Social networks and the wisdom of crowds

"Wisdom is not additive; its maximum is that of the wisest man in a given group."

There's a lot of buzz lately about the concept of social networking. You've got to admit the phrase is pretty silly. First off, it's redundant: Networks, by definition, are already social -- they connect humans. Plus, what's the opposite -- "antisocial networking"? Going online to tell everyone how much you hate them?

But despite the silly name, the basic concept is pretty exciting. It refers to the fact that networking technology is leading to all kinds of new and interesting ways for humans to interact. One of those is crowdsourcing -- the notion that people in the aggregate can provide more accurate information than individual experts.

The reasoning is that crowds can be self-correcting. If a large number of people are able to correct one another's errors -- whether made out of ignorance or bias -- the results will be overall more reliable than the output of any individual (or small group). The canonical example is Wikipedia, which is by pretty much any measure at least as accurate as a traditional encyclopedia, and considerably more timely. Another example is the reader reviews on Amazon.com.

So far, so good. I'm a huge fan of Wikipedia (as well as Amazon.com reviews). But we need to be aware of the limits of the wisdom of crowds. I'm reminded of the wonderful Robert Heinlein quote: "Democracy is based on the assumption that a million men are wiser than one man. How's that again? ... Autocracy is based on the assumption that one man is wiser than a million men. Let's play that over again, too."

Replace "democracy" with "crowdsourcing" and "autocracy" with "individual expertise", and you see the problem precisely. OK, I'm just a tad biased. My livelihood depends on my perceived expertise (however real or not), so I'm naturally a bit reluctant to imagine that I could be replaced by an anonymous crowd.

But there's a bigger reason I'm skeptical of crowdsourcing. It's another social networking phenomenon called information cascading. Cascading refers to the demonstrated fact people often change their opinions based on those of others -- without having any better data.

Researchers Duncan Watts, Matthew Salganik and Peter Dodds demonstrated this phenomenon in a study published last year in the journal Science. More than 14,000 participants registered at the Web site Music Lab and were asked to listen to, rate and, if they chose, download songs by bands they had never heard of. Some participants saw only the names of bands and songs; others also saw how many times the songs had been previously downloaded.

The upshot? Bands that had been rated highly by previous participants were more likely to be rated highly by subsequent ones. As the researchers reported, "The impact of a listener's own reactions is easily overwhelmed by his or her reactions to others."

Crowds, in other words, aren't quite as self-correcting as we'd like to believe. Once again, Heinlein put it well: "Wisdom is not additive; its maximum is that of the wisest man in a given group."

Join the newsletter!

Error: Please check your email address.

More about Amazon.comBIASHISPLUSWikipedia

Show Comments