Saturday, October 14, 2006

A Wobbly Chair

One of the arguable joys of being a program chair is dealing with unusual requests. I'm co-chairing CC 2007, a conference on Compiler Construction. CC resides within an umbrella called ETAPS, a confederation of conferences that meet every year in Europe. The advantage to working within the ETAPS framework is that ETAPS deals with the administrative details, from venues to publishing, leaving conference chairs to think thoughts of pure intellect.

Sort of.

I was recently contacted by a researcher who is a good friend who works for one of our leading technology corporations. (I will not even hint at their name: once you learn more about their lawyers, you'll see why. A wayward marksman is far more dangerous than a precise one.) He asked for the ``confidentiality policy'' that we follow. He clarified that they ``are filing a patent on parts of the paper'', so the lawyers were asking for the policy.

CC doesn't have a policy, and neither does ETAPS as a whole. It would be rather tricky for CC to have a policy, actually. Recall that ETAPS contracts the publishing? Once a paper is accepted the publisher takes control of the content, and that is done on terms settled by ETAPS. Therefore, any policy that CC did adopt would probably have very limited applicability, and worse, would have the potential to directly contradict the contract signed by ETAPS.

My friend pointed out that both ACM and IEEE did have such a policy and pointed me to the policy of ACM SIGSOFT. It makes fun reading from this perspective. Note all the things it says about confidentiality: First, it describes a meta-rule. Then it discusses the confidentiality of reviews (but not of papers). Then it discusses the confidentiality of discussion (but not of papers). Then it provides a delegation obligation, but still doesn't say anything about papers. Then it makes a statement about access control, but not about confidentiality. And finally, it specifically addresses to papers. To wit: "Neither SIGSOFT nor ACM guarantee the confidentiality of the submitted manuscripts".

This is the confidentiality policy that so satisfied the company's lawyers.

It's also worth noting the duration we're talking about. CC submissions are due mid-October. The decisions return early December, and final papers are due early January. This means the lawyers were trying to protect a window of not even three months. Of course, when it comes to the law even a day can matter, but that begs the question: if this patent mattered so much, why risk exposure of this content at all?

Given the number of conferences we have these days, chairs are, I think, expected to be utterly solicitous, cloying sycophants who will do anything for a submission. I couldn't resist. I told my friend that (a) we offered precisely the same confidentiality guarantee as SIGSOFT does, so it ought to please his lawyers, and (b) there was an extraordinarily simple and obvious way of protecting the confidentiality of the paper's content....

No comments: