This is an ambitious effort. Originally (years before HiRISE), Clickworkers was used to tag craters on Mars, helping pin down the relative ages of various regions. This time around, you identify a dozen or so possible feature types, then move on to the next image. So you have to be a little more discerning, though examples are provided.

I was just looking at the sizes of our images to date. We’re coming up on one thousand images that have been map projected. And it looks like we just recently passed the one million megapixel mark (one thousand gigapixels, or one terapixel!) in the geometrically projected ones (when rotated so that North is up, there tends to be a lot of empty pixels framing the images).

Assuming a standard screen size of 1.25 megapixels (1280×1024), that is 800,000 screenfuls. If you looked at one per second, it would take you almost ten days to view it all! But one thousand volunteers could get through it in a day, and spend 100 seconds per image, which seems reasonable. [Though of course they'll need time for sleep, etc!]

The idea of using human brain power as a sort of massively distributed computation engine (shades of The Matrix) has come a long way. Amazon’s Mechanical Turk pays volunteers for tasks such as identifying features, translating documents or answering questions. It was recently used in the search for a person (computer scientist Jim Gray) missing at sea. Volunteers viewed over a half million images, covering 3,500 square miles of ocean, though unfortunately his sailboat did not turn up.

Still, ‘crowdsourcing‘ (as Wired called it) seems like it will continue to be an efficient way to perform tasks that computers are currently very poor at. Here at the Lunar and Planetary Lab, it has also been used by Spacewatch to find Earth-approaching asteroids. So, essentially, you could help save the planet in a real-life version of the classic game Asteroids! Clickworkers also has a program where you can tag Mars Global Surveyor images, scouting interesting locations for HiRISE to target.

We can’t let the machines have all the fun!

This was the third test in a series of 4 tests. The purpose of this test was to begin ramping up the volume of products we released, to test our first JPEG2000 product, and to test the ability of the node software to distribute files that are larger than 2 Gigabytes. We are still learning how to make proper JPEG2000 files, but we did learn that the PDS node server software was successfully able to serve these large files. The ability of different clients to download such large files was a little more spotty, but this was not completely unexpected since there are many clients that are incapable of downloading files this large.

This test gave us a pretty good idea about how downloading such products might work by conventional means. However, our plans are to serve our JPEG2000 products using the jpip protocol which will allow clients that understand this protocol to pan and zoom around portions of our images without downloading the entire file. In most cases, you probably do not want to even try to download such large images in the first place.