Why the hyperbole, Judy? It's the co-incidence of two interesting items which caught my attention recently. One was an article about computational thinking by Jeannette Wing, who is head of computer science at Carnegie Mellon. The other was a news story about a search for the missing computer scientist Jim Gray.
Wing's article is a manifesto which promotes the teaching of the sorts of analytical thinking used by us computer scientists to everyone, as part of a basic education. As in: reading, 'riting, 'rithmetic and computational thinking. The general idea is that in learning how to solve computer science problems, we have developed a set of techniques for approaching problems which are useful in everyday life, and when applied outside traditional computer sciency problems. I was finding it hard to find examples of what this might actually mean, when my husband (also a computer scientist) told me about the search for Jim Gray.
Jim Gray is an ACM award winning engineer who works at Microsoft. He went missing at sea on January 28th, and although the coastguard called after the search after 5 days, since then his friends and colleagues (in organisations including Amazon, Google, Microsoft and NASA) have been hunting for him in a massive digital search effort. A blog entry on All Things Distributed comments somewhat cryptically that "Through a major effort by many people we were able to have the Digital Globe satellite make a run over the area on Thursday morning and have the data made available publicly." Once the satellite imagery was available, it was split into lots of tiny sections and placed on Amazon's Mechanical Turk web site, where thousands of volunteers each looked at individual tiny sections to spot anything unusual. Sections which were flagged by volunteers were then further analysed by experts. The Mech Turk page for the search says: "Mechanical Turk workers looked at more than 560,000 images from 3 satellites, covering nearly 3,500 square miles of ocean".
Now there are many amazing things about this story, including how on earth the search effort managed to get the relevant satellite data, and the moving goodwill of all those people who joined in the search, but what interests me most is the strategy used to solve this problem. It's a "divide and conquer" approach: breaking a seemingly intractable problem into bite sized chunks which can be easily solved with simple methods. In this case, the simpe method involved good old fashioned human eye sight. To avoid the computational worry of false positives (seeing something when nothing is there) they had experts check the small number of suspected sightings afterards. I'm not sure whether they had checks and balances to make sure there were no false negatives (seeing nothing when there really is something) but I bet you they had some system for it. This is a rapid application of well known computer science techniques to a solving a tangible human problem; a use of just-in-time computational thinking to help other people. I hope it is successful.
Labels: computational thinking