Wednesday, June 06, 2007

A Modest Proposal: Paying to Play at Conferences

I am on a program committee that is struggling with an explosion of submissions. There are way more than we expected (or, perhaps, can reasonably handle), and what really blows is that we apparently don't have the money to pay for quality conference paper management services.

In general, conferences have all sorts of problems with paper submissions. Every good conference is familiar with receiving a certain volume—sometimes a disturbingly high volume—of papers that are either too weak or too far off topic. These authors could easily save the program committee members (and themselves) a great deal of effort by just perusing past volumes, which would rapidly help them realize their work does not fit; but there is currently no disincentive in submitting (even the same paper to multiple venues). It's a sorry story all round.

I therefore have a fairly obvious modest proposal: charge each paper a submission fee of USD 10.

Here are various considerations:

Apparent disadvantage: It hurts those who can't afford that sum. However, this amount is absolutely nothing in the face of the other conference costs. Even if you lived around the corner from the venue, and the conference had a one-day registration fee, the cheapest you could do would be a minimum of about USD 250, making the submission fee a 4% overhead. But compared to the more realistic cost of a conference, it adds an overhead of about 0.66%, i.e., nothing.

Advantage: Conferences can use this money to pay for professional services for the submission, review and response phases. (The cynic will say, “But they won't!” But they'll have a harder time justifying why not.)

Advantage: Processing the submissions costs the program committee and administrative staff time and effort, so it seems reasonable to ask the authors to pay for it. (This is no different from college applications, etc.) Right now, they pay only if they get the (presumed) glory of an acceptance; but there is no (direct) cost in trying without success.

Advantage: It reduces the number of irrelevant, off-topic submissions.

Advantage: Some authors create havoc by submitting the same paper multiple times, etc. They'd be a bit more careful if they had to enter a credit card number each time.

Observation: This system appears to not have perverse disincentives. If, for instance, your friendly neighborhood oracle told you you have a 100% chance of acceptance, you undertake no risk at all in paying to submit. Thus, it hurts the best papers the least.

Mitigation of Disadvantages: You could waive the submission cost to select authors to encourage them to apply, just as we do application costs for other activities. Of course, choosing whose costs to waive can get controversial, and even counter-productive. One elegant solution would be to waive it for people whose papers were previously chosen for award nominations and the like, because these are precisely the people whom you want to have submit again (and they're a small enough group that they don't cost you much).

Concern: It may be more difficult to implement this using, say, credit-cards from some parts of the developing world. There are some alternatives we could consider in these cases.

I expect the first response most people will email me will be, “But then we won't have any submissions left!” Anyone so inclined is hereby expected to also explain why that's a problem. (-:


John said...

The analysis of costs isn't completely accurate. Some venues are so happy to have accepted work that they give rewards for it -- like free registration. So for THAT person, the cost of the paper has gone from 0 to $10.

One possible solution (proposed for mathematics journals about 25 years ago, but considered too radical at the time) is to say "You pay $10 to get reviewed. If you're accepted, we knock the $10 off your registration fee." After all, as an author I *am* contributing something to your conference -- you'd be in deep trouble without me and my kind. :-)

Another problem is the disincentive that the fee would generate for students (perhaps).

By the way, the problem of "horribly off topic" or "woefully bad on the face of it" papers can be resolved, in part, by giving the program chair "executive privelege" -- the right to reject any paper on these bases. If you don't trust the chair to do that fairly, you've got a broken conference already.

The real problem with your proposal is that it examines only one side of the issue: how to arrange for less work for the committee. As a frequent committee member, I find this laudable, of course. But if there are actually lots more good submissions, then the problem is "how to create a new conference with stature similar to this one, so people can submit some papers there." Of course, if there are ENOUGH good submissions, then you must also have lots of smart people who could serve on the committee of the new conference.

I actually believe that the lack of willingness to review ("I'm so busy and important that I cannot be expected to waste my time reviewing!") is a far more serious problem. Indeed, there's something to be said for a mechanism in which...let me see:

1. Each graduate student gets 10 points from the granting authority.

2. If you review a paper, the person who asked for the review will award you some points...say, zero to three. (People who write one-line reviews get zero points, for instance.)

3. When you submit a paper, it costs you, say, 2 points.

NOW you've got enough reviewers to handle your paper load :-)

-John Hughes

Jay McCarthy said...

I support the proposal! Definitely! I also am in favour of aspects of John's addendum of reimbursing the cost if you get accepted. You could even decide to reimburse the cost to anyone who gets above the "out-of-topic" review. This is essentially the same as a deposit to use a university facility, just to protect against people who will cause damage.

Philip Wadler said...

I doubt this would achieve anything except some hassle for submitters a lot of hassle for conference organizers.

Ten dollars means much less to me then the effort to write a paper, and I suspect that remains true even for the sort of submissions you intend to discourage.

One important step is to remind referees: spend the most time on the best papers. Think of the author as having earned your time by his effort to write a good paper. If the paper is woefully bad or way off topic, then the referee should have no compunction about taking little time to write a short review that says just that.

I can't see why you are worried about the cost of high-quality conference management, since you wrote CONTINUE, the best of the systems I have used. But I'll remind you that I helped SIGPLAN negotiate a bulk deal for conference management with START, so we have reasonable quality conference management available to every SIGPLAN sponsored conference (and SIGPLAN covers this cost).