Numbed by numbers

Couple things in my news feed this morning that really woke me up — one a startling statistic and the other a reminder of what statistics miss. The first from Nature, 'How to win a Nobel prize: what kind of scientist scoops medals?':

John W. Strutt, who won a physics prize in 1904 for his work on the properties of gases, has 228 academic descendants with Nobels — his students, their students and so on. … An incredible 702 out of 736 researchers who have won science and economics prizes up to 2023 are part of the same academic family — connected by an academic link in common somewhere in their history. Only 32 laureates … have no connection to the bigger academic family.

Meaning it's nearly impossible for 'true' outsiders to break in. Either you become part of the The Network or you have a very low chance of winning a Nobel Prize. Of course the prize-giving apparatus isn't a machine. There are humans making these decisions and clearly in a famously human way: not really paying attention to the consequences of their preferences or assuming that that doesn't, or even shouldn't, matter.

But what does The Network say about science itself, especially about good* science and where that gets done? That is, what institutional mechanisms and/or forces are (even passively) encouraging the scientists who do such work to clump together?

One factor that immediately comes to mind is funding: in the typical Indian experience, because most places of research have traditionally not been well-funded, the government or some philanthropic entity endeavours to set up a few facilities focused on research and funds them well, while the rest struggle on.

On a related note, should the diffusion of researchers who produce good-quality research (and know how to do it) into previously neglected locales be desirable?

Next, the reminder of what statistics miss:

The French researcher and physician Didier Raoult has been banned from practising medicine for two years. It is the latest and probably most significant sanction against Raoult after he became infamous during the pandemic for his enthusiastic support for hydroxychloroquine even though the drug lacked evidence of its efficacy against COVID-19.

His claims brought the spotlight on him as he probably intended but then expanded to reveal he had published too many papers — much more than should be humanly possible. But Raoult took pride in his research metrics, so even as research integrity investigators including Elisabeth Bik revealed dire problems in his published** papers — including image manipulation and ethical lapses in clinical trials that rendered them illegal — Raoult and his supporters came out swinging on social media.

He also filed a lawsuit against Bik alleging she and others were besmirching his name without reason. Raoult eventually lost these disputes and in the process the trust and respect of the research community. Now his medical license has been revoked. He was retired but the action was clearly symbolic: Raoult is done.

It took Bik's and her peers' scepticism to reveal the extent of Raoult's misdemeanours. His metrics betrayed nothing of it except through their largeness.

As if on cue, The Hindu published an excellent opinion piece by S. Swaminathan today about why and how we educate people, including those who become professional scientists:

The metrics-focused system has created a situation which implies that education is a market rather than a citizen’s right and the state’s duty.

* "Good" here means worthy of winning a Nobel Prize, not good per se.

** Remember that they were published, meaning the journals that did are answerable, too.