This morning I got an email message from a student asking if I wanted to go windsurfing today. Jo and I have to drive up to Brunswick, Maine to visit Jo's mother today, but it didn't look like a great day for windsurfing in any case, so I pointed this out and asked for a raincheck. It used to be much harder to check on the weather conditions at your favorite surfing locations but now you can get detailed weather updates every few minutes customized to your location from places like The Weather Channel and the Narragansett Bay site provides a nice graph showing the direction and speed of the wind for half a dozen sites around Narragansett Bay update several times an hour. The iWindsurf web site provides even more comprehensive forecasts based on both government data and their own network of real-time wind sensors. They have webcams set up at a lot of the best windsurfing venues, including some of my favorites, Kanaha on Maui (Hawaii) and The Dalles on the Columbia River Gorge (Oregon). I also found a webcam showing San Francisco Airport (SFO) and a nearby popular launch site for windsurfers. Many of my memories of surfing next to SFO involve jet fuel which was abundant in the area. You could smell it in the air, taste it in the water and feel it on your skin; the neoprene in wetsuits quickly disintegrated with continued exposure to the water and equipment used in the waters off SFO had a markedly shorter half life than any other location I've ever surfed in.

There's a mini weather station on our roof that has sensors designed to measure wind speed and wind direction. The sensors are wired up to a couple of analog meters in the kitchen so we can check the wind while sitting at the kitchen table. I've been considering wiring up the weather station to provide real-time information to one of the on-line windsurfing web sites. I could probably take output of the wind-speed and wind-direction sensors and design a circuit with a serial interface so I could read the sensors from my computer. In terms of the time required, however, it would probably be easier for me to mount a wireless webcam on the chimney pointing at the weather station and then analyze the images produced by the webcam to figure out the wind speed and direction.

What got me thinking along these lines was not windsurfing but sensors and how we are wired not just to one another through the web but to the planet and beyond. NASA's Visible Earth and Terra, the latest Earth Observing System (EOS) satellite, web sites offer a huge amount of timely data about the planet. And, while some of the more accessible sites like HubbleSite provide more of a public-outreach orientation, with a little more work you can get to more data and imagery from earth- and space-based telescopes than you'll know what to do with.

We're still some years away from building robots that can interact with the messy world of trees and grass and doors and stairways as well as humans manage this. Some of what's holding us back has to do with the need for better sensors, better effectors for walking and grasping, and smaller, lighter power sources so robots can run more powerful computers and operate for longer periods of time. That's not the only problem of course; we also need better algorithms and indeed algorithms for doing things we're not even aware we need to do yet. But significantly better hardware would accelerate progress in robotics dramatically.

The picture is somewhat different when you consider the notion of a planetary intelligence that could monitor the entire planet. Forget about the "intelligent" part - that word is too loaded to be of much use - just think about an organism that is aware of and responsive to its environment. We already have an extensive network of sensors and computers cranking away, generating summaries of the data produced by these sensors. Under suitable circumstances such organisms could come into being and evolve on their own in the form of viruses. A fantasy right now given the protections in place to prevent unauthorized computations on networked computers but who knows what loopholes hackers will fine and exploit or what economic incentives might arise to tempt people to sell or barter unused cycles on idle machines.

I better get off this subject before I head off on yet another warped tangent. I get excited when I see the following combination of elements: an interesting dynamic environment (what could be more interesting than the planet we live), an efficient method of monitoring that environment at a sufficiently high resolution (there are terabytes of data produced every couple of days by earth-based weather sensors and webcams and satellite-based sensors tuned to the infrared, ultraviolet, x-ray, and visible-light ranges - a kilobyte is roughly (2¹⁰=1024) a thousand bytes, a megabyte is roughly (1024) a thousand kilobytes, a gigabyte is roughly (1024) a thousand megabytes, and a terabyte is roughly (1024) a thousand gigabytes), and lots of idle computational power waiting to be tapped (check out SETI@home for a description of a project aimed at harnessing the power of millions of computers on the web to help in the Search for Extra Terrestrial Intelligence (SETI)).

Perhaps it seems strange that I believe a rudimentary planetary intelligence (ok, primitive organism) will emerge before we have robotic presence anywhere near as interesting as the current crop of biological organisms. Of course, we may not recognize the planetary intelligence because we won't know what to look for, but we're quick to disparage the intelligence of a robot that can't negotiate a set of stairs or regularly stumbles into walls or people. People possess a bodily awareness and common-sense physical intelligence that current robots aren't even close to emulating but check out Cog and Kismet for some interesting experiments in this direction.

I need to get ready for our trip to Maine. It's a four-hour drive so perhaps I'll take the laptop along and try to add to this entry on the drive up.