Hacker News Reader

Comment by pmoriarty | Hacker News Reader

It's interesting how sci-hub's papers on medicine dwarf those in many other fields like comp-sci, math, and physics. I wonder if that reflects the number of papers in those fields, or if sci-hub just has a non-representative sample. If the latter, why?

View on HN

I guess today is much easier to find new noteworthy, publishable facts in medicine than physics. New diseases are discovered every year, and old diseases are poorly understood (e.g Alzheimer disease), and the treatments for many of them are still sub-optimal, or even inexistent. Every patient is different, individual cases are research-worth. We only got antibiotics in the 1940s. On the other hand, most big breakthroughs of physics happened between the 17th century and the first decades of the 20th century. After the general case is cracked in physics, individual cases have very little publishing value.

> much easier to find new noteworthy, publishable facts in medicine than physics

Plus I would imagine that everyone wants their illness looked into, thus that's where funding tends to go. I care much less for physicists to figure out what dark matter is than how to treat health problem X that bothers me daily. (Just an example. In my particular case I'm healthy and would actually be quite excited about dark matter findings compared to any individual illness solution... but still.)

This comment would be worth a lot more with some stats about funding going towards the different fields, though. Not sure where to find that.

It does appear to be the latter. I just searched for several famous ML papers (attention is all you need, lottery ticket hypothesis, capsules, etc) and they are not there. I think if someone counted all papers that have been ever published anywhere, the picture would be a lot different.

So does that mean that vastly more people in medicine use sci-hub than do people from other fields?

Or is there some other reason for the discrepancy?

Could be. I've been an ML researcher for 8 years and I haven't used sci-hub until today. Ironically one of my (very obscure) papers is available there.

I suspect that a lot of ill people, and people with loved ones who are ill, spend an enormous amount of time reading every paper they can find on the illness, or trying to narrow symptoms down to a particular illness. That's also a situation where you can go through 50 papers easily for every useful one (or even intelligible one, if you're a layman) you find, so a definite sci-hub situation unless you're independently wealthy. It's the requests that draw (or once drew, right now I guess) stuff into the database.

I don't even know the scale of medical research vs CS in terms of number of schools.

The answer is right there on the page: the database covers >95% of publications among major publishers.

Math and comp-sci researchers use https://arxiv.org heavily and thus people usually do not need something like sci-hub.

Relevant xkcd - https://xkcd.com/2085/

I think it's due to the sheer number of biomedical papers published each year, coupled with comp-sci, maths and physics papers being less likely to be behind a paywall.

... which is due to the sheer amount of cash being dumped into medical research. Because congressmen get cancer.

There are several medical subfields that on their own get more funding than the entire NSF (i.e. all other science).

Reply on news.ycombinator.com