I think the biggest mistake is to over plan a panel.Stephen pointed us to a really great post by Derek Powazek - How to Moderate a Panel that has some really good advice on moderating. His first bit of advice is to get to know the panelists in depth. Couldn't agree more. He also tells you to not have the panel before the panel. Don't have them get together ahead of time. I also agree. Actually, I agree with everything that Derek is doing. His post is well worth bookmarking to remember.
I would add a few things to what Derek said:
1. Decide ahead of time what you want to get to in the panel and where you won't go. I've gone to many very frustrating panel sessions with experts on the panel and all we do is here about things at a surface level. That's completely the moderators fault. These folks are experts. The panelists get a total of about 15 minutes of airtime. As a moderator, you need to make sure you get them into depth.
2. Because of lack of airtime, I like to give panelists a sense of the topics we'll cover and the kinds of prompts I'm likely to use so they can think of concise answers or what they want to talk about. I don't ask them to talk with each other about this or even me (although I mostly know where they'll go based on conversations). It's still an impromptu answer, but I've found that panelists normally get there quicker if you let them know ahead of time.
3. Push the panelists. As Derek says, as the moderator, you have to be the audience's advocate. Panelists sometimes stop short or go astray of really interesting things. You sometimes have to help them get there.
4. Highlight contrasts and similarities. Contrasts (or better yet disagreements) are the most fun. But it's also great when you have Microsoft, Google and IBM on a panel (I've moderated that before) and point out where they are saying the same things in possibly different ways.
5. Involve the audience as much as you can. While there's not much total time in a panel, it's often really good to ask the audience for quick input on particular topics. It's a nice reality check and makes sure that it's not an us/them thing.
My experience moderating Stephen was quite fun - but honestly - I wasn't sure how it was going to be. He and I have a very different interpretation of what Derek is saying about preparation for the panel. Stephen would prefer to show up cold and let things go where they will go. I prefer to talk to each panelist ahead of time and have an agenda of things we'll talk about and some that we won't because we don't have time. Stephen will make sure he then includes all of those topics (actually he doesn't but he did get a good laugh by pointing out when he was mentioning a topic that was on my list of things we weren't going to be able to cover eveb though they were completely appropriate given the title to the session). In all seriousness, I truly believe that a deeper conversation with Stephen ahead of the panel might have allowed me to aim things a little better and pull him in better and know where I could point out how what he was saying was very much in alignment from what you were hearing from Tony O'Driscoll from IBM and where he aligned with me.
But to add the last point to Derek's post:
6. Have fun and make it a panel session you and the panelists would have wanted to attend.
And I think we did that - didn't we Stephen? and Tony and Brent?