Hacker News Reader

Comment by dinkleberg | Hacker News Reader

I suppose this is a meta question, but why is this in pubmed? While it is in the format of a research paper, this seems like more of a blog post coerced into that format.

I'm not an academic, so I'm curious, are papers like this just for fun or are they intended to actually contribute to our understanding of the world? Because I can't help but think that the methodology in the paper is dubious. Do we just the guy because he is a professor? I suppose ideally the reputation system is such that if it is revealed someone is just making things up they'll get booted out.

Not trying to hate on it, it was just an odd paper so it had me wondering.

View on HN

It is an article from the bmj “Christmas issue”[0] which is meant to be light hearted or satirical. My personal favorite was about the efficacy of parachutes when jumping out of a plane[1] - which fails to clearly describe that the plane was on the ground. Their point was about replication and clear thorough explanation of methodology in science.

The Christmas issue has been critiqued for exactly the reason you made this comment - indexers don’t pickup on the differences of the Christmas edition making it functionally invisible if you see it in pumped unless you know about it

[0] https://www.bmj.com/about-bmj/resources-authors/article-type...

[1] https://scholar.google.com/scholar?hl=en&as_sdt=0%2C11&q=bmj...

Because it's in BMJ (a journal), which is indexed. It has things like letters to the editor, obituaries, and "observations", which are indexed as well. Cf. https://pubmed.ncbi.nlm.nih.gov/30842093/

This is a fantastic research paper, better than 90% of the "real" ones. It's natural science in its purest form, grounded in lived experience.

1. Introduce novel stimulus

2. Observe response

3. Hypothesize about cause

So much of the statistical significance and p-values and citations is oppressive bureaucracy at best, cargo cult busywork at worst. This is the good stuff.

Except that the conclusion is dogshit. It's such a wild leap that Superman would wonder if he could follow. There's a reason we use formal stats, and it's because our intuition for these things is not reliable.

And you can see why the conclusion's bad. You can imagine a way to prove the author wrong: a hypothetical observation that you expect, that's inconsistent with their conclusion. Or, as we say in the business, “further work”.

No, the reason it's bad is that it's 100% non-sequitur. There's no reason to even consider further evidence before dismissing it.

Other commenters have pointed out that this is basically a parody article, which is the only way it makes sense. But that means anyone who thinks it represents the pinnacle of science has eaten the onion and needs to reevaluate their criteria for scientific work.

> why is this in pubmed?

[...]

> because he is a professor

Ding ding, we have a winner!

(EDIT: to clarify, I'm not being snarky, if you've spent your adult life contributing to a certain class of publication [which happen to be those indexed by PubMed] perhaps it becomes second nature)

Reply on news.ycombinator.com