Dynamic Crowd Wisdom
The voting system is straightforward: users rate each submission on a 0-to-5 scale and offer comments that range from the constructive to the unprintable. Still, some submissions never make it to the voting stage, usually because they ignore format rules, raise copyright issues or, sometimes, are simply “awful.” (Kalmikoff says that eliminating ugly designs before a vote is an infrequent but sometimes necessary measure to “protect the experience” of Threadless voters.) While most winners have scores of 2.6 or higher, one recent batch included a design with a score of 2.0. That’s because the final decision about which T’s actually get made and sold has always involved a bit of nonpublic number crunching. For example, Threadless looks at how many 0s and 5s a design gets; designs that inspire passionate disagreement often get printed because they tend to sell, Kalmikoff says.
Seems anything but straightforward! But that’s okay, I think it reveals several truths around wisdom of crowd stuff. Neat how the decision process is iterative and cumulative, as the community gets smarter and tries to game the system, and as Threadless gets smarter and tries to right the system. This sort of evolution is completely unacceptable in politics, say, but seems to be innovative when done by Threadless.
I love the blunt naivete of putting forward X choices and having people pick, and then the sophisticated noodling that comes out later as the community grows in sophisticate. It’s not unlike the elaborate hierarchy of individuals, monitoring, and other checks-and-balances created by Wikipedia, outlined by the NYT magazine last week. A simple idea and simple implementation becomes arcane and complex by inches. Is this entropy? Human nature? Evolution? Line extensions?