She posted this picture on some well-read forum and within minutes the readers resolved the actual location and resolved other questions about the identity of the individuals in the picture.
This shows me that certain small tasks with a desired result (identification, problem resolution, etc.) can become very easy as long as you have a large enough audience to process it. There was a large amount of noise generated by this audience, but inbetween some very interesting resolutions and suggestions.
There are probably certain limitations on the efficiency of such an effort:
- The end result must be clear and objective. This rules out open questions. (how to....) is not a good question to put in a processing task.
- It must be a small enough task that only requires knowledge and creativity to solve. Actual manual effort would be very difficult to organize and coordinate.
- It must be a problem that lies within the expertise and experience of the audience.
- The community should be able to share references to documented knowledge or images.
I can imagine that for the single person to develop a good resolution path to this problem it would take much longer than 50 hours and it might not be optimal. There is also no knowledge sharing (since the result is published otherwise).
Is it better if the result is developed and shared automatically by all people involved? Can the audience that is participating steer itself in the best direction for the resolution and auto-correct eventual mistakes?
I found some interesting sites on this topic that also touch on something else, human-computer interfacing on the brain-level:
Integrating computers with human brains
Now, on another note, already scientists have developed software and hardware that allows a paraplegic to move a cursor on a screen. There are two things to consider here. With bionics we can augment human functions and allow the human to improve, but vice-versa we can also create human interfaces that would complement a computer so that some automated process becomes much easier.
In a way, Flickr is something like this. It receives keywords from people browsing the photo's, so-called tags and then indexes photo's accordingly. The classification of things thus happens with people whereas the rest is simpler to automate by machines.
Maybe we should run some experiments by thinking of some not too difficult objectives to achieve and see how we perform in the resolution of that problem.