Too much reviewing is bad for you!

I clearly have been accepting too many requests for participating in Program Committees. I am reviewing for the top conferences in Computational Linguistics (ACL and COLING), AI (AAAI and IJCAI), and theoretical computer science (LiCS and ICALP). And there are also the smaller, more specialized meetings, which also require a serious amount of work to do a half-decent job. As a consequence, instead of doing my own work, I am really tired of trying to understand other peoples' work, at a breakneck speed.

Of course, the temptation to accept these invitations is huge. These are my three main fields of activity and a few years back I'd kill for "one" such invitation. So this embarrassment of riches is disturbing, to say the least.  (and I am not even starting on the big  issue  above of working for free to make rich bastards richer.) But sometimes I am reminded of why being a reviewer is not only hard work. You do actually get to know plenty of other work that is really interesting and that you want to know more about.

So I only recently got to know about Conal Elliot's work on "The simple essence of automatic differentiation". You can read/watch about it from Conal himself here. It seems that there's an awful lot more about it around though. That, I don't know about it, yet. But I also got to know about Conal's suggestion of writing papers, by writing blog posts. This seems a nice idea, if a little dangerous. If you're trying to do a Polymath project, everyone knows that sometimes things do not work. (This is research, not development, after all!) But in a research-by-blogging situation, what happens if nothing works? oh well, I guess the only way to know is trying it.

Meanwhile from Elsevier, a nice picture only on their blog.