flight patterns

When I started at SGI, the company was a lot more than over-priced graphics workstations. There was a tonne of work in data visualisation; taking huge quantities of data and presenting them in a visual format that’s easier to understand than just the raw numbers.

One of the most compelling demos we had was a time-lapse visualisation that showed call traffic following the San Francisco earthquake. It displayed point-to-point call volumes by super-imposing bars of colour between major exchanges. In the hours prior to the quake you see the odd bar (which required 1k calls to show up, and different colours as the number of calls between the exhanges incremented) between points, and then as the earthquake hits you see how the news travelling affects the phone system.

As time passed, the calls between different points became a non factor, and all the exchanges started to show volume to the SFO area. Call volume continued to increase as the news got out and people started calling anyone they knew in the SF area to make sure they were ok. Eventually SF looks like a red (the colour used to display the highest number of calls in a trunk) sun with rays reaching across the United States as the system nears collapse. Pac Bell then switched the network over to priority calls only, at which point the map went dark. Really cool stuff, and a very effective demonstration of how powerful visualisation can be.

I ran across the Flight Patterns work this morning that was used as a stepping stone to the Celestial Mechanics project. The Flight Patterns work is visualisation of FAA data over the United States, and shows flight paths using vectors drawn from repeater data from the FAA. My fave is probably the visualisation of the flights by aircraft type.

Cool stuff, and this was the kind of work that was being done by the folks at SGI over a decade ago. So much potential, but boy was it wasted.

Check out Flight Patterns, it’s pretty cool, and reminds me a lot of the talent that was once at Silicon Graphics.