Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Thousands of scientists publish a paper every five days (2018)

nature.com

Post reply on HN
Print view View original

111–120 of 157 posts

Re: Thousands of scientists publish a paper every five days (2018)

#111

Post by YeGoblynQueenne »

UncleMeat wrote: post #90
YeGoblynQueenne wrote: post #85

Earlier quoted context omitted.

Nothing happened to Ioannidis. He did his job as a scientist and said what he thought was right, even when everybody disagreed with him. So he was wrong? So what? To be a scientist means to blunder a lot. In fact, to be a scientist means to (be willing to) blunder much, much more than everybody else because that is the only way to learn. Science, ultimately, only gives you the tools to know when you're wrong .

Well, he also decided it was appropriate to insult the physical appearance of grad students on twitter who disagreed with him. He definitely didn't come out of covid looking fully dispassionate.

I have no idea what incident you are talking about but, from the context, I have no doubt that it is something inconsequential that social media made a big deal about just to have some reason to be outraged at Ioannidis, with whom they were already outraged.

Re: Thousands of scientists publish a paper every five days (2018)

#112

Post by UncleMeat »

YeGoblynQueenne wrote: post #111
UncleMeat wrote: post #90

Earlier quoted context omitted.

Well, he also decided it was appropriate to insult the physical appearance of grad students on twitter who disagreed with him. He definitely didn't come out of covid looking fully dispassionate.

I have no idea what incident you are talking about but, from the context, I have no doubt that it is something inconsequential that social media made a big deal about just to have some reason to be outraged at Ioannidis, with whom they were already outraged.

I didn't say you should be outraged. Feel what you want to feel. I list this example where he definitely did more than "his job as a scientist."

Re: Thousands of scientists publish a paper every five days (2018)

#113

Post by sillysaurusx »

borroka wrote: post #110
dylan604 wrote: post #100

Earlier quoted context omitted.

> The fact that most published papers have no value is also true. Is that true though? If I'm about to embark on a bit of extensive research, the first bit of that research would be to see if someone else has already done it. If they have, then there's value to me in not having to spend time/money in repeating what someone else has already done. Or it could show me where other things can be tried instead of just repe…

You are referring to the publication of "negative results", which are of course informative, although rarely published due to a bias shown by journal editors and scientists themselves towards "positive results", that is results supporting the hypothesis tested. Negative results are published, yes, but in the context of popular or highly contested hypotheses or models. That's valuable research. Some examples of papers…

> Some examples of papers that have zero value are papers testing hypotheses nobody is interested about

I broadly agree with you, but this point is probably mistaken in an illuminating way. The path to breakthrough science sometimes passes through uninteresting work. Few scientists were interested in natural selection before Darwin and his contemporary. There are many other examples.

Re: Thousands of scientists publish a paper every five days (2018)

#114

Post by dahinds »

londons_explore wrote: post #19
stenl wrote: post #5

John PA Ioannidis, one of the authors of that paper (which is from 2018 btw) has published an average of 60 papers per year for the last decade, and more than 1100 in total ( https://pubmed.ncbi.nlm.nih.gov/?term=Ioannidis+JPA&filter=d... ). I’m pretty sure he doesn’t even have time to read most of his own papers.

A paper is only a few pages long usually. A read of a paper takes perhaps half an hour. And in that time, you might even be able to check some references, make a few edits, or drop an email to someone to gather some more results for a figure or two. Doing this once a week seems very reasonable.

A close reading of a scientific paper of any consequence does not take half an hour. These days a published paper may typically be only a few pages but it isn't at all uncommon for the paper to be accompanied by 20-50 pages of supplementary text. When I'm asked to review a paper it takes me the better part of a day, at least. If I were a senior author on that paper that would be the bare minimum that I'd consider an acceptable commitment of effort.

Re: Thousands of scientists publish a paper every five days (2018)

#115

Post by __seka__ »

rockostrich wrote: post #8
stenl wrote: post #5

John PA Ioannidis, one of the authors of that paper (which is from 2018 btw) has published an average of 60 papers per year for the last decade, and more than 1100 in total ( https://pubmed.ncbi.nlm.nih.gov/?term=Ioannidis+JPA&filter=d... ). I’m pretty sure he doesn’t even have time to read most of his own papers.

It looks like he's second to last or last author on most of those papers which typically means he's the advisor for the graduate student/post-doc that's actually doing the work. It's still an insane number of papers to be a part of but considering Ioannidis is director of the Prevention Research Center and co-director of the Meta-Research Innovation Center at Stanford it's not really that insane.

> Prevention Research Center and co-director of the Meta-Research Innovation Center at Stanford

Aren't those roles supposed to hold a significant amount of responsibility and therefore require a certain level of commitment? A day has only 24 hours. How much time do you think someone in such a position can dedicate to reviewing papers?

Re: Thousands of scientists publish a paper every five days (2018)

#116

Post by lliamander »

UncleMeat wrote: post #90
YeGoblynQueenne wrote: post #85

Earlier quoted context omitted.

Nothing happened to Ioannidis. He did his job as a scientist and said what he thought was right, even when everybody disagreed with him. So he was wrong? So what? To be a scientist means to blunder a lot. In fact, to be a scientist means to (be willing to) blunder much, much more than everybody else because that is the only way to learn. Science, ultimately, only gives you the tools to know when you're wrong .

Well, he also decided it was appropriate to insult the physical appearance of grad students on twitter who disagreed with him. He definitely didn't come out of covid looking fully dispassionate.

> He definitely didn't come out of covid looking fully dispassionate.

No one did.

That said, despite the serious amount of professional scrutiny and and public harassment he was subject to, I think he made valuable contributions to the conversation around COVID-19 public policy.

Re: Thousands of scientists publish a paper every five days (2018)

#117

Post by dguest »

Bedon292 wrote: post #88
elashri wrote: post #58

> The vast majority of hyperprolific authors (7,888 author records, 86%) published in physics. In high-energy and particle physics, projects are done by large international teams that can have upwards of 1,000 members. All participants are listed as authors as a mark of membership of the team, not for writing or revising the papers. We therefore excluded authors in physics. While this is true in general, I'm part of…

It definitely feels a little odd. If everyone is an author, then does authorship really matter? Is authorship broken out in any way? I would think it could be useful to do something like credits on a movie. Here are the writers, the directors, the producers. And then here is the lighting team, and this team worked on these special effects, this other team on these other effects, etc.

To answer your second question: no.

On a few of the biggest experiments the author list is completely flat and alphabetical. From outside the collaborations there's no way to know who actually wrote the paper or did any of the work described therein.

The best argument to keep it this way is that fewer people fight about who gets credit. But anyone with internal access and a bit of experience knows where to look. If you want to verify someone's contribution from the outside, you just have to find one of the other 3000 authors and ask them to do some sleuthing.

Personally I'd answer "no" to your first question too, but that's a bit more subjective and depends on the funding agency. Particle physicists have, for better or worse, thrown a wrench into the whole idea of authorship metrics, but they did so in a way that would benefit them by default: if the funding agency has a publication quota, everyone passes!

Of course this causes a few problems:

- If your funding agency hasn't caught on and you aren't in particle physics, you are effectively being cheated out of a better ranking.

- Many agencies are now disqualifying publications with over 1000 authors, in which case particle physicists get zero publications.

To deal with the second case, it's actually quite common to see smaller groups of scientists from these collaborations writing papers using their own simulated data, as a way to get publications with fewer authors. I would definitely count myself in the group that considers this situation absurd, but that's where we are.

Re: Thousands of scientists publish a paper every five days (2018)

#118

Post by borroka »

sillysaurusx wrote: post #113
borroka wrote: post #110

Earlier quoted context omitted.

You are referring to the publication of "negative results", which are of course informative, although rarely published due to a bias shown by journal editors and scientists themselves towards "positive results", that is results supporting the hypothesis tested. Negative results are published, yes, but in the context of popular or highly contested hypotheses or models. That's valuable research. Some examples of papers…

> Some examples of papers that have zero value are papers testing hypotheses nobody is interested about I broadly agree with you, but this point is probably mistaken in an illuminating way. The path to breakthrough science sometimes passes through uninteresting work. Few scientists were interested in natural selection before Darwin and his contemporary. There are many other examples.

We are talking about different things. We know about speculative research (which I referred to in my first comments), which gets discovered later on because of new methods, technologies, changes of paradigm. It rarely happens, and it makes a "splash" because it is good narrative and it is unlikely. But it happened, no doubt.

I am talking about, just to offer an example, the vast majority of research that is published (please take this as a gross simplification) in journals with impact factors that start with a zero, theoretical work that re-invents linear regression, work based on data collected like a stream collects rain during a storm, that is everything gets in unfiltered.

The justification for thousands and thousands of worthless articles published worldwide every year cannot be, "maybe among them there is the next natural selection-like theory of evolution that will revolutionize our understanding of the world". But unfortunately, left unchecked, scientific production has increased tremendously in volume, and I am quite confident that increase did not lead to an increase in "good science".

Re: Thousands of scientists publish a paper every five days (2018)

#119

Post by bee_rider »

throw_pm23 wrote: post #18
version_five wrote: post #9

I did my PhD in a group with ~20 students and postdocs including many co-supervised with some other department. The prof in charge of the group had his name on > 90% of the papers we published, I believe justifiably. He would have read all of them, and written sections where appropriate, and had a hand in directing and organizing the research as well as challenging the results and strengthening the presentation of th…

I was in a similar group a decade ago. At the time, I found it impressive and efficient, but looking back now, I find much of those papers not to amount to much. I wonder what your experience is: were those 30 papers per year good science overall?

I’m not sure what to compare against. Of course most research papers don’t amount to much, but are the ones produced by big groups with only a little hands-on work by the professors much worse on average, I wonder?

Re: Thousands of scientists publish a paper every five days (2018)

#120

Post by pvaldes »

Bedon292 wrote: post #78
lost_tourist wrote: post #70

Earlier quoted context omitted.

So I still don't understand why these professors don't have their name second after the student author. It's still prestigious to guiding some of the best minds but also very honest, maybe even have a special symbol to indicate it as such in parenthesis.

I am sure it can be different everywhere, but at least my advisor in grad school was great about making sure the students were ahead of him in the list. The student(s) who were primarily responsible were first and then he would be on it after. Out of curiosity I checked and he has 161 publications on Google Scholar. Of the publications with multiple authors he is first author on only 7 of them. And then 7 are single…

This is the standard protocol
Post reply on HN