/* */

09 September 2007

Brandy with the Mechanical Turk

We have long been staunch advocates of commercial overhead imagery systems, and from the earliest days of their uncertain launch and performance we would vocally proclaim their potential to any and all – most of whom who would darkly mutter about NIIRS, classification, and mission availability.

However, commercial IMINT has so far exceeded any hopes we could have dreamed in the early days of the 90’s, having been brought to widespread use by Keyhole / Google on a scale previously unimaginable. The resulting sophistication of new processing techniques – from humble mash-ups to the surprising benefits of the open source model of many eyes – has given us great reason to reconsider the infamous “bottlenecks” in the architecture of other national technical means. Perhaps it is not the nature of the problem that has so bedeviled those of us seeking to expand the percentages of raw take examined for interpretation and analysis, but rather the model of the solution which is defined so very much by implicit and unquestioned assumptions.

The strange case of a missing billionaire “adventurer” has given us further evidence that other production models may well exist. High resolution imagery was acquired by a private donor’s tasking of the DigitalGlobe constellation, and it has been populated into Amazon’s innovative Mechanical Turk system – a fascinating technology designed to facilitate the development of architectures of participation to effective harness ad-hoc volunteer efforts. The system’s name, of course, derives from the famous chess playing automaton of the 18th century Austria-Hungarian court.

We wish the searchers bonne chance, for we know well the damnable difficulties of wide area search problems. We would also hope that any of our readers in the professional geospatial community might consider lending their unique talents and expertise to the search. (Those wishing to do so may contribute at the Steve Fossett Missing Mechanical Turk page.) We hope that a suitable flask of good brandy – in the model of the Saint Bernard rescue dogs – is close at hand on the day of the searcher’s success.

The effort is a dramatic example of innovation, however, that will have utility to the GEOINT community long beyond this crisis. This may well be the esteemed Mr. Fossett’s greatest legacy to the intelligence community – although we do hope he returns safely to provide many more candidates for nomination. We could easily see a trial program initiated for similar analysis of open source imagery and geospatial extraction, through an albeit less public architecture. USGIF certified schools and other intelligence studies programs would be a natural first home to such an experimental platform. This may also bring IMINT and geospatial analysis instruction into the mainstream of the intelligence studies curriculi, despite having long languished in the backwater of a narrowly perceived technical corner.

h/t Slashdot

Labels: , , , , , ,