Most of the day was taken up in preparation for the evening presentation on Flex with James Ward. Our goal was to create an anti-bullet-point experience because people have already had a couple of days of bullet points by then.
We used two screens, and I roughly represented the right hemisphere of the brain -- more pictures and big concepts -- and James roughly represented the left, with specific examples and details. We interacted and changed speakers frequently, and the whole effect was to provide lots of stimulation. Plus we managed to keep the length down to 25 minutes, a new (short) record for me, so it was kind of like an extended lightning talk. Since I tend to fall asleep in presentations, we tried our best to create something awakening. I felt successful when one of the AV guys, who had no programming background, told us afterwards that they had found the presentation stimulating.
The party afterwards was also a success, in this case because it was held in a space filled with pool tables. Most conference parties I've seen have either been a room with food and drink, or some kind of passive entertainment where the partygoers are watching a band or something. Kind of an eyes-forward party. It's obviously good to have food and drink (and our party had that; quite good food, really), but people have been doing eyes-forward all day already. It's much better if people have something they can actually do (which is the common theme of the events I put on in Crested Butte). Everyone seemed to have a good time playing pool, and the conversations formed around the activities. The Java Posse and some of the Roundupees came and stayed until the Google party.
A lot more people came to the party than to the presentation, and it's occurred to me that if we had videoed the presentation we could have been replaying it in a corner of the party for those interested.
I spent a good portion of the Google party talking to Patrick Chanezon, the Program Manager for Google's Checkout API. He had emailed me in response to my blog on the need to make other aspects of ecommerce as easy as creating a buy-now button. Although I'm averse to RSS on moral-technical grounds (it's publish-subscribe, darn it, and the only reason we don't want to do it that way is that the identity people have been dragging their feet for years), he convinced me that slow polling would be the easiest way to simplify the problem of discovering new purchases and extracting the information. If that were in place, then creating a fully-functional purchasing system becomes only slightly more work that just putting in a buy-now button. I could make use of it right now.
Earlier in the day I went to the SunSpot booth and ran into an old friend (at least, we both agreed that the other one looked older) Bruce Boyes, who's been running Systronix since before the days of Midnight Engineering magazine. My first book was about using higher-level languages to control devices, so the SunSpot has been interesting to me from the beginning, and Bruce had just finished a marathon circuit-board fabricating process so that the SunSpot could be attached to a small set of tank treads and you could program it to drive around, sensing edges with infrared LEDs and sensors.