Collective outsmarting
Using collective intelligence to build better products
What do Google Flu Trends, Waze and delicious.com have in common? They all tap into passive collective intelligence to outsmart their competitors.
We call passive collective intelligence a distributed knowledge that emerges as by-product of our daily lives. Whether we are navigating the web, having conversations on social medias or buying online we are leaving behind us a lot of data. Taken individually, this data does not mean much, but put together, it exhibits trends and behavior that can help building more relevant, closer to real-time products at a lesser cost.
Google Flu Trends provides estimation of influenza activity by aggregating a lot of health related search queries. Because data is collected automatically and rely on an empirical formula rather than survey and expert analysis, Google Flu Trends can predict the flu 10 days before they are reported by the Centers for Disease Control and Prevention !
Waze is a GPS-based geographical navigation app that relies on its users to map road-traffic. As opposed to other community-driven products like wikipedia, the application does not require users to be fully active. Just open the app while driving and Waze will learn by itself where the traffic jams are by analyzing car speed data. This passiveness is what makes Waze both scalable and very cost effective.
Delicious (formerly del.icio.us) is a social bookmarking web service that allows users to tag web pages. The process of co-tagging, i.e the combined view of all users bookmarks is highly valuable. First the coverage in term of number of pages tagged is tremendous because it relies on the concurrent action of a lot of distributed actors. Second, it is very accurate and relevant because the final tags are the result of a consensus between those actors. No expert can achieve that.
This is collective outsmarting.
Having those examples in mind, I decided to build my own collective intelligence app. Being a big fan of Twitter, I have noticed that people are sharing a lot of quotations from famous authors and I thought it would be a good idea to tap into this data source to build one of the best database of quotations. I called it twisdom.li (a portmanteau word from Twitter and wisdom). It extracts most of the quotations shared on Twitter by monitoring the Twitter stream through their API. In less than a week I have already collected 30,000 unique quotations in more than 10 languages and it’s adding up. You may say it’s just another quotation website and you will be partially right. The only difference is that I have spent almost zero effort building the database and that I can infer the popularity of a particular quotation by looking at the number of times a quotation was shared. If you are looking for a good quote to share, you’d better pick one from the top 50 of twisdom.li rather than using a database made by experts.
Is it the end of expertise?
In a sense, yes. Automatic data collection and algorithms are making experts analysis more and more obsolete. We started by delegating memory to the machines and now we are also delegating some analytical skills. Nevertheless, we always need to prove ourselves discreet in front of automatically generated knowledge because it is subject to bias and abuse. Indeed Google Flu Trends is said to have drastically overestimated peak flu levels and one of the most popular quotations in my database is:
“The problem with quotes on the internet is that it is hard to verify their authenticity” — Abraham Lincoln
