May 2007

Okay, class. Here’s a brief review. The original web was a place where you went to get information. Then Web 2.0 came along. You were no longer just a web surfer but you were also now contributing vast amounts of content to web sites such as YouTube (videos), Flickr (photos), and MySpace (social networking), helping to create storehouses of organized content enjoyed by everyone.

That was then. And now the next big thing: crowdsourcing.

The idea is simple. Think of the Internet as a vast web of interconnected intelligence — millions of people sitting at millions of computers. Why not marshal that intelligence in a coordinated fashion to accomplish specific ends? It’s sometimes referred to as “hive intelligence”: a decentralized, multicomponent intelligence like that of ants and some species of bees.

Although the definition is still emerging, it seems that YouTube may not be an example of crowdsourcing because the end is so sort of amorphous: a data bank of user-created videos organized in various ways.

Instead, crowdsourcing could be defined as tapping into the collective intelligence to accomplish a specific task.

Wikipedia, the encyclopedia that anyone can edit, is a good example of crowdsourcing: Internet users worldwide create and modify articles on specific topics. Another example is open source software. Programmers around the world freely contribute to creating software used by everyone such as the Firefox Internet browser or the Linux operating system.

But in the past year, crowdsourcing has taken a new twist. Unlike Wikipedia and open-source software, from which no one profits, crowdsourcing is increasingly being used by businesses.

One major area is journalism. For any given news story, it seems like the collective intelligence would have a greater range of relevant information than any single journalist could feasibly gather by himself. Why not use that? Why not tap into the natural tendency of people to share tidbits that they know?

In March Wired magazine announced Assignment Zero, an experiment to see if widely scattered people, working together voluntarily on the net and reporting on something happening in their world right now will be able to “tell the story more completely, while hitting high standards in truth, accuracy, and free expression.”

The Gannett newspaper chain has already jumped into crowdsourcing, creating mechanisms whereby people can contribute to emerging news stories. Also, Topix, which is perhaps the Internet’s largest aggregator of news, announced in April that readers will be able to submit local news stories.

And herein lies a bit of controversy about crowdsourcing: Topix will be reaping revenue from sales of ads that appear on the pages where the user-contributed stories appear. In other words, businesses are outsourcing to some poor suckers who are doing the work for free.

In other cases of crowdsourcing, the model works differently: the bees get a small amount of money. A good example is iStockphoto. It’s sort of like Flickr, except that people who are posting their photos aren’t just posting snapshots but are uploading photos that are good enough quality to be used in publications. And publications that use these photos pay the photographer a small amount of money — from $1–5 per basic image.

While that might seem like a pittance, many of the photographers already have these photos on their hard drive anyway. Why not upload them and make a few bucks per image? Of course, publishers love this. If they went to a professional photographer for photos, they’d likely have to pay $100–300 per photo.

So far we have two types of crowdsourcing: many hands freely contributing to a specific task, and many hands making content available that’s used for specific purposes by business, in this case publishers.

A third model entails posting problems so that many people can take a crack at solving it. And if they do, they’re amply rewarded. A good example is InnoCentive, where companies such as product manufacturers pay $10,000–100,000 for solutions. Often hobbyists are able to figure out a solution to a problem that has stumped in-house industrial scientists.

And a fourth model is Amazon’s Mechanical Turk, where companies can post tasks that require human intelligence and pay people small amounts of money, from a few cents to a few dollars, to do things such as transcribing podcasts, tagging specific items in images, and reviewing data extracted from contracts — tasks a computer is unable to do.

Crowdsourcing may be just the beginning of a sort of emerging global intelligence. And you’re part of it.