Over the last week or so, the world of science became enchanted by a pair of obscure preprints claiming the synthesis of the first room-temperature ambient-pressure superconductor LK-99. So far, the story has everything you might want in your Netflix drama: a breakthrough claim, scientific outsiders, sceptical experts, in-fighting among authors, a race to replication, demands for retraction… What time to be alive!

So far, there are several good scientific arguments to be made against the claim of superconductivity in LK-99. The primary one is that the magnetisation curves do not display a phase transition, hence don’t add up with the I-V data if it’s a superconductor. There are also good arguments of “statistical” nature, for lack of a better word. First, claims of room-temperature superconductivity have been made and discredited before. Second, the authors are not part of the scientific mainstream, hence less likely to know the state-of-the-art tools and methods for analysing potential superconductors.

And then there’s the third argument used to dismiss LK-99, the good old ad hominem. Because why bother engaging with the claims, if what is written can be rejected simply based on how it’s written?

The preprint stands little chance against the grammar gang.

For example, the words “Volt” or “Volts” are used instead of “voltage” in some figures.

The X-ray diffraction plot is badly squashed

The paper is written in MS word for crying out loud!

Admittedly, the more you read, the less pretty is gets

Oh hey, why not switch language sometimes?

These things are amusing, but I feel really uncomfortable with how many scientists on Twitter and Linkedin take these issues as proof that the paper is wrong.

  • Indeed, the figures are clearly rough. Maybe the authors were rushed through? Maybe different plots were compiled over many years by many collaborators, stubborny resisting to use the same tools? But hey, while horrible, they’re comprehensible, aren’t they?

  • And indeed, the language is clearly unpolished. Maybe the authors speak very little english though? Maybe they’re writing on tight deadline? Maybe they don’t know about tools such as Grammarly? But hey, while irritating, the text is comprehensible, isn’t it?

  • And indeed, the formatting is clearly sad. But hey, maybe the authors prioritised writing content over debugging LaTeX? Maybe they submit to a journal who does their own formatting? And hey, maybe they are simply less into the LaTeX look than you are?

Please, let’s be less shallow than this!

I do feel some sympathy for the grammar gang. More often than not, it’s an amazing heuristic for choosing what to read and what to ignore. And it really does help that basically all crank papers on “THE NEW QUANTUM THEORY OF EVERYTHING” are horribly formatted. However, we must not over-apply this heuristic.

To start with, the more trial-and-error the process is, the more likely it is for an outsider to make an important contribution. You are exceedingly unlikely to build a good quantum computer without a complete understanding of what you’re doing. On the other hand, the history of materials and chemistry research is full of stories where chemical synthesis is attempted for one purpose (or with one specific hypothesis in mind), but the product has unexpected properties or applications. Outsiders finding new materials every now and again should not surprise us that much.

Finally, the grammar heuristic is basically a Bayesian process for updating the likelihood that the paper is correct. The thing is that most papers are boring and useless, even if correct! If I see that a paper entitled “Numerical evaluation of out-of-order correlators for non-projective measurements in plasmonic ququarts” is badly formatted, then in my head, this only moves it from a category of “Probably thorough but boring - do not read unless absolutely necessary” to “Probably wrong and boring - do not read unless absolutely necessary”. In other words, (probability of correctness) x (impact if correct) $\ll$ (threshold of interest)

However, the preprint “The First Room-Temperature Ambient-Pressure Superconductor” is in the category “Holy shit, this is unlikely to pan out, but if it does, it’s gonna be big”. The typos and grammar move to closer to “Holy shit, this is very unlikely to pan out, but if it does, it’s gonna be big”. But still, (probability of correctness) x (impact if correct) $\gg$ (threshold of interest)!

So don’t dismiss the typos, but don’t let the ad hominem critics take over. Let’s give the outsiders a chance every once in a while, and let’s get properly excited about low-probability high-impact findings like LK-99.