Let's review talks [Updated]

[Update: I see I'm not the only one talking about talks this week.]

At ISIT 2012 this week on the campus of MIT, I've noticed that ISIT has changed very little since the first time I attended, in 2003 in Yokohama. In fact, as far as I can tell, the format of ISIT hasn't changed much in decades.

To be clear, ISIT is a popular, prestigious, and high-quality conference, so I wouldn't want to go all "new Coke" on it. But there are two things that bug me a bit:

• Too many parallel sessions make it hard to find the most interesting talks; and
• In general, there's lots of great work but not so many great talks.

The plenaries are always great talks, and I would learn so much more if ISIT could be one track, all plenaries. What makes the plenaries so great? These are specially invited talks from senior professors, who can be depended upon to give great talks. But if this were model for the entire symposium, it would be unfair to students and junior professors, who would never be asked to present their work.

So here's my suggestion. We have the technology to record our talks. I do it all the time, and you don't need specialized equipment -- I own at least five devices that can make a video recording, and I bet you do too (phone, laptop, ...). So why not record and submit your talk for review, along with your paper?

Papers should still be accepted or rejected on their own merits -- but only the best submitted talks should be accepted for presentation at the conference. So we would have fewer, but higher quality, talks; ideally, we could cut ISIT down to a single track of very informative and engaging presentations. We would all have to think about how to make a good talk, and even the rejected talks would get feedback for improvement. (In this world, accepted papers with rejected talks could go to an expanded poster session, or some other means of offline presentation.)

And if you want some pointers for your own talks, here's Frank Kschischang's famous Giving a Talk guide.