As someone who matriculated in the year Nelson Mandela was set free and the ANC unbanned, questioning text books just happens naturally. I can’t help but to have more faith in collaborative data from a wide range of people, as you would find on Wikipedia, than in any article by only one author. But reading James Surowiecki’s The Wisdom of Crowds this week convinced me that there are at least four prerequisites for groups (including the MoreThanMoney community) to function intelligently.

Groups work well when there’s enough:

1. Diversity: The more diversity in a group, the greater the number of opinions and potential solutions that the group can produce.

2. Independence: Independent members are more likely to add new information and can prevent mistakes from becoming correlated. It takes an independent, highly confident and outspoken individual to break the spiralling destruction of an information cascade, also known as ‘hype’ or ‘groupthink’.

3. Decentralization: This organisational structure allows members to specialise their labour, interests and attention, which in turn helps the group to benefit from local, detailed knowledge.

4. Aggregation: A system is necessary to feed all the information to a place where it’s visible to the entire group (and preferably filtered through debate and a ranking mechanism).

Google is a prime example of how we, the crowd, help you find the exact information you were looking for within seconds. Google uses the democratic link structure of the web to decide how important any particular page is. A link from page A to page B counts as a vote by page A for page B. Google not only takes into account the number of votes that a pages receives, but also the importance of the page that casts the vote – weighted aggregation.

www.experts-exchange.com is another group attempt at problem-solving. A dazzling number of insights are aggregated every day, with members constantly filtering the good ideas from the bad. As a result, the group has come up with solutions to more than two million technical problems.

Groups fail to come up with intelligent answers when:

1. Participation is skewed: The Presidential Aids Advisory Panel of 2000 is a local example of how lopsided participation can lower group intelligence. President Thabo Mbeki requested a secure internet facility reserved exclusively for debate around the HIV/Aids causality. But the site mainly attracted the so-called ‘Aids dissidents’. Very few mainstream medical specialists entered the debate and posted their findings and supporting evidence. At the end of the two-month collaboration, it was evident that the panel was unable to provide a balanced view.

2. Members guess what the rest of the group thinks: For example, stock market prices are not only determined by what members think a company is truly worth, but also by what they think the rest of the market thinks it’s worth (with re-selling in mind). When members anticipate that other members will be willing to pay almost anything for an e-commerce company, for example, an information cascade or stock market bubble starts forming.

3. Too powerful or less-informed members speak first: The order in which information is received is particularly relevant in the case of forums. The first contributors generally create the structure of the discussion. When contributions are received in quick succession, often without members having time to read the comments, there is normally a larger pool of different solutions.

4. Members lose their identity within the group: Group psychosis can lead to terrible decision-making, as at soccer riots, for example.

5. Members are not properly incentivised: Real or play money or just boosting their esteem has been found powerful enough to focus people’s efforts to provide more accurate answers. A prominent journalist was once connected to the Kennedy assassinations in his Wikipedia profile after a prank entry. Maybe, if the prankster had more reason to protect his own reputation online, he would have provided more accurate information.

6. The outcome is not considered that important: With fashion trends, little independent thought goes into deciding whether those skyscraper heels or skinny jeans are really wise – it’s a frivolous decision. As a result, the outcome of the crowd’s decision-making process usually isn’t very smart.

Why we still desperately need experts

James Shanteau recently investigated between-expert agreement in a range of fields, including investment management and clinical psychology. He found the agreement to be less than 50% for most fields. Not only did the experts whom he studied struggle to agree on the facts, but they also found it difficult to calibrate their judgements, i.e. remain conscious of how certain they are of their findings. No expert is without her blind spot.

Still, looking at Surowiecki’s four requirements, the collective intelligence of the crowd depends greatly on independent experts with diverse opinions. It is enriched by every member’s specialised knowledge. Now the many perspectives of online aggregation tools can only help to illuminate those blind spots.