Collective Counting: The Results
In anticipation of the opening weekend of college football, we sent out a quick online experiment. We wanted to know “What is the total number of skittles in this jar?”
Theory & Explanation
The hypothesis was that individually the answers would vary and most definitely be incorrect, but that the sum average of the answers would produce a close to accurate count.
On a Radiolab* podcast episode way back in season 1, they shared a story of how in 1908 a crowd of 800 county fair-goers individually guessed the weight of an ox. When the average of all their guesses were calculated, the average was within .08% of accuracy. This was in 1908! More recently, we learned that the ox weight counting story comes from “The Wisdom of Crowds” by James Suroweiki. Since he wrote this book in 2005, our social online connection has increased exponentially. According to Social Engine Watch, online social media involvement has increased 64% between 2005 and 2013.
This got us thinking…what does this mean for architecture and design? We want to see if we can work with collective wisdom to shape our future buildings, streets, towns and cities. User input in architecture is not a new idea. It happens with town forums early on in projects. It occurs during programming when you want to learn how each person uses the space. Those examples, however, are much more about collecting opinions and individual ideas rather than looking at a large amount of ‘data’ for decision making. We are interested in seeing if the general population has a collective sense about making our cities a better place.
Our hope was that this experiment would draw the same conclusions as the countless experiments, puzzles and questions that others have done in pursuit of collective wisdom. Here’s what we found:
Need More Guesses
Our data needed more guesses. We were very excited to have a variety of participants in our experiment, however, a larger field (i.e.: more guesses) would have continued to alter the results and we believe would have seen that average get closer to the actual number.
The Answers Support the Theory
Of the 46 guesses, only 1 was within 10 pieces of the actual amount. Most were really incorrect, but when the average is taken-it does get that number closer. This speaks to the hypothesis completely-individually, we may not know what’s right or what’s best, but collectively, we are able to steer in the right direction.
This experiment was very much black and white (or rather, orange and purple). There is a correct and singular number of skittles in the jar. What happens when crowd wisdom is used to determine something less numeric; less definitive. With design considerations and program selections, there can be many approaches to a solution. We hope to continue pursuing this idea of using social media as a means to communicate with a large amount of people and translate their feedback into data that can truly influence and better our cities.
Radiolab-we strongly recommend you do. It makes any car trip more fun!