It is no secret that industry labs are competing with academia at a pace never seen before. Most of the tech conferences now are dominated by industries with superior funding, human resources, and data.
What surprises me is that papers from industry are often written by a large number of authors. In fact, whenever I see a citation/reference with a whole string of authors, I can immediately predict that it comes from industry labs. Why is this the case?
For example, this very short and simple looking paper from Google is authored by 9 people.
Here is another very brief paper from Google, similar to the style of an undergraduate project. Now why is 12 people needed to put it together?
Take a look at another paper from Facebook. What looks to me to be a survey paper with no simulation or any equation required 17 authors.
Or this, again, very short joint paper from Apple, Facebook, Google. Why 11 authors?
Why does this Google paper (that has three labeled equations in total) require 16 authors? Are these people gaming the publication/citation count system or what.
Why does this other Google paper require 31 authors? Am I to believe that they all contributed equally?
Can someone please chime in as to why these papers require so many authors? I say many because I've seen student projects or theses written by a single person that (more than) rivals the depth of those papers that require some 15 people to write.
Answer
The work mode of an industry lab (or, even worse, of a group of day-to-dat software engineers in industry who just happened to become involved with a paper) is typically different to your average academic lab. To wit:
Academic project (simplified, of course):
Somebody (a PI, the student themselves, or somebody in between) has an idea for a research paper. A person (often, but not necessarily, a student) takes responsibility for executing it, and collaborates with their advisor and, maybe, one or two other students on executing it. Importantly, for this "driver" of the project it is usually of high importance (because it goes into their thesis in the case of a student), and they invest a significant portion of their time on it. The other collaborators usually only invest time in specific roles (giving feedback, doing specific analyses, helping with paper writing, etc.). In short, there is often one person responsible for bringing the project all the way.
Industrial project (particularly in the case when a software company is reporting on aspects of their practice, as in many of your examples):
Somebody (often a member of the research division of the company, sometimes an academia-affine practitioner) has an idea for a paper. They start working on it, but they cannot commit more than a fraction of their time on it (because even for industrial researchers, writing papers is often much less a core focus of their job profile than for students or academic staff), so they gather collaborators, who likewise cannot put more than a few hours here and there on the project. Additionally, for such "report of a practice" kind of papers, many people have typically contributed to establishing the practice, and it is bad tone to leave out key players (unless they don't want to contribute to the paper). In short, while there may be a coordinating person as well, nobody really has the time to work on this paper full-time, and many people have somehow contributed to the work being reported on, so automatically the author list swells.
TL;DR: results in industrial practice are often much more of a product of team work, rather than the "every student works on their own thesis" model that we usually have in academia. This is also reflected in the length of author lists.
No comments:
Post a Comment