direwolff's dashboard

I think there is a learning aspect to collective intelligence that collaborative filtering as a term misses out on. Programming Collective Intelligence (the penguin book) is one of the best books O'Reilly has published in years.

I think Tim's right on. Collaborative Filtering is specific type of collective intelligence which relies heavily on statistical techniques for collecting the, um, intelligence.

How the ideas or information are aggregated is what makes amazon different from wikipedia and others.

It seems to me that collaborative filtering is one of many examples of internet-enabled collective intelligence. I use the term collective intelligence as a broad umbrella to cover all kinds of collaborative activity in which "all of us are smarter than any of us." For example, the swarm editing of wikipedia is an example of a collective intelligence activity sometimes referred to as crowdsourcing. And that is very different from collaborative filtering, a statistical technique that uses similarities between customers to make recommendations. (Amazon's "people who bought x also bought y" is a good example of collaborative filtering.) Google's PageRank is another example of collective intelligence, as for that matter is Google's keyword auction. Digg's voting algorithms that propel news stories onto the front page, Techmeme's or Google's crawl to do the same thing, eBay's reputation system, Facebook's emerging social graph -- these are all aspects of collective intelligence.

Just look at the overall span of the phenomenon we're calling Web 2.0, and you'll realize that every success story is harnessing collective intelligence in one way or another. There are many successful techniques.

Many of the techniques are statistical or algorithmic in nature--Toby Segaran's book Programming Collective Intelligence covers many of these--but some are merely architectural. For example, it seems to me that Flickr's "default option is public" was a collective intelligence breakthrough, as was their idea that you can follow someone's photostream without asking for confirmation. (Other social networks are starting to follow that insight.) Xobni's mining of the implicit social network in email is another great example.

FWIW, O'Reilly once owned some of the seminal patents on collaborative filtering, which were initiated by a guy named John Hey as part of a video kiosk application that we took onto the net in the mid-90s as part of a site called moviecritic.com. Moviecritic was spun out into a company called Likeminds, which was bought by Andromedia, which was in turn bought by Macromedia. When Macromedia shut the product line down, I believe the patents were sold to IBM.)

What you're doing here with Satisfaction is also a kind of collective intelligence.

You should ask Tim Wolters, CTO/co-founder of CollectiveIntellect...bet ya he knows ;) Also check out: http://en.wikipedia.org/wiki/Collecti... and http://en.wikipedia.org/wiki/Collabor..., which should provide some good added fodder for the discussion.

Collaborative filtering is a specific form of collective intelligence. It deals with attempting to make predictions about one person's tastes based on the tastes' of other and the connections between the people.

"Collective Intelligence" is obviously a broader term that could probably include collaborative filtering.