July 2012
« Jun   Aug »




Rant: A breed of well-informed idiot

I’m not liking the peer review system right now. Since I’ve started grad school, I’ve had every one of my papers rejected. This leads me to the following theory: reviewers are some breed of well-informed idiot. They may be smart in the field, but I think they don’t read closely. I know that I’m biased into thinking my writing is top notch and that part of the bias comes from putting an inordinate amount of effort into the product, but the reviewers responses on my most recent rejected paper give me the impression that reviewers didn’t understand what I was trying to sell. It’s probably the case that the well-informed nature of reviewers gives them a selective focus, so that they miss what I’m trying to communicate.

Now, I shouldn’t take the review system too personally, because most of the conferences have a 30% acceptance rate (some 12% even). That means reviewers reject by default. Since many academic teams are producing good work, the competition compounds. You must have good work, it must be well-presented, and the paper must both tickle reviewers fancy and simultaneously avoid offence. However, this string of rejections has left me depressed and one of the reviews picked on a sentence that I felt necessary to include in the paper. So some self-righteous indignation is in order (and what a drug it is!).

My technical writing proceeds very slowly. I hem and haw about what words I’d like to use. I try very hard to produce active sentences in E-prime. I try harder to avoid self-referencing (e.g. ‘this paper’ or ‘this work’) because I feel it distances the reader. After some collection of words finally makes it through these hurdles, I then mill it further. I slice and dice. Rearrange clauses, re-order phrases. Vary words and structure to avoid repetition. Most sentences are rewritten 3 times and contain at least 2 ideas before I can finally move on to the next sentence.

Maybe after this extended and thoughtful process I expect too much of my reader. Many ideas are left implicit or implied. Especially, when those ideas are but one logical step away. Reviewers probably prefer ‘short, declarative sentences’ and have neither the time nor the capacity to read carefully and think critically. After all, they have a whole stack of papers to hand out to their grad students. I shouldn’t expect that they should strain their minds in order to perceive details only tacitly suggested.

In this last paper, I was trying to sell the idea of first-class information flow labels. It requires one syntactic change to the JavaScript environment within a browser: the introduction of a labelof operator. All other changes, including the introduction of new FlowLabel objects, do not technically count as syntactic changes, as they are defined by the host environment and lie outside the JavaScript language specification (just like the History, XmlHttpRequest, and DOM objects).

All three reviewers expected to read about the underlying semantics of the information flow labeling system and rules of label propagation. These were not included within the paper for two reasons: (1) the idea of exposing the labeling system as first-class JavaScript objects is incremental and (2) independent of the underlying propagation system. Under these guidelines, I made the mistake of submitting a short paper. I thought the idea of first-class labels was straightforward enough that reviewers would see its independence and recognize that the extended discussion about the supporting framework is nonessential and a distraction. Remember that I said reviewers are a breed of well-informed idiot? I didn’t pander to their knowledge base. I didn’t give them what they expected. So they didn’t identify what I was selling, in spite of the fact that the TITLE of the paper clearly and concisely and DECLARES the idea.

Finally, the last reviewer picked on one of my sentences. If I may quote:

As a minor comment, the sentence “To align with common JavaScript development practice, our labeling system follows the functional programming paradigm.”? is quite meaningless.

This is one of those sentences that I had to change because, even after going through my rigorous sentence assembly process, my post-docs couldn’t identify what I was trying to say in the first draft. The end result, what you see quoted above, is much more clear, and I assure you, quite meaningful. A four page paper doesn’t give one a whole lot of room, so I had to really cram ideas into this sentence. I’m basically saying, as concisely as I know how, all of the following:

  1. JavaScript developer practice the functional programming paradigm. They use lots of closures and lambdas.
  2. I don’t want my system to feel foreign to such developers.
  3. Therefore, it is also design with a functional paradigm interface.

I can only conclude that this reviewer doesn’t know much about JavaScript development, which is not a huge fault. But to miss entirely the fact that I’m making extra effort to follow the programming language paradigm used by the developers that my system targets so that they can feel comfortable when using it? I must be expecting too much of my reviewers. It’s clearly a strain for them to see the missing middle 2 when shown the sequence 1 and 3.

Well that rant was longer than I expected. Self-righteous indignation is a hell of a drug.

Leave a Reply