Thursday, April 30, 2015

Co-chairing Eurocrypt two years running, reflections of a program chair (Part 1)

How do you get selected as program co-chair in the first place? Well having been asked for ‘two' major IACR events (CHES 2008, and then for the pair Eurocrypt 2014 and 2015) it seems to me that the process is that somebody drops your name in a board or steering committee meeting … you better chose your ‘friends’ carefully! Once asked there is no way out as such opportunities do not come often and so you better go with the flow! 

I was asked often over the last couple of years how the ‘co-chairing’ was for me.  Now, because CHES embraced the tradition of having two chairs from the beginning (one selected from academia and one from industry) I have only ever have been a co-chair and I know nothing else. Hence, my standard answer was ‘fine’ (remember also my prior remark about friends) ...

Silly jokes aside: I cannot imagine doing such a role alone. Being part of a small team suits me. I was also blessed by working with three exceptional colleagues: Pankaj Rohatgi during CHES 2008, Phong Nguyen during Eurocrypt 2014 and Marc Fischlin during Eurocrypt 2015. I had a great rapport with all three, which contributed to being able to enjoy the madness! Over the three times that I co-chaired, I experienced different ways to distribute the workload: by dividing up the set of papers into two subsets, by dividing up into topics, and alternatively by simply time-slicing: one takes the lead on the various tasks when the other one can’t. All these strategies worked. 

Another question I was often asked  was how much time it takes. Well ‘a lot’ is my answer. I am not a fan of exact time keeping (no matter how many work sheets my employer asks me to fill out ... I have a ‘template’ anyway), because it costs me time that I don’t have! But if you want to get a feel for the input required then think about looking after about 200 ‘researcher entities’. They send you a paper and then they want feedback, which you get based on other people’s opinion. This should not be a time consuming task once all PC members are appointed, as the number of interactions is limited to a few which require manual involvement (thanks to Shai’s effort with our Web Review System). If only that was the case. My Inbox lists just under 1000 emails for Eurocrypt. Removing automated notifications for submissions and results and final versions, this leaves around 400 emails. Because I can’t afford a personal assistant, I have to read/think/act/respond/ myself, which in some cases takes a few minutes and in many cases requires many more  minutes during the ‘act’ and ‘respond’ phase. Luckily I have held a personal Fellowship since 2011 and so could spend plenty of time …

Yet another area of interest was that around how we (i.e. Marc and I) managed to convince the PC to go with our suggested tweaks for the review process (we introduced a two stage process and changed the submission requirements around paper length). My answer was ‘with difficulty’. There was at first a number of people who expressed their concerns in a very vocal manner. Which is fair enough. After quite some email exchanges, it became clearer that not everybody who was more ‘quiet’ was against our proposal. After some people offered us their support more openly, we were lucky enough that everybody ‘gave it a go’. 

It’s useful, I find, to experiment, but only if afterwards one also asks for feedback and reflects on it. So after all notifications had been sent out in 2015, and people had some time to ‘cool down’, I sent out a Doodle poll asking people for their views. Here are the results (121 people responded, I list the number of votes each statement received):

I like the fact that there were two rounds: 92 
I hated the fact that there were two rounds: 12
I think it was useful to only elicit rebuttals for the second round papers: 63
I think that everyone should have been allowed to write a rebuttal: 38
I think the increase in page limit was useful: 86
I think we should stick to 12 pages as limit for always and ever: 12
I want to have a page limit of 20 pages LNCS in the future: 38
I want to have a page limit of 25 pages LNCS in the future: 19
I want to have a page limit of 30 pages LNCS in the future: 31

I also received anonymous comments which I will supply to the IACR board in my report. They mainly touch on the perceived lack of impact of the rebuttals that authors wrote on the final decision. I think this is a valid point. As all second round authors were given the chance to write a rebuttal, most authors did take that opportunity. But for many papers reviewers hadn’t directly required clarifications from authors and so these papers probably didn’t benefit much then from that opportunity. Some papers however did get questions very directly to be addressed during the rebuttals and there the rebuttal had a useful role to play (even though it was limited to a single
iteration, i.e. we refrained from allowing lengthy conversations between reviewers and authors, which we inevitably would have had to moderate to preserve anonymity). 

I was to do some chairing again (which I will not any time soon) then I would want to discuss the question of only giving selected papers the opportunity to respond (based on reviewer recommendations). But one thing I took away from the review process is that the two stages, where one is short and the second one focused on only a smaller set of papers makes a lot of sense. 

This is the end of part 1 … 

... come back to the blog in a few days time to find out about what it was like to make decisions about accepting/rejecting papers, our attempts to create parallel sessions, and to read up on how the attendees though about the experiment! 

No comments:

Post a Comment