I wasn’t sure whether to post this: it seemed kind of self-centred and not really relevant beyond my talk at ELC 2012. But I’ve wanted to address the subject for a while. This is somewhat incomplete, but it’s all I’ve had time to think about it.

I’m a bad public speaker. I either over-prepare or I under-prepare, and I tend to forget a lot of my material in the heat of the moment no matter what. In the most recent case (which has been a common scenario), I compulsively over-prepared, fearing being unready to give my talk. This ended up being so bad that what happened was this: I had some notes on paper, but because I had a lot of my talk in memory, I was able to skip most of them. But I missed a lot of the nuanced points I’d written down.

Then I’d hit a point where I couldn’t remember what I was supposed to say next. So I looked at my notes, but I’d gone fast from memory and had to flip several pages to see where I was up to. That caused a break in my talk. Then, later, I realised I’d gone out of order and missed almost an entire section out, so I had to go back. The result was that my talk, instead of being a neatly informative upside-down pyramid, was more of a confusing upside-down cello.

I’m going to attempt to right this particular wrong by posting an essay version of my talk (which was actually how I wrote it in the first place) on this site in the next couple of days. I’m sure when the video of the talk goes up I’ll realise it wasn’t that bad.

(The other problem with my talk was that I didn’t realise how disagreeable some of the things I was saying were to my audience. I get the impression that talking about a new “practical Lisp” to a crowd that turned out to be full of Clojure fans was not such a great idea. Doubly so when you’re not clear about what differentiates your language from Clojure.)

In the general case, what could I do to improve this? I feel actually that I wish I didn’t have to have slides, and that they directed the flow of my talk in a very un-natural way. But I had to show off code examples somehow, and with no working implementation that accurately reflects the current state of the language as I envision it, live coding was not an option.

Slides are out in future, I’ve decided, as are notes with pages. I’m going to try using a teleprompter — not to read from directly, but so I can have my notes always roughly keeping pace with where I am in my talk, without having to do anything to advance them. (Ideal app: something that uses speech recognition to work out where I’m up to by looking for key words in my speech and my notes, and advances automatically.)

Coming later this week:

• A report on Strange Loop.
• The aforementioned essay version of my talk.