Why do TCS papers have author names in alphabetical order of their surnames?

Question

I am currently doing a Ph.D. in Theoretical Computer Science, and any research paper I encountered so far has the author's names in alphabetical order of their surnames.

For example consider the most fundamental book on algorithms: "Introduction to algorithms" by Thomas Cormen, Charles Leiserson, Ronald Rivest, and Clifford Stein.

Also consider the book: "Parameterized Algorithms" by Marek Cygan, Fedor V. Fomin, Lukasz Kowalik, Daniel Lokshtanov, Dániel Marx, Marcin Pilipczuk, Michal Pilipczuk and Saket Saurabh

Basically take any paper in the TCS domain, all follow this pattern. This pattern is not followed in other domains where the authorship is decided based on the individual contribution of the authors. In other words, a person having the most contribution to the paper is given the first authorship. Likewise, a person with less contribution would have his/her name appear later in the list of authors.

I consider this norm, fundamentally flawed. Can somebody provide a good reason as to why such a norm is followed in the TCS domain?

Answer 1

The American Mathematical Society has released a statement (pdf link) about the commonly accepted practice of listing authors in alphabetical order in mathematics. Their reasoning applies, in my opinion, to TCS as well.

This kind of theoretical research is often done in small groups of 2-5 people. In close collaboration it can be difficult to determine who came up with which idea and/or whose ideas are more important for the main results. Using the alphabetical order lets us circumvent arguments about it and default to "everyone contributed equally", which is often (but not always) more or less the case anyway. It's also easy to explain to non-mathematicians by e.g. showing the above statement. When you apply to math/TCS positions, the hiring committee will be aware of this convention and won't hold your last name against you. My impression is that researchers in fields close to math/TCS tend to know about it as well.

In biology and other fields where research is based on experiments, papers are written by large groups whose members have distinct roles. A professor has an idea and obtains funding, one researcher designs the experiments, some lab assistant runs them, someone else analyzes the results, others write the paper. In this context it makes more sense to differentiate the authors by their contribution.

Answer 2

In theoretical computer science (and mathematics), there is also the issue that the work develops and heavily changes over time:

• You state and prove a lemma.
• After one week, the lemma has become irrelevant since one of your co-authors has proved a much stronger statement.
• After one week, this much stronger statement also has become irrelevant since the structure of the entire approach has been simplified by another co-author.

So the individual steps that the research has taken are not necessarily visible in the final paper. How do you want to assign credit to the various co-authors? How do you want to rank them? Should only authors be listed who have contributed something that explicitly shows up in the final version?

Answer 3

This is complicated and somewhat nonsensical, like most social practices involving humans.

In mathematics, there is a very strict tradition of alphabetical ordering. This is not "optimal" in terms of informing readers how much each author contributed, but it has the advantage that there is no way to game the author list.

The other tradition, used in engineering and the sciences, is to order by contribution, so that the person who made the biggest contribution is first author. This is perhaps more informative, but it is also more vulnerable to efforts to game the system, and also to disagreements about who contributed how much. Eg, if someone in your author group will be going on the job market real soon, there is a temptation to make them first author.

This leads to further innovations like reverse-seniority ordering, which is an attempt to avoid the cynical careerism of gaming the author list, while retaining its benefits. Most likely the least senior people are going to be on the job market soonest, right? Also it tends to lead to more motivated grad students, especially if their last name is Zubhasamanvit or something.

In algorithms and complexity, the usual pattern is to follow the example of mathematicians, and alphabetize the author list. Areas of CS more influenced by engineering, like operating systems, will use the engineering order. And some areas, like programming language theory and semantics, have no agreed standard at all, since it's a hybrid of pure mathematics (eg, categorical logic) and engineering (eg, compilers).

The one good outcome is that the whole state of play is sufficiently incoherent that AFAIK very few people take author list position all that seriously when hiring. People notice, but it's rarely if ever a decisive factor.

Answer 4

Think of authors order on paper as a game for credit, which in turn influences how scarce limited resources (and in case of academia very limited) distributed. It is positions, founding money, opportunities, ... The distribution of resources are exponential on the other hand: some get top positions in top universities, with prestige, good students, and funding. Most get nothing and leave the field.

Think if contributions of people in the field as a bell curve. There are a few stars, they would be recognized easily and get resources. There a few no-chancers, they would be recognized easily and not get resources. Around the mean you have a lot of concentration where it is not possible to easily distinguish, and side factors came into play.

Paper authorship is one of these. Instead of evaluating and pedicuring judging if someone would really go to contribute significantly, we rely on these place holder metrics.

Some fields try to solve the problem locally and allow to authors to sort out who contributed how much. This in turn leads to some times dishonestly ordering authors, e.g. the most senior person with tenure tried to put their soon to be graduate students as first, etc. Mathematicians and TCS as a subculture feel that they are above this kind of human politics, so they instead try to systematically avoid the problem. However the field is full of human politics, as any other area. It takes other forms.

Generally academic mobility is relatively low. If you are a star, you might get tenure in a good university. Otherwise, if you are contributing more than another person, but they have advantages like being in a more prestigious university, working with more famous people, etc. it is hard to break in. Those who get to decide how to allocate resources often turn into like minded circles who naturally like to only work with people similar or familiar to them. Most don't go to do any fundamental works later.

As a general advice, think about why you are really in the field? If you are one of them people who are naturally driven to it, that is your inner satisfaction comes from skiing the research and talking about it and learning and discovering, then keep that in mind and don't worry too much about these political games. Enjoy doing good research. If your satisfaction comes from being smarter than everyone else, either you really are, and most likely get what you want, or you really are not, and would soon learn that. If a major factor for you is resources, pick a field with more abundant resources, if you are even an average graduate student in an average university, you have the skills that can make you earn more than what top academics earn in the industry, or have impact beyond those of typical academics who spends years improving the epsilon that no one uses or really cares about (most academic papers even at top constructed are of that category). And so on.

No system is perfect or immune to human politics. Instead of worrying about the system being flawed (which one is not). Don't expect the system to be fair and give appropriate recognition. Often results are children of their time, despite our huge egos, most of us know that for almost all results, if it was not us, someone else would proven it, maybe a few years later. Even Godel and Turing's famous results were children of their time, similar ideas were present in the works of other people at the time. Cold war was a great social experiment in science where the west and Soviets independently proved major results roughly at the same time.

So check with yourself what you really care about and focus on that and don't worry too much about things like these. If you find yourself worrying too much about fairness if distribution of resources in academia, know that there are much better paths to get resources than academia, particularly anarchy fields like TCS and math.

Answer 5

Because in a good collaboration, the total contribution of the individual authors exceeds in total a 100%, at least emotionally. All authors should feel it is their paper. If you start ordering authors by other means, then you have to start ranking the contribution of authors - which is a big can of worms that nobody wants to handle (because agreeing on ranking among people is impossible, see Arrow's theorem).