Friday, April 09, 2010

More on Authorship

My last post, on what I called "60-40" papers, where one author does non-trivially more of the work than the other(s), seems to have generated some interesting comments, worth following up on.  There seem to be multiple issues in the comments that, to me, appear essentially orthogonal:

1)  How does the community properly assign credit for 60-40 papers?  Should we use author ordering or some other mechanism to assign credit?
2)  What about advisors who do minimal to zero work but put their name on the paper?
3)  At what point has a person involved in the project done so little work that they should not be included as an author (by either withdrawing willingly, or possibly by being told "you're not an author").  (I think of this as separate from the "advisor" issue.)

Let me start with item 1, assigning credit.  I promoted the approach used in theory (derived, apparently, from mathematics) of alphabetical order, claiming credit comes out through things like letters and who gives the talk, and is determined more clearly over the course of a career.  Many question this;  indeed, many other fields use entirely different systems.  Many fields use author order to signal the level of contribution in some way, so that being "first author" has significant meaning.  At the extreme, the journal Nature, for example, suggests that author contributions should be fully specified in each article in their guide to authors:

"Author Contributions: authors are required to include a statement to specify the contributions of each co-author. The statement can be up to several sentences long, describing the tasks of individual authors referred to by their initials."

Graduate student and postdocs, in particular, are more concerned with systems that clarify credit, and this is understandable.  They have short career track records, and want a job;  making sure that they get their proper credit often seems, to them, quite imperative.

I'd like to defend the alphabetical, no-explicit-credit-assigned system, and then provide a couple of stories.  (If you find that indulgent, you can skip the stories.) 

One philosophical approach is to try to start from a blank slate.  Forget about your current situation, and how your field does things.  Your starting point is that you're just starting a career in science.  What sort of system do you want to use?  I'd argue you'd want to use a system that would lead to long-lasting, productive collaborations;  that would have minimal overhead;  and that would still provide meaningful ways of calibrating people over appropriate time periods.  I think pure alphabetical does that.  It removes the need to fight over (or even discuss) who contributed exactly what, leading more easily to frequent and repeated collaboration.  To be clear, I have a strong bias:  collaborations, I think, are great for scientific production, and on the whole make research much more fun.  Alphabetical order is clearly easy.  And while it's weak on allowing someone to find out how much each individual author contributed to a specific multi-author paper, over the course of several papers, I think the calibration works, especially when augmented with additional information such as letters in job searches and promotion cases.  Further, it's not clear that other systems are really stronger in terms of assigning credit.  Authors can disagree on contributions -- how does this get settled, and what does it do to future collaborations;  in multi-author situations where order ostensibly matters many advisors will game the system, for example by putting students first regardless of their contribution in order to prep them for the job market or out of professional courtesy;  and it's not clear how, for example, to value different types of contributions, like ideas vs. data collection and analysis.  My bias is that the blank slate scientist starting their career would pick the alphabetical order system.

I have at least one data point for this conclusion: myself.  (Here's where the stories start.)  In graduate school, a bunch of us students got together and wrote a paper.  This was a case where I was definitely the 60 author, and I thought it would be best if I was first author.  The other students didn't object, but since I knew it wasn't standard for theory, I asked my advisor.  (He wasn't a co-author for this paper, so his view was not biased in that regard.)  He told me it was my choice, but that I needed to recognize the following:  I would possibly get more credit for this paper, but, from then on, I would have adopted a system where, for every paper, I'd have to face the possibility of constructing the author order with my co-authors.  Did I want to have that discussion for every paper down the line?  I went with alphabetical order and have never looked back.  I always recommend alphabetical order, although when I work with people in other areas I do defer to whatever system they want to use, and tell them they can put me wherever they like in the ordering.  (It is true that, with tenure, one can care much, much less about such things.) 

On the other side, another story.  When I applied for my CAREER grant, apparently I was on the borderline, and it took quite some time to get the final word.  I asked the NSF officer for feedback -- especially in case I needed to resubmit.  (Apparently, enough money came through in the end to fund me.)  One thing he said was that a lot of my work had been co-authored with very talented people, and it wasn't clear what my contributions were.  This was a case where, obviously, there were no recommendation letters to draw from.  Still, I was offended then by the comment, and looking back I still find it ridiculous.  At that point, I'd written multiple papers with these other authors (who were not my advisor) -- clearly they thought I was contributing something worthwhile.  And why was the assumption that they were the 60 contributor, instead of me?  It's not clear that using author ordering would have helped in this case, or that such cases are at all frequent.  But it does help me understand alternative points of view on the underlying question.

44 comments:

Anonymous said...

If you are upset by this NSF guy, try being a woman author in a collaboration with men. I have heard such comments not once, but many many times, even when I has the 60 author and my (men) coauthors were in a somewhat different area than mine.

Anonymous said...

Alphabetical order seems to privilege those who won the "alphabetically lottery." Even if only subconsciously the first name is most easily remembered(especially if there are more than 3 authors). A random ordering is clearly superior.

Another issue: how do people know what ordering is being used? If there are two authors it could be hard to tell if the ordering is alphabetical or in order of contribution.

A crude solution would be adding a footnote specifying the reason for the author ordering, but this seems inelegant.

But just very generally, the question of assigning credit is always going to be very tricky. Everyone is influenced by everyone else and ideas rarely emerge in a vacuum. And yet we need credit to decide who gets the job and the award. Which is not to say that some contribute a lot more than others, just that untangling the contributions is generally not easy.

Anonymous said...

Michael: You started with (1) but did not go on to address (2) or (3). What do you think about those?

I am currently in both situations. My advisor does nothing for any papers except proofread and pay our salaries but demands to be an author on every paper in the lab. On another paper there is a student coauthor who should ask to be taken off (he was not even present for any of the research nor did he help with writing the paper or making figures or anything), but he hasn't volunteered.

In neither case have I considered telling them that the gig is up and they have to do actual research with me if they want credit for my research results. But your story about problems with grants because your papers had too many authors makes me think it might be time to start cutting dead weight if I want to get grants for myself in the future. Of course a possible grant in the future is worth nothing if I have no salary today (and no letter of recommendation) after confronting my advisor.

Charles Sutton said...

I'm in an area of computer science which is author order by contribution (for the first author at least). You do have to discuss the order, but I've never seen that discussion be acrimonious. Usually the equitable decision is obvious: When there's a 60 author, all parties are happy to acknowledge that publicly. Perhaps I just have unusually gracious collaborators, or perhaps I haven't been on, say, 30/30/10/10/10 five-author papers, where the first-author decision really is unclear.

Now, we use the same informal methods of assigning "real credit" as you describe for the alphabetical scheme, so I don't think that there's necessarily a real reason to favour one scheme over the other.

I will give two advantages to the order-by-credit scheme:

1. Even if everyone "in the know" correctly assigns credit to the 60 author, it's nice to be able to signal that to people who are outside of "the know".

2. All of your in-text citations aren't of the form Aaaanderson et al., making them easier to read.

Avik Chaudhuri said...

A couple more stories:

1. As a PhD student, I published some papers with my advisor and then some without him; for the latter, he simply withdrew (and I'd never even thought of writing papers without him before that). Towards the end he actually triggered collaborations for me with other authors, but still remained out of the fray whenever appropriate.

2. Much later (during my post-doc), in one of my interview talks, I ended by showing pictures of people I collaborated with on various projects (there were several threads of work with some common theme). I did this because (a) I believe these projects were greatly improved through these collaborations, even though I've been a "60 author" on almost all of these projects, and (b) I thought this would provide evidence that I "liked" collaborating. Surprisingly, one of the interviewers got the wrong idea about my contributions, even though the bulk of the talk should have made them unquestionable.

Anonymous said...

I work in the area of systems, and having alphabetical ordering for authors is just plain bad. Almost every one in systems, lists the authorship in the order of contribution, occasionally choosing the main contributor as first author, and then alphabetical. It will be confusing if you list in alphabetical, as it is difficult to tell who did what.

The reason is obvious, in systems, only one guy (usually the PhD student) does the most of work, like coding. It may not be the most intellectually challenging activity, but it would be completely inappropriate for everyone to get equal credit, because they suggested a cute hack to the main algorithm.

Flavio said...

During my first year of Ph.D. studies (in TCS), I used to wonder whether the lexicographic approach was really better than something that had some correlation with actual authors' contribution.

I remember reading some post of Michael where he talked about the paper where he was the 60 author... that made up my mind.

It really takes more time to convince people of why some ordering is wrong, than to do actual research :-)

Michael Mitzenmacher said...

Anon #3: Given the length of this post, issues (2) and (3) are relegated to later posts of their own. Also, it wouldn't hurt for me to think a bit more about them before writing.

Flavio: I thought I had written something about that paper before, but a brief search didn't pop it up. Now I feel I'm getting repetitive! Although I hope it's a story worth repeating.

Michael Mitzenmacher said...

Anon #6: What about systems papers where one person gives not just a "cute hack" to the main algorithm (you sound like a true systems person there -- are you denigrating theoretical contributions?), but actually comes up with the algorithm, which the student then implements? (Of course, with some "cute hacks" of the non-theoretical variety that speed it up x%.) Should the main credit go to the coder, or the algorithm designer?

And what about cases where actually two (or more!) students did the coding together, and you really can't say who did "more". Rumor has it that this actually happens.

You know the joy of alphabetical ordering? You don't have to worry about these cases or any other special cases that come up. I've admitted you lose some accuracy on assigning credit for individual papers; it seems a small price to me.

Anonymous said...

I don't like the argument that it balances out over the course of a career, coming from a tenured professor. For most people, it will not balance out, since they won't get a research job. I think a better argument is that it keeps collaborations friendly. If a paper really is 90-5-5, then the authors should not use alphabetical order.

Flavio said...

Michael: Not repetitive at all...

In fact, after having queried repeatedly a number of search engines, I found out that you wrote about that paper not in a full blog post, but in a comment in Suresh's blog...
I feel it's quite better to let this story have its own blog post :-)

http://geomblog.blogspot.com/2007/11/author-contributions.html?showComment=1195660804701#c3997303295925072912

Anonymous said...

If a paper really is 90-5-5, then the authors should not use alphabetical order.

I think the suggestion is that if a paper really is 90-5-5, then the author order can be anything because it should be a singleton set and the other two authors become acks. The blog post is about what to do with 60/40, where a 40 person clearly deserves authorship of some kind.

Michael Mitzenmacher said...

Thanks Flavio -- no wonder I couldn't find it by searching mybiasedcoin. :)

steve said...

Did I want to have that discussion for every paper down the line?

My fear is that when we adopt systems that cause us to stop talking about things like this (that is, things that are more about career and credit than about the science), we forget about those who stand to lose the most from not being able to have those discussions. In this case, I suspect that women and other underrepresented groups in alphabetical-ordering fields are hurt more by not having the opportunity to specifically defend their contributions than men in those fields, and hurt more than the same groups in other fields where they do have that opportunity.

This is not a scientific data point, but I will note that biomedical science (in which author contributions are actually detailed in the paper) has a much better male/female ratio than other sciences. And within CS, AI fields (in which author order is often by contribution) have better gender ratios than theory and systems.

These conversations may be uncomfortable, but in my opinion they are extremely important for making sure everyone has a fair opportunity.

Anonymous said...

Hi Michael,

I am Anon #6, you were referring to. About your comment

"What about systems papers where one person gives not just a "cute hack" to the main algorithm (you sound like a true systems person there -- are you denigrating theoretical contributions?), but actually comes up with the algorithm, which the student then implements? (Of course, with some "cute hacks" of the non-theoretical variety that speed it up x%.) Should the main credit go to the coder, or the algorithm designer?"

I am not denigrating theoretical contributions. If someone comes up with a great algorithm, and someone else implements and solves all the systems problems, then they should write two papers: one in theory conference, with algo person as the primary author, and another one in systems conference with systems guy as the primary author. I am not arguing about who's contribution is better. I am talking about the case, where people should be given credit for what they have done, and to me non-alphabetical ordering seems to be the right way to do it.

"And what about cases where actually two (or more!) students did the coding together, and you really can't say who did "more". Rumor has it that this actually happens."

Yes, this happens. I was in one of such projects, and we decided to go with alphabetical ordering.

"You know the joy of alphabetical ordering? You don't have to worry about these cases or any other special cases that come up. I've admitted you lose some accuracy on assigning credit for individual papers; it seems a small price to me."

I guess that's where I disagree. It's not a small price, it's important for a young student to be recognized as the primary author, give talks, take ownership, and of course take blame for the work.

I think alphabetical ordering works best in theory/mathematics/science, where multiple people work on a single algorithm/theorem/problem for long time, and they all collectively come up with a solution. It's very difficult to find one major contribution by one person.

This can happen in systems too, but many times it's very clear who has done majority of work quantitatively.

Paul Beame said...

Martin Tompa produced a nice theory of alphabetical author order for the 25th FOCS follies. (See Figures of Merit from SIGACT News.)

A few items:


(1) Because of the way that "best student papers" are done in theory -- ALL authors must be students -- there is a good incentive for advisors not to put themselves on papers where they don't belong. (I find the designation/criteria of such papers in other fields to be quite strange - it seems odd for faculty brag about their "best student papers".)

(2) If the size of the author list is already at least 2 then adding an additional author does not diminish the credit to any of the authors. (This can't strictly be true but it seems to work up to roughly 5 authors.)

(3) As a grad student it is useful to have at least one paper co-authored with your advisor so that people can place you in their mental model of the field despite your unfamiliar name.

(4) A final incentive in theory for not putting marginal senior authors on a paper is the theory rule about PC members not submitting to conferences. If you potentially are a marginal author (just as easily covered by an acknowledgement) who happens to be on the upcoming major conference PC then it is good form to offer not to be an author.

In theory I have seen little of advisors putting their names on student papers, maybe for some of the reasons above. However, I have often seen an advisor who has made major contributions on a collaborative paper argue for putting their student's name on the paper when the other collaborators have not interacted with the student.

Anonymous said...

Was just chair of a search where we interviewed a candidate all of whose papers were with the same 8 co-authors (7 students in a lab and the advisor). The search committee had discussed whether the candidate would be able to establish an independent research program. After the interview, it was clear the candidate would not be able to do so (at least not a good research program).

Anonymous said...

Of course, whether or not authors are listed alphabetically, a job candidate needs to demonstrate that he or she can pursue an independent research program.

D. Eppstein said...

The rule I tend to follow is that there are no 1% coauthors. If you haven't contributed an idea that you can point to as part of the paper, or done some of the hard work of turning ideas into proofs, or done some of the implementation or data analysis, you don't get to be a co-author.

But funding and copyediting don't count. Also, unlike the Hardy-Littlewood rule, being present in the room while the research was being discussed does not count, and past collaborations do not count: if you were a co-author on the paper that was a direct prequel to the present one, that's not enough to list you as co-author.

But once the author set is decided I'm happy enough with keeping it alphabetical and avoiding the nuances of who gets to be placed where, despite the difficulty it raises when I end up writing recommendations for people I haven't worked with.

Anonymous said...

Some stories:

Two great researchers were working together for a few years. At some moment, the one who was always the 2nd in the alphabetic order asked to be the 1st author in one paper. They published a joint paper and the other author often hears comments that most likely he was a 10% author, for otherwise they wouldn't change the order. The paper (as they claim) was 50-50.

A junior researcher was working in for several years and didn't have a single strong paper until he moved as a post-doc to an experienced researcher. The experienced researcher gave him an interesting topic to work on, some key ideas, but then the junior researcher spent a lot of time working on the problem, and they gotten two papers. The junior researcher was upset that the experienced researcher was a coauthor. Then he moved to other place and didn't have a good paper for a few years.

I also often hears some complains from students who are saying that they feel that they contribution is undermined by their advisor. In a few cases I know that students contribution was mostly in trivial parts and I think their advisors are too generous giving them 40% or even 50% credit. But also in a few cases they're right, as someone mentioned earlier, just reading the paper and fixing some typos may be sufficient for an advisor to be a couthor.

Anonymous said...

I don't know why one would think that alphabetical ordering hurts women/minorities.

What I've seen with respect to people who may be more isolated, is that they tend to publish more single-author papers. But then when the time comes for hiring/evaluation, people in judgment use that as a negative. Instead of appreciating the fact that the candidate can work independently, they say that the candidate does not appear to be a good collaborator, and the independent work will actually hurt the candidate.

I'd think any conventions that improve collaboration in general would be better for women/minorities/people who are isolated, but don't want to be.

Anonymous said...

Just read it today:

"when (...) came across an article titled "The Sloan Digital Sky Survey: Technical summary" in The Astronomical Journal, in many ways it seemed "pretty ordinary".

It was 5,230 words long, including the text of its 39 footnotes, and had 45 references.

Yet it was also an article with "more authors than any other publication I have ever come across in any of the areas in which I have worked", Professor Fairbairn said.

A total of 144 authors were listed - equating to a mean contribution of 36.3 words each."

Anonymous said...

Just to have some fun on the topic:

http://www.phdcomics.com/comics/archive.php?comicid=562

I agree with David's policy, which seems both practical and fair.

Anonymous said...

@Paul Beame: Probably you are saying what you do. In the couple of theory papers that I have wrote with my advisor, he has no technical contribution. He is the one who told me that I can work on those problems and listened to my solution once I have solved it. He also helped out in writing the introduction and proof-reading the paper. So he is an author in the paper. How much credit will you give to him for this ?

I also collaborate with others and my advisor is totally cool with that. He doesn't know the problem, I haven't discussed the problem with him even once, so he is not part of the paper. He thus have students who have written papers (and have got best student paper award) without the advisor's name. But that reflects nothing about his 10% contribution in the works where he is a coauthor.

I believe I also get the same kind of credit for doing nothing if I become a professor some time. So that keeps me motivated :)

Anonymous said...

"And what about cases where actually two (or more!) students did the coding together, and you really can't say who did "more". Rumor has it that this actually happens."

Yes, this happens. I was in one of such projects, and we decided to go with alphabetical ordering.


Just to throw in another story:

I was once on a (systems) paper like this, where I and a student did most of the work, and there were 2 other authors. They were the last two, and we went in alphabetical order for the first two.

The trouble is that to an outside observer assigning credit, there is absolutely no way of telling why the order for the first two authors was what it was. Thus ironically, the system inherently and unfairly favors those with earlier last names.

Additionally, the non-alphabetical system discourages me from collaborating 50-50 with other people whose names come before me, since I know I will not get fair credit.

The alphabetical-only system would have done better since there is no expectation of a correlation between author order and contribution.

Stefan Savage said...

I think, as with most things, this is one of those cases where there is no single best solution and thus different communities make different trade offs. What matters as much as anything is the understanding and expectations around such things (i.e., what is understood to have been communicated by those reading the author list and what do the authors expect about that).

Rather than expound on why one scheme is the best, let me share what I've observed that the academic systems community does, what I do in particular and why.

In general, the systems community does not follow an alphabetical model, but encodes some meaning in the author order. There are two near-universal aspects to this: 1) the first author is the one who is believed to deserve the most credit, and 2) students precede faculty. The first is meant to recognize leadership/ownership of a project and while it is imperfect (e.g., I've worked on papers where it would be hard to assign ownership to any one person) by and large it works. Moreover, its not like you can punt on this decision by going alphabetical because you still need to pick someone to give the talk. The second aspect reflects an implicit acknowledgment that the work done by faculty and students is of a different character and that students should be promoted. While not a "rule" per se, its generally considered a bit tacky if a faculty member has their name first on a paper (there are a few exceptions, but they'd take more time to explain than its worth). The middle range (i.e., between the first student author and the faculty authors) is ad hoc... it gets done lots of different ways. The premise is that people in the systems community remember the first author and they know to look to the last author(s) for the adviser(s) but the middle is largely undifferentiated by the reader. Some groups try to reflect some secondary notion of contribution, but it becomes tough to differentiate by the 3rd or 4th author.

In the UCSD systems group we further tend to order the _faculty_ authors inversely with some marginal sense of contribution (i.e., the most involved faculty member is at the end). In truth, many of such distinctions are small but it breaks up the monotony of consistent placement when you work with the same people for many years.

In general, this approach creates more opportunities for conflict (e.g., if two students could both be called "lead"), but has the advantage of communicating information in a way that persists. Moreover, it makes it "cheaper" to add smaller contributors to the author list because their presence does not "de-value" the contributions of the true lead authors (I like this aspect quite a bit because many of my projects benefit from an array of contributors and I like to acknowledge them).

Finally, as I said before, expectation is key here. The scheme I described above is "what the community does". A systems paper with alphabetical authors (or faculty before students) would be seen as "other" in the same way perhaps as a math paper in contribution order might be seen as unusual.

Anonymous said...

I believe I also get the same kind of credit for doing nothing if I become a professor some time. So that keeps me motivated :)

I appreciate this sentiment, especially given the crazy competitiveness of the job and grant market, but I cannot imagine going into science thinking that this is inherently a good thing. I went into science because I like thinking about problems and solving them, whether alone or collaboratively. I could never imagine going into science not because solving problems is fun, but because taking credit for solving them is fun. To each his own, but I figure that there are more lucrative ways to make money doing something you don't really enjoy. I understand the reality that the more senior you are, the more of your time is sucked into non-science, but the moment I am not spending any time actually doing science, I will leave and never look back.

And yet here we are, with scores of professors doing exactly this, participating in no part of the project, from the initial direction and ideas to the final writing. They are essentially being paid to push around piles of money from the NSF to grad students and postdocs, while skimming a big heap off the top as the middle man.

And to those professors who believe that the existence of such unethical advisors can be disproven by counterexample, by stating that they and a few other professors they know always participate at least somewhat for authorship, I say this: Look around your department. I don't care if it is Harvard or Hayfield State University. Someone in your department is abusing grad students in this way, even if you aren't. I guarantee it. Use your tenure for something positive and say something about it. Tell them to stop it. Their grad students have no power but you do because you can't lose your job for standing up for what you believe.

Anonymous said...

"He is the one who told me that I can work on those problems and listened to my solution once I have solved it. He also helped out in writing the introduction and proof-reading the paper."

Introducing a good problem is a big contribution. Helping to write the paper is a MAJOR contribution. If you want to take ownership of a paper, then start by writing it yourself.

Anonymous said...

If you want to take ownership of a paper, then start by writing it yourself.

Often students undervalue the supervisor contributions, which reminds me of an old comic strip:

- I'm beginning to suspect that when the teacher asks a question in class is not because she herself doesn't know the answer

- You dunce! Have you just noticed this or are you joking?

- I'm just joking!

- Don't waste my time then.

[silence]

- ...and me answering all her fake questions using a stupid paternalistic tone!


We all know cases of students whose supervisor pointed out to them a problem; the supervisor has already worked out more than half the solution in his head; the supervisor often got the student unstuck at critical times, only for the student to say at the end"I did all the work, my supervisor just added his name to the paper".

Anonymous said...

This is almost a troll post. The subject itself is pretty unimportant scientifically, and I would say even career-wise.

Nevertheless, let me say that there are certain persons who always think they have done more than others. This is a psychological issue. Alphabetical ordering is thus very good in this respect.

Those students complaining about their supervisor are also wrong of course: initiating a research, by contributing the problem to work on, and maybe even hinting on the solution or technique to be used (even if common knowledge for some), is a major contribution.

Anonymous said...

We all know cases of students whose supervisor pointed out to them a problem; the supervisor has already worked out more than half the solution in his head; the supervisor often got the student unstuck at critical times, only for the student to say at the end"I did all the work, my supervisor just added his name to the paper".

...

Those students complaining about their supervisor are also wrong of course: initiating a research, by contributing the problem to work on, and maybe even hinting on the solution or technique to be used (even if common knowledge for some), is a major contribution.

Let A = either paragraph above. Let B = "There exist advisors that contribute nothing to a research project, not even unappreciated but valuable background work.". Then A ≠ B.

It amazes me that among people who prove theorems for a living, there are so many who are reluctant to employ basic logic when it comes to dealing with personal problems. You cannot disprove the existence of abusive advisors by demonstrating the existence of non-abusive advisors. Some students have advisors who do valuable background work, and if the students are immature they may wrongly believe that their advisor is doing nothing at all. Of course this happens and no one disputes it. In no way does this imply that for all students everywhere who have problems with their advisor, the only credible explanation is that the advisor is providing valuable support and the student merely fails to appreciate it. Sometimes the advisor really is just taking advantage.

BM said...

interesting battle among anonymous students and anonymous advisors

Anonymous said...

Indeed, troll posts are known to instigate a lot of battles.

Anonymous said...

I strongly agree with one of the comments above about some profs being just middle man between funding agency and PhDs, post doc.
They have lost all their scientific skills and are only money manages.
The funding mafias, a menace which should be eradicated. Such people better of being CEO of some company or banks. At least, this way the research environment will remain pure.

Anonymous said...

interesting battle among anonymous students and anonymous advisors

Here's the most interesting part. If a student non-anonymously calls out their advisor for unethical behavior (through a blog post or more directly through the university), they likely lose their jobs and kill their careers. Even if they still graduate, no one goes anywhere in this business without an advisor's letter of recommendation. Professors on the other hand have tenure. It's easy to criticize posting anonymously when you have tenure, a privilege whose primary purpose is to ensure that you cannot be fired from your job for saying things your colleagues or administration or government don't like. But those with tenure should consider for a moment what sort of power imbalance that creates between a tenured professor and a grad student, who can be booted out for basically any reason whatsoever.

So what can a student in such a situation do? Well I don't know. But whenever I see a post (not accusing Michael of this but rather some commenters on this thread) that is based on the underlying assumption that all professors are saints and the only problem with advisor/student relationships is how to best communicate to students that every problem is due to their own misapprehension of their advisor's saintliness, I feel like there is a bubble of ignorance surrounding tenured professors that must be popped.

Let me be clear: Some advisors are saints. Some advisors who are saints have students who wrongly believe they are devils, and these students must be educated to think more carefully about the duties of their advisor before criticizing. But some advisors are devils, and their tenured colleagues are best in a position 1) to notice this, since they know from years of experience what proper and improper behavior of advisors looks like, and 2) to do something about it, because they don't face the same risk of career-destruction faced by that devil advisor's students.

Anonymous said...

@anon-3:48: Yeah, right, banish the profs who seek money to becoming CEOs of banks. Let's see you do research on an empty stomach.

People who make comments such as the above, and also those that suggest that their advisor "only suggested the problem" seem to grossly underestimate the efforts involved in getting grants and identifying interesting research problems.

Michael Mitzenmacher said...

Stefan --

As usual I thank you for your comment (and willingness to use your name). I agree with you that expectations are key, and that it is hard to break away from what is being done traditionally within a community. However, I feel your comment helps make my point that alphabetical order is to be preferred.

Specifically, statements like "While not a "rule" per se, its generally considered a bit tacky if a faculty member has their name first on a paper (there are a few exceptions, but they'd take more time to explain than its worth)." and "The premise is that people in the systems community remember the first author and they know to look to the last author(s) for the adviser(s) but the middle is largely undifferentiated by the reader" make the system seem a bit complex and perhaps even arbitrary in how it's "assigning" credit.

I'd like to think that even in systems alphabetical order is a possible alternative, perhaps used to denote that everyone worked on the project. I've used it for several systems papers in the past.

Anonymous said...

It amazes me that among people who prove theorems for a living, there are so many who are reluctant to employ basic logic when it comes to dealing with personal problems. You cannot disprove the existence of abusive advisors by demonstrating the existence of non-abusive advisors.

Heed your own advice: anonymous 9:16am never claimed nor implied that were no abusive advisors.

Anonymous said...

I would argue that the best approach depends on the field. In fields where the nature of the work creates very unequal collaborations, order by contribution is best. In fields where more equal collaborations are the norm, alphabetical is best.

To some extent, arguing which is better is like arguing whether uniform or exponential distributions are better. Clearly, it depends!

Michael Mitzenmacher said...

Anon #39 : I think you and Stefan have a point; perhaps, in fields with greatly disparate contributions, order-by-contribution is best. Although, in that case, I think the approach by Nature -- all authors should list their contribution -- is arguably better. I'm also not clear if things are really like that in most fields. I do understand it CAN be that way for systems papers, but is it really the norm?

Stefan Savage said...

I think its fairly common, by design, that systems papers have a "lead". Certainly in our group we try to create this notion a priori -- that one students will lead in a particular project while others will help, and that this relationship is reversed in other projects (this happens naturally when students come up with the project idea, but we try to engineer it even if faculty come up with the ideas). This reflects a reality that for many systems projects there are a range of "moving pieces" (this is not limited to artifact building, but there can also be different pieces of measurement or analysis work) far beyond what any one person can do alone and yet there is also a necessity for someone to own the "big picture", do integration and make sure the whole project succeeds. This structure helps students to work on more than one project at a time (there is a communal expectation that you'll help on the project's other people are leading just as they'll help on those you lead) and encourages them to develop non-technical skills (developing new research directions, research integration/management, leadership, etc) as they work in both roles.

As a consequence, this notion of "lead" is easy to identify and in the 80 some papers I've been involved with I think there were only a handful of cases where it wasn't clear (and perhaps only two where this became problematic).

Again, it comes down to culture and expectations. We're very clear with our students how we do things, we create opportunities for everyone to participate at all levels and we encourage all of our students to welcome co-authors rather than worry about credit (I think its self-destructive to worry about credit too much). The culture of author order is thus an extension of the culture of how the research gets done.

In principal we could identify what each person did (ala Nature), but I think this would hurt the esprit de corps that develops in a team project (instead of being part of a team that put together a cool result, you're now just the person who developed a technical widget). When it matters such things come out in letters... its not clear to me what's served by putting it in the paper.

Anyway, this is probably as much as I can say on the topic (indeed, far more than it deserves). Other approaches have their own advantages/disadvantages and may indeed be preferable at times (I've use alphabetical in at least one case for this reason). Thus, I'll be happy to simply concede the point after this rather than defend what the systems community does in general (bearing in mind that I don't see the systems community changing any time soon :-)

Yisong Yue said...

You can also just indicate that the first 2-3 authors made roughly equal contributions, e.g.:

http://people.csail.mit.edu/mcollins/papers/egjournal.pdf

In a multi-author computer science paper, people are unlikely to care about who is the 4th or 5th author. So long as the people who did the most work are recognized, it is fine.

Anonymous said...

I am also a grad student in (2) and can understand the angst of the other grad students in similar situations. To anonymous no. 3, I can say that they're lucky that the advisor contributed so much... my advisor is only a passive listener. In the beginning I would go into his office every other week and present things that I had thought about seeking to get feedback but he didn't seem sufficiently interested or involved. Never asked me to read any paper, didn't seem to have references for any of the problems I was facing that he was supposed to be the expert on. In the beginning I was a little confused, I thought that he just wasn't interested in the problem I had chosen. He would say the problem is interesting, but I never got any useful comments from him, in fact other grad students in the department I talked to and who weren't even collaborating on the problem had more useful feedback than him. Slowly it dawned on me that all he was interested in was results, a paper that he could put his name on by doing some editing, it seemed he had no problem in demanding (confirming politely) his name be put on the paper I had written, with the problem being picked by me and the entire result having been worked out all by myself. I don't think I will ever confront him on this, but I have no respect for him, even though he watches out for my interests by trying to get funding for me. In my opinion he is a very unethical person, and my personal experience makes me think about how much his contribution really is in the papers he has co-authored with other subordinates. I only hope that he is an outlier.

Anonymous said...

I have no respect for him, even though he watches out for my interests by trying to get funding for me.

Don't feel guilty about this for a second. Maybe some of that funding might come your way but it is not altruism. Your advisor would be applying for it whether or not you are there and no one, repeat, no one pays graduate students out of charity. They expect research in return and even if success is more difficult to measure compared to workers on an assembly line making Fords, research is a job like any other and you don't have to feel guilty that you took money in exchange for your work. Maybe your advisor doesn't mind that getting those grants also benefits you, but he's watching out for himself, and you have to do the same.

Good luck to you and I hope you end up swimming and not sinking long enough to get out and let your advisor fend for himself.