Blog

Another interesting ReadWriteWeb article, this time on how life is shaking out in the social bookmarking world.  The article has a number of interesting types of analysis, but the main focus is on user-ship.  On this dimensions, del.icio.us dominate, with more than 10x more users than the next best (Magnolia). 

My predictions is that there should be a moderate vortex effect in social bookmarking, with benefits to the category leader because of the greater volume of other users creating value by adding their bookmarks.  On the other hand, in the absence of true personalization in this category, the vortex should be limited, since above a critical mass the additional users add more noise than additional value.  In fact, we should expect to see a number of secondary social bookmarking sites spring up with the goal of attracting a clientele that shares similar tastes.  In a sense, this will let people find their own "neighborhoods" of others with whom to share bookmarks, rather than sharing with the unwashed masses.  Unlike with automated personalization, these neighborhoods will most easily form when they are structured around easy-to-understand syntactic categories, so the usual suspects — religion, politics, and sex — seem most likely as the fracture points.  

No, it probably is not coincidence that these are the traditional "off-limits" topics for casual conversation.  These are topics about which people prefer to talk to people with whom they agree.

I’d personally rather see a more personalized approach from one of the top social bookmarking sites.  Such an approach might lead to a much stronger vortex effect, and the opportunity to dominate the category.

John

ReadWriteWeb talks about a new feed reader that has strong social features, making it easy to share items with other users in rich ways.  They say that FeedEachOther also has interesting recommender algorithms to help people find other feeds to read that are similar to feeds they have read in the past.  Sounds very interesting! 

John

I just finished reading "Bounding Rationality to the World" by Todd and Gigerenzer, published in the Journal of Economic Psychology in 2003.  It’s an interesting article with the core theme that the reason we see cognitive limitations in human decision making is that people may be optimized to make decisions in a particular environment.  In particular, this argument suggests that the apparent deficits in human decision making discovered in various economic experiments may be due to those experiments creating environments that do not correspond well to the sorts of environments humans regularly have to make decisions about.

Several of the issues seem to have interesting implications for recommenders, including:

• people are often making decisions in an environment in which information is not free.  collecting more information to make a better decision might lead to an overall decision that is worse, ironically.  (This reminds me of Colin Powell’s talk in which he explained that he was trained to make decisions with 2/3 of the information; waiting for more information would mean that the decision would not be available in time to help during a battle.)  Can a recommender be set to help a person satisfice appropriately?
• how do known stable cognitive illusions like the overconfidence bias and the hard-easy effect relate to recommenders?  Can and should recommenders be tuned to avoid these illusions?
• knowing more sometimes decreases decision performance.  Can this sort of problem be seen in recommenders? 
• one common decision approach is sequential choice, in which a decider must make a decision without having the opportunity to choose a prior choice once it has been passed on.  What would a recommender be like for a sequential choice problem?  Could it improve performance?
• the paper mentions research that suggests that humans benefit during language acquisition from having limited cognitive ability, so higher abilities don’t get in the way while learning the basics.  This feels related to concepts in machine learning, like those that limit the complexity of the function search space.  However, humans over time increase the complexity of their cognitive function.  What does this correspond to in machine learning?

Overall, recommender research that explores the ways recommenders interact with human decision processes would be fascinating.