For a business shift that’s about inviting in non-experts, hobbyists and hackers, there’s a lot of insider lingo around crowdsourcing. This begins the first of a series of straight talk on crowdsourcing principles to help us all put the theory into practice.

What is collective intelligence? Jeff Howe, the guy that came up with the term crowdsourcing, says it this way, “A central principle animating crowdsourcing is that the group contains more knowledge than individuals.” James Suroweicki says, “Even if most of the people within a group are not especially well-informed or rational, it can still reach a collectively wise decision.” This is the science that explains why when asked for a lifeline on Who Wants to Be a Millionaire, the crowd guesses 91% correctly, whereas experts have a 61% likelihood of getting the right answer. The answers that come from crowdsourcing are called collective intelligence or wisdom of crowds. Yes, two terms for the same thing.

So what does it take to achieve collective intelligence? Will any group of people do? Crowdsourcing has three unique requirements to deliver collective intelligence – (1) a diverse crowd, (2) a qualified crowd, and (3) the right sized crowd.

Crowd Must Be Diverse
Crowdsourcing requires a mix of people who aren’t too much alike.
If you had the undivided attention of every genius on any given topic, you would get very accurate answers to questions. The problem is, as Bill Joy Cofounder Sun Microsystems says “No matter who you are, most of the smartest people work for someone else.” So is there another way to get to genius results? Contrary to popular belief, a group of the most highly accredited or accomplished brains on a topic do not produce the best answers. Crowdsourcing shows that breakthrough solutions more reliably come from a diverse enough set of minds to go at the question from different angles without the bias inherent in like-thinking people. American Idol wouldn’t work to accurately predict best-selling pop stars if everyone voting had identical musical tastes. Goldcorp would not have found a total of 8 million ounces of gold if they hadn’t invited not just geologists but grad students, consultants, mathematicians, and military officers into its Goldcorp Challenge.

Crowd Must Be Qualified
Crowdsourcing takes people who are interested and capable.
Qualifying a crowd is about finding people with an interest in the subject with the technical skill to provide the answer or perform the task. For answers on how to better operate your business, the best crowd is likely your employees and partners. They know the company and are motivated to make it work better. For a research breakthrough on a diabetes drug, it would take a crowd with competence and passion in medicine – but don’t just think doctors or others in lab coats. As InnoCentive has proven, solutions to scientific challenges may come from students, retired members of a field, or in the case of a drug discovery, why not a technically astute patient with experience on what drugs have worked? Procter & Gamble has a publicly-available connect + develop crowdsourcing site to seek new product insight as the company recognizes that everyone is a consumers or potential consumers of its millions of products.

Crowd Must Be Right Size
The simpler the crowdsourcing question the larger the crowd needed.
Crowdsourcing involves many models where the simple majority does not rule, instead different members vote with different weight or perhaps expert moderators have influence. To keep it simple though, let’s talk about crowdsourcing where the majority will pick the winning solution. How many people does it take for the majority answer to be right? The answer depends on the simplicity of the question: Who should be the world’s next mega pop star? That’s a question every pop music listener is qualified to answer. More than 100 million crowd members cast votes via telephone and text-message to determine the last American Idol (of course the controversy ensues about vote rigging). What is the next cool software worthy of Intel’s chips inside and their investment? That’s a more specialized question, with fewer qualified to answer. Profitable insight there took a crowd of Intel’s 80,000 employees. The jury is still out on the minimum size of a crowd for collective intelligence. Jeff Howe mentions 5000. Some say it could be fewer. Like other crowdsourcing pioneers, we continue to test crowd size and other principles to hone the reliability of crowdsourcing results.

What have your experiences shown about the principles of crowdsourcing? We welcome your comments.

photo by Sreejith K