danielknox.net

I was excited to read this afternoon that Microsoft is on the verge of rolling out a new version of the MSN.com homepage to its 100M+ users.

A preview of the redesigned homepage is available here. The redesign finally gives the site a more modern look (farewell “iconic” blue background) and introduces some cool features around search, social media and local news. However, the following feature is potentially the most interesting:

Headlines on the page will now be customized based on user behavior, so, for instance, people who tend to be more interested in entertainment news will be more likely to see those type of stories.

I love this direction….  given the loyalty of their audiences, the big portals like MSN, Yahoo and AOL should be able to do a really good job of dynamic personalization (the sort of personalization that happens in the background without requiring the user to make choices or selections). As far as I can tell, Yahoo and AOL aren’t doing dynamic personalization (yet), so it seems that MSN may get a headstart in this area.

Hopefully MSN will be forthcoming with performance stats over the coming weeks / months as users adjust to the new design and features.

Last week Google Reader introduced two new features to help users find interesting content. Google claims that one of these, the new and improved Recommended items section, has items “selected just for you”.

Today was the first time I had the chance to play with the Recommended items section and I have to admit to being totally unimpressed. For some reason, Google Reader thought I’d be interested in ctrl+z stationary (above), glass toilets, little people and some research about women’s preference for hairy geeks! In fact almost every item in my Recommended items section was a popular (100+ ‘likes’) photo or cartoon with no relation to the hundreds of items I read every day through Google Reader. If there is personalization happening it is either incredibly subtle or not very good.

People trust recommendations more when the engine can explain why it made them.

Interesting observation in What is a Good Recommendation Algorithm? post by the ever insightful Greg Linden.

While he concedes the company didn’t do much technology innovating in its early years, he says that’s changing. Digg hired Anton P. Kast, a former assistant professor of mathematics at the University of California at Berkeley, to assemble a small research team. Kast et al have produced software that links people with similar interests and makes recommendations to people based on their preferences. “This is not something you can build over a weekend,” says Rose.

From Whatever Happened to Silicon Valley Innovation? (recent BusinessWeek article on innovation in Silicon Valley).