Waste Paper

It’s paper-writing season, which is a few months of the year chiefly spent in LaTeX writing up results for a bunch of games and AI conferences. Papers are the classic measurement of academic output: you describe your work, your peers decide whether to accept it for publication, and thus publishing more papers indicates more of your work was considered good. I don’t think anyone really thinks this is a good system for assessing academics any more, so I won’t go on about it here, but in theory there are lots of other reasons we write papers: to summarise, to archive, to disseminate. Even if the system of evaluation through paper-writing is stupid, it’s still a worthwhile activity.

Recently I’ve been feeling differently about this. In late December I began work on Danesh, a tool for analysing procedural generators. In January I wrote a paper for FDG about the tool, and it felt pretty okay. It’s satisfying to write about work for the first time, and I felt productive (partly because writing papers becomes a ritual that makes you feel good about yourself, I think).

I submitted the paper towards the end of January, and it’s currently up for review. Assuming that the paper gets accepted, I will discuss it for the first time in August of this year, seven months after the paper was written and eight months after I started work. Seven months is a pretty fast turnaround, too, partly because I built the system so close to the paper deadline. Had I started the work a few weeks later, I might not have submitted a paper to FDG until the following year. A rejection pushes this even further back: at ICCC 2015 I presented work that I had completed in 2013, but had been rejected from the 2014 conference. There was no other suitable place to submit it, so I had no choice but to wait another 12 months and resubmit it with improvements.

Danesh is already markedly different from the tool I described in the paper I wrote three weeks ago, and many of the hypotheticals described in it are now implemented ideas or ruled-out dead-ends. Should the paper be accepted into FDG, Danesh may be unrecognisable by the time I come to present it. Of course, I can write the presentation to reflect this, but anyone who checks the proceedings and reads the paper will be reading ideas that are almost a year old.

Perhaps there’s no way around this and this is simply the best way we’ve found to work. All I know is, it’s making me feel distanced and demotivated about the cycle of paper-writing. I can talk to researchers, developers and other potential users about Danesh right now, I can share my code with them, upload videos, get feedback and adapt my ideas. I get a lot of value from doing so. In their current form, papers are now something I write for the benefit of the system I’m in, and that’s about it.

I think that’s all I have to say for now — I just wanted to follow up on some fairly glib tweets I made last night before falling asleep. Off to go write another paper now. Cheerio.

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.