Saturday, April 10, 2010

Whodunit?

I know many examples where senior researchers were listed as authors of a paper even though they didn't contribute to the paper. In some cases I doubt they ever read the paper. I know people who looked at the arXiv in the morning to find they have "written" a paper they didn't previously know of. Especially when it comes to conference proceedings, that are often just repetitions of already published results, adding a previous collaborator is more an act of politeness than a statement about the contribution to the act of writing. That won't surprise anybody who works in the field.

In some cases, interpreting the different author's contributions to a paper can be more subtle. Supervisors for example are frequently named as co-authors not because they contributed to the paper but because they are the ones with the grant who made the experiment/research project possible to begin with. The problem here is of a practical sort. With bringing in the money to make a project possible they do arguably make an essential contribution. In return, the funding bodies want to see their money put to good use and having the grantees' name on a paper increases chances for future funding. Especially for researchers too young to apply for grants themselves (application typically requires a PhD), adding the supervisor is thus an act of self-interest. The problem starts at exactly the point when the paper is submitted to a journal and it is declared that all the authors made significant contributions to its content.

How to read an author list is tacit knowledge that differs from field to field. In some fields, the first author is the one who actually did the work, the last author is the one with the grant, and the ones in the middle might be ordered by some obscure ranking. In other fields, author ordering is strictly alphabetically and being a first-author simply an ode to your family name (Abado, A. A. et al).

I was thinking about the meaning of author lists yesterday when I read this ridiculous article in the Times Higher Education: Phone book et al: one paper, 45 references, 144 authors. It can be summarized as: Professor for Ethics comes across a summary-paper from the Sloan Digital Sky Survey and counts 144 authors. Since he hasn't seen such long author lists in his field, he concludes there must be something wrong with physics. Clearly, people like to have their names on such long author lists because "Careers depend on number of publications."

Now I have written many times on this blog, most recently here, that the use of metrics for scientific success can indeed hinder progress and should be done with caution. But the ethic professor's implicit assertion that hiring committees are not able to distinguish between a single-authored paper and a collaboration's summary paper simply shows he has no clue what he's talking about. Even when it comes to the above mentioned papers with few authors, the question who made what contribution is typically (extensively!) addressed in letters of recommendation accompanying a publication list. The reality is that in experimental physics such long, or even longer, author lists are not uncommon. It's simply a consequence of these experiments being enormously complex in the technology and software used. If anything, the THE article shows that comparing ethics to physics is like comparing fruitflies to the homo sapiens. I'll leave it to you to decide which stands for what.

In any case, the obvious solution is that there would be a way to better declare what the author's contributions to a paper were. This has been discussed many times previously, and I am hopeful that sooner or later this will become reality.

On that note, YoungFemaleScientist had a post this week on the Ethics of Publishing, and hits upon more relevant problems caused by the pressure to perform according to a certain success standard. That's the praxis of splitting up papers into "least publishable units" or dumping all sorts of stuff together in the hope that it will overwhelm the referees and something of it will make a splash. The latter is not very common in hep-th though. I guess that's because there are too many people working on too closely related topics, so everybody tries to get even smallest results out as soon as possible because otherwise they risk being scooped.

19 comments:

Phil Warnell said...

Hi Bee,

You’ve thrown out a lot of thoughts in respect to papers and what the significance is to those names that appear at the top. However, I’m not sure how things work in Germany, yet in Canada a lot of this what I call forced collaboration begins in high school and proceeds on through university, even at times into the graduate school level. The reason often given by the instructors is it helps people learn how to approach problems and tasks collectively, yet the primary motivation is so there will be less papers to grade and therein less work for the instructors.

Subsequently there are many students that survive their respective systems, not so much resultant of having personal abilities much beyond their political ones. So I’m afraid a degree of this mindless bureaucracy gets built into those who comprise the ivory towers, more out of laziness, rather than any higher ideal of meaningful efficiency. It could be argued that in nature there be some of this also to be found, yet even drones serve a function which is made equal in terms of their influence and eventual fate.


Best,

Phil

Bee said...

Hi Phil,

Well, I'm not sure how "enforced collaboration" relates to the case of authorship, in which case it's more the lack of collaboration that is the issue. It seems to me it's two different points. Also, there are very clearly benefits to working with more experienced people, so I am not sure in how far this has to be enforced or how much one should be complaining about it. As far as physics is concerned, I think it is sensible to have supervisors till the PhD level. After this point it should be left to the researchers if they think they better work with or without supervision. Most people are at this time in their late twenties already and are perfectly capable of being responsible for their own research. Other's might need guidance for somewhat longer. In any case, it should not be enforced. Best,

B.

Phil Warnell said...

Hi Bee,

I think it all comes down to being a matter of difference relative to perspective and method, with you being a self admitted phenomenologist and myself more concerned with what stands as being perhaps the fundamental underlying principles, as to how they are derived. That’s sort of like the difference between looking at illness as a set of symptoms or rather these symptoms being indicative of a specific disease. Then it is left to decide whether by way of solution if it better to address the symptoms or learn more about the disease

Best,

Phil

Jérôme CHAUVET said...

Guten Abend Bee,

Your profound concern about what makes a good science and what makes a bad science, plus all the arguments you bring out to us so as to rigorously demonstrate that there is something wrong in the science world proves that you are what I would call a genuine scientist, as you are adressing the right issue, no matter if this deals with a subject as "sensitive" as carreering.

It could not indeed be clearer : nowadays, the best solution for carreering long and high in science does not fit with performing a science that is fair. And if science is not fair, then science is wrong. My sense is, it is the way scientists adapt to their environment, so they become the best fit to survive in it. Hence, though being not fair, it is however natural.

If I am not mistaken, I think I saw once an article in PNAS in which each author's assignment in the published work was explicitely written (i.e. write the article, performs experiments, design the research). However I am not sure whether all articles must give this information. Those forever-long publication records look anyway ridiculous : Who can have seriously worked on research subjects with the frequency of one paper a month (or so)?... Recruitement committees who pretend to believe this are just ones who do the same for their own, so they are just lying by accepting it : they are just well in their position, we are happy for them. A small self-stabilizing system has emerged, and one cannot do anything to hinder it.

To ameliorate this, the system would have to think it over, and then decide to oblige those who are well-embedded in their place to change... But the system is precisely made of those who are well-embedded in their place : it's just another type of aristocracy, so what we are asking looks like the French revolution.

I just hope my interpretation of the situation is wrong.

Regards,

Arun said...

The butler?

:)

jack said...

Bee,

This is no major surprise - how is it we have some researcher with literally hundreds of papers? There's no way one person can write that many papers. Rather he/she is the PI who gets the grant money, so automatically gets named as a co-author. It's standard practice.

ErkDemon said...

So we set up the equivalent of www.imdb.com, to deal with published science papers and books the way that the "Internet Movie Database deals with movies and tv programmes.

We nick the whole format. :)

Each paper gets a "cast list" cross-referenced to separate entries for each person, each person gets a profile page that cross-references every paper they're listed on, and the "paper" pages can include trivia, media clips, photos, background information and supplementary data.
Eric

Bee said...

Jack,

It is very well possible to write a hundred or so papers. I know people who have that many papers and they *did* write them themselves. It just adds up over time. See, if you write 3-4 papers every year, and you do so over 3-4 decades, you'll end up with more than 100 papers easily. And that doesn't count the numerous repetitions as in talk summaries, conference proceedings, popular articles, etc that are usually just a different version of casting the same result. If you work in a field where there is lots of data and you work on a numerical code to analyze/fit that data, you can end up writing >10 papers per year easily, which vastly accelerates that process. In summary, the number of papers itself is not an indicator for success, but neither is it an indicator for there being something wrong. It just vastly differs from field to field, and even within fields. Best,

B.

Bee said...

Hi Eric,

Well, I'm sorry to see your "Starmeter" is "Down 56% in popularity this week" ;-) But yes, I guess it would be something like this. The question is though whether it won't just increase the politics behind the authorships. Eg that there will be some unspoken agreement that The Guy With The Money always had the idea for the project whether that's true or not. Or that the student always did the detailed calculation because he needs to find a postdoc position next year etc etc. Best,

B.

Bee said...

Hi Phil,

Your comment is funny in that I was just yesterday thinking about fundamental research versus phenomenology versus experiment, how they relate to each other and where they overlap etc. I'm not happy to be put in a drawer that in your account excludes the search for fundamental principles, that's certainly not a very accurate description of my interests and also not, so I think, what drives most people who do work related to mine. In any case, maybe that should be content of another post, it's somewhat off-topic here. Best,

B.

Bee said...

Hi Jérôme,

Sadly enough, I think your account is right. This survivor bias is one of the main reasons why the system with all its problems is so resistant to change. Best,

B.

Phil Warnell said...
This comment has been removed by the author.
Phil Warnell said...

Hi Bee,

Well I should have realized as is demonstrated by your most recent published paper, that you of all people would stubbornly refuse to be restricted by a box:-) Also please don’t take my comment as a slight, yet simply an observation I’ve made over the years as to how physicists approach the problems they are presented with. I find it interesting you had similar thoughts and then very curious just what they might be and thus will leave my own comments and arguments to whenever it is you decide to address it in your blog.

The only thing I would like to have made clear is I meant no disrespect, as in fact two of my favourite physicists of all time being what I would find as one of each, with Albert Einstein representing being a foundationlist and J.S. Bell the phenomenologist. You might also find it interesting that I consider what you managed to do with your box problem as significant to quantum gravity as J.S. Bell’s inequality is for the foundations of QM. However, like you say my thoughts on this should be left for a more appropriate post which I hope one day you might write.

Best,

Phil

ErkDemon said...

Hi Bee!
Maybe I'm an optimist ... but I think that if there was an IMDB-style website for published papers, hooked into the DOI/ISBN/ISSN databases, and it let authors upload additional comments and details for their experiments after publication, then a lot of researchers would gratefully seize on the opportunity to cut free of conventions and protocols and flesh out the details of their work, and say what really happened ("Here's a group photo of the team standing in front of the equipment. Yayy us! This is Dave. Dave kept the machines running, and fetched pizza").

If they're unsure, they can leave their entry blank, or showing the defaults.

It'd be great to have somewhere that people could leave comments on papers, and maybe also have a wikipedia style "Discussion" page commenting on how well (or not) the commentary and coverage was doing.
For instance, I've seen a major peer-reviewed experimental paper saying that the author didn't see how the Ives-Stillwell result could have been as accurate as claimed, given the hardware used ... but I've lost the reference. If I //had// that reference, I could add it to the IS paper's entry as a "criticisms ..." forward link, and if I didn't, I could post a note in the comments section asking if anybody else knew what it was. At the moment there's no real central point for aggregating data about science papers.

I think it could be a great resource, and it'd have the advantage over imdb that the "skeleton data" to get it up and running would already be freely available.

IMDB is supposed to have been hosted for a while by Cardiff University. If we can do this for movies and individual episodes of TV programmes, why not for science papers?

As to whether people would still be less than honest on the new database even after their papers are safely published, well ... I'd like to think that researchers tend to be honest people, but if some aren't, and if the habit of trying to succeed by faking data is so ingrained that they even do it when they don't have the excuse of journals breathing down their necks, then ... to me that'd then raise questions over the authenticity of other aspects of their research.

Christine said...

In any case, the obvious solution is that there would be a way to better declare what the author's contributions to a paper were.

It would be nice.

It is not simple, though, for various reasons.

It is not always the case that you can, e.g. just put a percentage indicating the relative amount of work of each author.

For instance, it may happen (as it happened to me) that one colleague offers an intellectual contribution (it can be a single idea or note) that changes the course of investigation or the analysis of the problem in such a significant way that, even if the person does not put his/her hands on the "dirty" part of the work, the main author may find that that person deserves to be an author as well. (Sometimes it happens that the person in question just asks for an acknowledgment in the end of the paper).

Once I wrote a software application that was useful to someone else's work and I was asked to be an author, but I declined, I saw no reason to be an author, but an acknowledgment was fine.

Sometimes it is a political thing to put someone as an author, an unfortunate common practice.

In any case, it would be nice to have a space, e.g. at the end of the paper indicating explicitly each author's contribution and its relative importance.

Bee said...

I think we should just split the keyboard up among the authors. Then you could say I did the a,j,k,r,f and all the numbers, and he did the p's, and the student had the special symbols or so ;-)

Christine said...

BTW here in Brazil academic or governmental (permanent) positions are filled through very competitive contests which usually involve written and oral exams, CV analysis, project and memorial defense in front of a committee, interviews, etc.

In my experience, candidates who go for a postdoc abroad to work in large, well-known groups usually return with many papers with long lists of authors. Such postdocs return to Brazil (if they return) with a big weight in their CVs because of the increased number of papers as compared to a postdoc who have not participated in such large groups.

I mean, although such experiences are undoubtedly well worthy per se, a postdoc who did a small contribution in a large group resulting in several papers will have a big advantage over a postdoc that stayed in the country in a smaller research group, even if having published as a single-author (or smaller list of authors) in excellent journals, but with a smaller number of papers in their CVs.

So, again, it would be nice to have each author's contribution to a paper written out explicitly in the same paper so that a more adequate analysis could be made, at least in cases of contests.

Rastus Odinga Odinga said...
This comment has been removed by the author.
ErkDemon said...

I just thought, it must be really dispiriting being in modern research if your name starts with a "W".

You must think, "Even if I'm the top researcher in a group, I'll always get lowest billing, assuming that I don't get relegated to an 'et.al.', while some upstart teaboy called Aaronsen gets first credit on everything, and the papers end up being cited as "Aaronsen et al".

So is that why John Wheeler put so much emphasis in writing books in his later career, because that way he knew he'd always have his name in the credits?

Maybe someone could do a statistical analysis comparing the number of scientists with surnames starting with each letter of the alphabet, against the number of papers produced, against the number of authors per paper, against time. We could track whether the higher-letter researchers have a paper-count that trails off faster as they become dispirited, whether their number of collaborators is smaller, and whether this effect (if it exists) is getting worse.