ONE doesn't publish many physics articles on the whole, but it has done a few along the lines of the above, for example this article that seeks to rework general relativity - probably equally dubious but one I have less background to assess.

What are we to make of Dr. Kipreos' article? Univeristy of Georgia seemed rather pleased with the publication. It also notes he is a molecular geneticist, not a physicist. Hmmm. Some version of the press release (with no comment from any actual physicists) spread quickly around the internet; the article lists 8522 "views" on the PLOS platform right now. Despite the promise of "post publication peer review", there is only one incomprehensible comment on PLOS ONE itself, and nothing in PubMed Commons or PubPeer.

I did track down two apparently knowledgeable critiques in blog form from physicists Brian Koberlein and Matthew R. Francis. There is no sign of any response or acknowledgement from Dr. Kipreos of the problems they raise.

As Dr. Francis notes, articles "proving" Einstein was wrong are extremely popular among the less-well-informed. I recall when I was a young man, probably about 12 after having read some account of special relativiely probably by Asimov, there was something that occurred to me that I thought I saw clearly, everybody must have overlooked it! I'd be famous, get a Nobel, etc. etc. When I tried explaining to my Dad (a chemist) he rather patiently suggested maybe I needed to study a bit more about it. Sure enough, when I understood it more, my insight had already been long accounted for.

Sadly, there are a few people who never seem to grow out of that stage of certainty that they've discovered something simple that others have missed. Physics journals routinely receive these "crackpot" papers - I've heard something like one in five papers submitted to journals that cover gravitation and relativistic physics are in that category. Their editors can spot these papers a mile away.

And Dr. Kipreos' paper is clearly one of them. The first sign of trouble is the very limited use of math. The only mathematical expressions appearing are simple algebraic equations in scalar variables. Anybody tackling special relativity these days would need to address the impact of their theory at least on electromagnetism and other foundational theories that have been built on top of it over the last century. Kipreos talks about light but there is no deep analysis supporting his assertions about it; in fact his theory appears to require light to move (in one direction) faster than "c" at times, and slower than "c" in return, so the average comes out to "c" - that would greatly violate several foundational principles that have been accepted now for over a hundred years.

A second bad sign is the list of references - there are certainly a respectable number of them, but the only references on theoretical questions are very old (1940s and earlier), and there are rather a lot of what look like textbooks or popular summaries. The author does find a lot of experimental articles to reference and discuss, but it's not clear he really understands how to reconcile his theory with what's already well understood.

In particular, his discussion of the GPS satellite relativistic effects is just wrong. The author references a good review article by Neil Ashby but seems not to have understood it. Figure 2 in Ashby's article illustrates clearly the complexity, showing that for close orbits (like the space shuttle) special relativistic time dilation dominates (clocks run slower on the shuttle) but at GPS and Geostationary distances, the difference in gravitational redshift dominates, so those clocks actually run faster than clocks at Earth's surface. Ashby documents five different sources of relativistic effects contributing to time-keeping differences in orbit and on the surface. Note the contrast between the mathematical complexity and sophistication of Ashby's account with the naivety in Kipreos'.

For reference here is Kipreos' completely incorrect discussion of GPS time "dilation" (as noted, GPS satellite time is not dilated, but runs FASTER than Earth surface time due to our time being much more dilated by Earth's gravity than time at the GPS orbital distance):

Satellites of the global positioning system (GPS) are in inertial reference frames because they are in free-fall orbits around the Earth, similar to the inertial reference frame of the ECI that arises from its free-fall orbit around the Sun. It is well established that the ECI functions as a PRF for GPS satellites, with the satellites experiencing directional time dilation based on their velocity relative to the ECI [16]. Clocks on GPS satellites undergo time dilation of ~7 µs per day relative to the Earth's surface, which is calculated by applying the Lorentz/ALT time dilation formula independently to the speed of the satellite relative to the ECI and to the speed of the Earth's surface relative to the ECI [17]. Correcting for the Lorentz/ALT time dilation is essential for proper positioning in the GPS system, as the 7 µs/day difference translates to a localization error of 2.1 km per day [17]. The Sagnac effect, which is important for the communication of GPS satellites with rotating ground-based receivers, is irrelevant to the time dilation experienced by the satellites as they move relative to the non-rotating ECI [16]. The communication between GPS satellites and ground-based clocks continuously reveals the absolute and directional nature of the time dilation.

More than all that, the whole logic of the paper is simply incoherent. How can you have independent "preferred reference frames" at the center of every major gravitating body? Kipreos starts (in the introduction) by claiming his ALT is "describing absolute simultaneity and invoking a preferred reference frame (PRF) relative to which time dilation and length contraction occur in a directional manner" - you can only have ONE such preferred frame, not hundreds of billions (in our galaxy) times many hundreds of billions more (all the other galaxies out there). The whole concept seems almost pre-Copernican, putting the Earth in a special place in the universe. He claims not to be doing this, but he shows no indication of how simultaneity between two different "PRFs" can be reconciled with one another.

So the paper is clearly, almost certainly, complete nonsense. Yet it appeared in a major journal that does practice peer review, and received considerable publicity and thousands of "views" - far more than the refutations from Koberlein and Francis are likely to get. While it's not likely to have any substantial impact on physics research, incidents like this bode ill for the trustworthiness of science as perceived by the public.

The second example I'd like to discuss is from a very new entrant in scholarly publishing - The Winnower. Perhaps it's a little unfair to criticize as it's still labeling itself as "BETA", but I think there are some lessons to be learned here. The Winnower is another attempt to change peer review to a post-publication system, perhaps the most explicit attempt so far along those lines. Basically anything submitted to the journal goes online and then is opened up for "review"; the author can invite people to review their paper, and others can also post reviews; the journal encourages a back and forth, and also allows explicit (0 to 5-star) ratings from the reviewers for "quality" and "confidence". In one place a $100 fee is mentioned, but elsewhere it claims to be "free"; the fee might be a good idea to set at least a minimum bar for seriousness on the part of the author. The journal has published at least a few dozen articles so far, mostly in biosciences.

So how is the post publication peer review process working? The vast majority of articles seem to have zero or at most one review; frequently the only review is actually a follow-up comment of some sort from the author. But right on the front page, highlighted as "most reviewed", we have Fundamentals of Relativization by Hontas Farmer (the author is a physicist affiliated with several Chicago-area community colleges). Farmer has a couple of other papers in The Winnower as well, one listing 5 reviews, and another where the only review is a comment from the author. The 10 reviews on the "most reviewed" paper are by three separate individuals, only two of whom seem to have provided ratings, leading to an overall 1 star (out of 5) on the top. So - the reviews are negative. But they are from a limited set of people - do you trust the reviewers, or the author to be right, or just leave as undecided?

Looking at the paper itself, it's not obviously wrong in the same ways that Kipreos was. There's plenty of math, references seem suitable (though on the previous, first paper in the series there were only 4 references listed, one of which was Einstein's original General Relativity paper). The author isn't claiming Einstein was wrong at all - to the contrary, the approach is to try to develop a new version of quantum mechanics that somehow fits better with General Relatvity. And that may be the main clue there is likely to be a problem here: uniting quantum mechanics with general relativity (quantum gravity) is probably one of the toughest problems in physics today. It's highly unlikely a solution will appear in just a handful of pages of math and text - some deep conceptual problem needs to be better understood. There is no sign of that deep insight here; rather the author is claiming something that seems almost obvious (as far as I can tell) - that one can make some minor adjustments to quantum theory to get it compatible with general relativity. It's not clear that the approach taken here actually does what it claims (the reviewers express many doubts). The idea itself in a general sense is certainly not new - there has been work on quantum fields in curved spacetime for many years. A simple exapmle is the Unruh effect, known for about 4 decades now.

Aside from what the "reviewers" talked about, an issue that concerns me is the problem of time. As noted in this arXiv paper, there have been a huge number of approaches attempted in trying to reconcile the time evolution of quantum theory with the time embedded in spacetime in general relativity. Farmer touches on none of this complexity - and I see no mention of time dependence at all in the paper, other than a claim that the approach allows for dynamics. Is this relativized quantum mechanics timeless like the Wheeler-DeWitt equation? Whether or not the author has anything substantially right, if that's true this all seems rather useless.

Farmer's claimed insight seems about the same caliber (though with more complex math) as the naive insight of my early years, which evaporated on further learning. I hope this author will put in that effort and reassess this work in time. In the meantime we are left with a host of negative reviews and protests from the authors that the reviewers didn't understand something or other. This could be rather confusing to the outsider. It's not clear to me what The Winnower is planning for articles like this one, that have been so negatively reviewed. The intention appears to be to archive it permanently - it has a DOI, it's essentially "published". Farmer's most recent article seems to have been just ignored, which is perhaps the best outcome, but still it seems unfortunate to leave something like this in such an ambiguous state. Situations like this will necessarily erode trust in this journal and the whole idea of post publication peer review. Is there a better solution?