Here's a video of a talk I gave in Tromso a few weeks ago about the state of academic publishing, and about why a system whose flaws are obvious to almost everybody is as robust as it is.

+Mark C. Wilson reports on his eventual success, after three years of patient effort, to obtain information about the spending by New Zealand universities on Big Deal subscriptions with the major commercial academic publishers, and draws some conclusions from what he has discovered.

A newspaper article that sets alarm bells ringing

I don't mean in the obvious sense that earthquakes are damaging. Rather, I think there is quite a good chance that the research that it describes is nonsense -- though I haven't read the scientific article itself, so it could be just the way the journalist has presented it that gets me suspicious. Anyhow, here are the passages that bother me.

In their study, Bilham and Bendick looked at earthquakes of magnitude 7 and greater that had occurred since 1900. “Major earthquakes have been well recorded for more than a century and that gives us a good record to study,” said Bilham.

They found five periods when there had been significantly higher numbers of large earthquakes compared with other times. “In these periods, there were between 25 to 30 intense earthquakes a year,” said Bilham. “The rest of the time the average figure was around 15 major earthquakes a year.”

The researchers searched to find correlations between these periods of intense seismic activity and other factors and discovered that when Earth’s rotation decreased slightly it was followed by periods of increased numbers of intense earthquakes.

And then later on:

Exactly why decreases in day length should be linked to earthquakes is unclear although scientists suspect that slight changes in the behaviour of Earth’s core could be causing both effects.

The increases in day length are by about a millisecond. For all I know, that's huge on some sensible scale, even if it sounds small. But it does sound as though the researchers looked at a lot of potential correlations and managed to find one, and that is often a good way of finding a correlation that has nothing to do with causation.

I almost wrote "I hope I'm wrong about this," but of course that's not true: for the sake of potential victims, I hope I'm right about it.

Last weekend I took two of my children to the Pompidou Centre in Paris. The photo below is most of a picture displayed there. It consists of four square grids of equal sizes superimposed on each other. They are rotated by 0, pi/8, pi/4 and 3pi/8. (Note that rotating by pi/2 takes you back to the original grid). That doesn't uniquely determine them, but it does up to translation. I'm not sure how the artist decided what translates to take, or even if any systematic decision was taken. Anyhow, it's fascinating to see the pattern it makes, with little "flowers" surrounding points that happen to be close to points of all four grids. It's Diophantine approximation turned into art in a surprisingly simple (conceptually at least -- actually creating the picture must have been a technical challenge) way.

The painter is François Morellet (1926-2016), the painting was painted in 1958, and it is called "4 double trames, traits minces 0 - 22.5 - 45 - 67.5" (where those numbers have degree symbols after them, which I don't know how to write here). For those whose French is rusty, that means "4 double grids, thin lines".

Hat tip to +David Roberts for the information that a new diamond OA journal is starting up. I shall certainly submit to it when I have a suitable paper (and a coauthor who doesn't mind submitting to a journal that doesn't yet have things like impact factors -- unfortunately some people find themselves in countries where a publication that isn't on some approved list of journals is treated as though it didn't exist). The journal is called the Annales Henri Lebesgue. It aims to be a high-quality generalist mathematical journal, and it has a high-quality, and predominantly young, editorial board. It also has plenty of financial backing -- the founders clearly mean business. They give the following reasons for submitting to their journal.

1. Publishing in journals such as the AHL will contribute to the success of such open and ethical journals.

2. The journal is and will remain free of charge for authors and readers.

3. Authors retain full intellectual property rights over their work.

4. The AHL is a peer-reviewed, generalist Mathematical journal addressing a wide readership. The AHL aims to select the most beautiful mathematical results through a rigorous and rapid editorial procedure. If preliminary reports are favourable (to be notified within a month of submission), the journal aims to obtain a full report within five months of submission. The final decision is taken collectively by the Editor in Chief and at least two other editors.

5. The journal is committed to ensuring that the editorial board represents all fields of mathematics by regular renewal of the board.

6. The journal will ensure the distribution and conservation of its papers using public platforms.

7. The journal is supported by the CNRS, learned societies and four internationally-recognized research institutes.

8. The journal is run by mathematicians for mathematicians.

9. The journal's vocation is to support similar initiatives and journals, certainly not to engage in competition with them. The AHL is convinced that there is enough beautiful mathematics for all free journals; many journals based on this model are currently overwhelmed by submissions.

10. Submitting to the AHL is easy!

I would add that I thoroughly agree with their point 9. I am keen that Discrete Analysis, which is run on similar lines, should be a success, and I see the emergence of other journals like this as a huge help rather than a hindrance. (It's a little bit like the well-known phenomenon where you get lots of similar shops in the same parts of some towns, and this turns out to make good economic sense.)

If we had sensible systems of dissemination and evaluation of academic output, it would be easy to rectify mistakes. As it is, we have to make do with things like this ...

Here's another interesting division of the plane into regions, again spotted in a French museum -- this time an old-fashioned natural history museum full of skeletons and body parts, including exhibits that are not for the faint-hearted. This photo is of part of a glyptodon clavipes, which lived between 780000 and 13000 years ago (I am told). They were giant mammals and lived in South America and belong to the same family as sloths.

What interests me is that the pattern looks rather like a Voronoi diagram (what you get if you scatter some points in the plane and divide up the plane according to which point is nearest). I'm also interested in how, like the superposition of four grids, it looks somewhat systematic and also somewhat random. I know people, including Turing, have thought about mathematical models for how biology creates patterns like this: I'd be interested in what they had to say about this one. I tried Googling "random Voronoi diagram" but those, like fully random collections of points, tend to have ugly clusters. Here there seems to be some mechanism that makes the centres of the cells more evenly spaced than they would be if they were purely random.

Now Springer follows the trail blazed by Cambridge University Press ...

Springer Nature, whose publications include Nature and Scientific American, acknowledged that at the government’s request, it had removed articles from its mainland site that touch on topics the ruling Communist Party considers sensitive, including Taiwan, Tibet, human rights and elite politics.

The publisher defended its decision, saying that only 1 percent of its content was inaccessible in mainland China.