I was reflecting again recently on an obscure example I use in presentations that demonstrates a useful point. A few years back, people started thinking about online collaboration projects and experimenting with what could be created by a group of random, unaffiliated people charged with completing a small task. "Mob art" was one term that developed around this. One such project was the Typophile Smaller Image Generation Experiment, which attempted to create a collaborative font. Users who went to a site were asked to mark a grid to make each letter. The grid markings were averaged - in theory this would result in what the collective intelligence thought each letter should look like.

The results were good. It is pretty clear that all letters look like they should.



Another experiment took off from this point and asked users to create simple pictures of known things. For example, users were asked to create a goat. Sample results are shown below, representing the average of roughly 2000, 5000, and 7000 entries:



The picture starts off looking pretty ragged, but then, by the time 5000 people have had at it, it looks like a goat. By the time 7000 people have worked on the picture, however, it starts to deteriorate.

Interesting mind-candy, but what does this tell us anything useful? It turns out that there is a message. A letter is a very standard thing - across all common, non-decorative fonts an "A" is an "A" is an "A," plus or minus a serif. Everyone understands what an "A" looks like. A goat is harder - people generally don't see goats very often, and when they do, might see different kinds in different poses. There is much less guidance, from common experience, as to what a goat is. Some of the deterioration could be decay in interest, but some of it is probably that more tries does not necessarily bring more clarity because the subject is less well defined.

So here is where things get more down to earth. In many cases organizations are interested in getting a group of users - members of a key audience - to do something: contact a policymaker, internalize knowledge, give money, an so forth. Too often organizations expect users to make the connection between what they are providing and what they want done. It is important to provide direction - explicitly and implicitly - for users to follow. This message is also borne out in the success of MoveOn and the like - they provide a clear action and path for users to take it. Putting even highly valuable information online and not directing users on how to use it is like asking them to [drum-roll, please, for big punchline] draw the same goat.

This is even more pronounced for collaborative projects. Educating and energizing audiences through collaborative projects is an emerging tool for bringing about desired changes in their behavior. These projects can be successful - look at Wikipedia - but they require clear guidance. They need a well understood and very specific mission, prominent operational rules and code of conduct, and a highly defined end-product (or set of end-products). Setting up a project without this structure is a recipe for failure. If any activity does happen - and without the structure it may not even get this far - it will likely be confused or even contrary to the original goals. Take the time to provide guidance for online projects - it is an up front cost but will pay off in much higher likelihood of success.