Writing Papers

This is written from the perspective of writing papers, but much of it also applies to writing grants. Also, writing papers is hard; this is a living document as I/we learn.

General Approach

How many should you write?

Useful Resources

Authorship

Schedule

Communication

Technology

Open Science

Revisions and Rebuttals: Working with Reviews

General Approach

In research, our goal is to generate and share new knowledge. If we do not communicate the results of our work, we might as well not have done the research. Despite all of their flaws, papers are still the primary way we document and communicate that research, and so investing in becoming a better writer will pay off for you and the community.

How many should you write?

This is not the right question, but it’s often asked. My general view is that we write a lot of papers in our community, and in some instances, we’d probably be better served by combining papers to tell a more complete story. So, my general preference is to write fewer papers that teach the reader more per paper.

That preference is tempered by a few things. (1) Once we’ve learned something worth sharing, the community benefits from us getting those ideas out there. (2) Many in our research group do cross-disciplinary work, and might need to write separate publications to share ideas with different communities; e.g., one paper in a health journal and one in an HCI conference. (3) You will apply for internships and jobs, and the market values having publications out there.

The heuristics I have (mostly) settled on for deciding whether to publish:  


From the career perspective, I think it’s a good goal to be proud of and confident in the quality of any publication an application reviewer might choose to check out from your CV.

Useful Resources

Authorship

The simplest heurstic for whether someone should be an author is whether they have contributed substantially to the intellectual framing of the work described in the paper. Note that this is distinct from contributing to getting the work done.

All authors should be given the opportunity to edit and comment on the full manuscript. In turn, when you are lead author, you should ask them to do work to help you.

Also see the ACM guidelines. Below, I include some notes about authorship inclusion and order, but I first want to say: this is hard to navigate, and I’m happy to talk with you about this early and often.

Authorship versus Acknowledgement. If someone did a lot of work or gave pivotal feedback that advanced the work, they should at least be acknowledged. For someone who contributes intellectually to the work, the difference between acknowledgement and authorship can be harder to navigate. I often look to sustained or critical contribution as the distinction.

This distinction is also more easily navigated up front. As you look at a planned project and where you might want help, you can ask people for help through lab meetings / one off feedback sessions (explicitly as not an author) or you might ask for more sustained involvement as an author. This clarity helps everyone. You all work on interesting things, and when you ask others to be involved, it’s easy to get caught up in it and find that one has gone from giving feedback to drafting entire sections.

As for deciding whether to ask others to join you as an author, my general approach is to be inclusive. Adding more authors does not diminish your contribution (especially if you are the lead author), and they can help strengthen the work and anticipate issues that reviewers or readers would later find. (For some of the health work we do, it would be irresponsible to proceed without a health expert on board from the start.) More authors can, however, increase coordination costs to a point that it is not useful. When writing more “viewpointy” papers, more voices can also water down the argument especially if they disagree about the main points, and so you may prefer to keep your author list shorter in those cases. However, when you can get a large team to agree on a message, that can also increase the weight with which it is delivered.

Please check with all authors about which funding sources should be acknowledged.

Order. HCI and related fields have made a mess of author order, to the extent that other than the first author position, order has littel meaning or even contradictory meaning. In general, my approach is: People without doctorates in decreasing order of contribution, followed by people with doctorates in alphabetical order.

There are many exceptions to this. When a faculty member leads a publication, they will be first. When a colleague from another discipline with more normative expectations about author order joins us, we may revise the order to conform to their field’s expectations -- especially when publishing in a field in that venue.

Some medical publications use contribution statements. While that comes with its own awkwardness, I increasingly find it appealing for reducing ambiguity (and thus allowing teams to be even more inclusive with authorship, since there’s less of the “we can’t add N because they did so much less than A & B”). I am happy to talk with you about whether that is the right choice for a paper.

Schedule

I encourage the following milestones, though being earlier is better particularly around deadlines for which many in the lab are writing:

These deadlines may need to move earlier based on authors’ travel or vacation schedules. For the last two, if these deadlines are not met and I have multiple papers for the deadline, we may need to defer the paper until a later deadline. Rushed writing leads to rejected papers (or, worse, accepted but bad papers), promotes unhealthy work habits, wastes reviewer time, and squanders our reputation.

Also, to repeat: I discourage long hours and, especially, all-nighters. They aren’t sustainable, they harm you, and they harm the work. Occasionally we misjudge how long a paper will take and may decide it’s worth a late night to finish it, but if you find yourself doing this repeatedly, we should discuss how many things you are taking on and how you are planning your time to complete them.

Avoiding long hours is easier said than done. Some techniques: setting aside a short amount of writing time each day or otherwise booking meetings with yourself, decomposing the task into smaller goals (even a paragraph!), joining a writing group, finding an accountability buddy.

Communication

The lead author is responsible for engaging authors around where you need help and support, as well as coordinating schedule. If editing asynchronously, make and communicate decisions such as: (1) we’ll use this thread to pass paper versions around (please don’t) or (2) we’ll keep the paper in this shared Dropbox folder.

Every author should have the opportunity to read and approve of the submitted version. It is important you engage people and provide drafts early enough to facilitate that.

In addition to having boundaries on when you will work, you should inquire about and manage your writing process to respect your co-authors’ boundaries.

Technology

The lead author, in consultation with other authors, is responsible for picking the technology. I generally encourage:

  1. Outlining and initial drafting -- Google Docs, because it allows synchronous editing and a focus on content
  2. Final formatting - Overleaf (LaTeX) or MS Word.

You can also use whatever reference manager you want, but unless it’s contained within the editing platform (e.g., BibTeX for LaTeX), you’re on your own. Reference managers rarely play well with collaboration.

Open Science

To help others interpret our results and to help others reproduce our studies or similar studies, I try to share as much of our research product and process as possible (while noting that participant privacy comes first). This includes sharing:

To share these materials, we might submit them as supplementary materials with the publication (if the venue allows) or create a git repository which we then cite/footnote.

Revisions and Rebuttals: Working with Reviews

After submitting, you’ll get reviews back. For some venues, it’s a final decision -- no rebuttal, just an accept or reject. For others, we’ll need to revise and resubmit or to write a rebuttal.

Whatever the decision -- yes, even for accepts -- the first step is to read reviews. Some may seem quite wrong to you, especially on first read. Put them away, go do something else. Do not tweet or post about them on social media. Even when a review is so problematic that we decide to do something about it (in seven years as a faculty member, I’ve run across just one of these as an author), public comments are not the way to handle it.

Once you are past the emotional reaction, the next step is to sort the reviews in a table, using Google Docs (not sheets!). I recommend three columns: one with the review text, one with brainstorming / reactions (and, yes, any snark or frustration we need to get out), and one with the constructive changes we plan to make or response to why we are not changing it. Even for comments that seem off-base to us, it’s often valuable to examine our paper to see where we invited that misunderstanding: if the reviewer misunderstood, so too will some future reader. I also think it is valuable to remember that reviewers are volunteers: we may not have gotten the outcome we want, but by and large, people in the community are generous with their time and perspective with the intent of helping make the work better.

The steps after that depend on whether this an accept, a new submission to another venue, a revise and resubmit, or a rebuttal. There’s lots of advice out there on those; we can discuss when we get to it.