TechFold - Bold tech & web commentary

Bold tech & web commentary

TechFold is technology discussion, commentary, reviews, and opinions from well outside the valley. There's no koolaid to drink here, and TechFold is not in SL, or on Twitter.

Keegy: Geo-relevance rankings

Keegy has an interesting concept: Its a website that serves up “relevant” news by correlating what other people from your same geographic region clicked. Localization is a good buzzword to be touting at the moment - Keegy’s execution seems to be confused on a number of levels.

1. Geography? Arguably, when it comes to web browsing, geographic location is a poor basis for relevance correlation to begin with. Clickstream behaviour is motivated by interests, which in internet land are loosely if at all correlated with location. i.e.: A gamer living in Winnipeg is more interested in international “gaming” news, not news that other people from Winnipeg may have enjoyed. It seems to me that location should be one spectrum of correlation - not the only one.

2. Transparency - Its unclear how Keegy relates my location to the content it serves up. At what level does the relevancy calculation take place? Province? City? Country? Can I change it? Given that there doesn’t appear to be a way to get an un-modified view of the site, I need to know “how” my world view is being generated.

3. Ranking - imagine if the Digg homepage didn’t tell you the number of diggs any given story had gotten. Would you find the page more or less trustworthy as a news source? Just how relevant are results are on Keegy is currently a mystery - i.e.: how much clickstream data is the page that Keegy served up based on? What I want is a “digg equivalent” score for each story that clarifies its ranking: “Viewed by 28 Winnipegers, 76 Manitobas, and 891 Canadians” would be a nice summary.

The Keegy press release helpfully offers this non-explanation:

“In a personalized news service users interact with the site and their activity anonymously generates statistics for each city/country. Using this information, an artificial intelligence algorithm ranks the posts for relevance according to a visitor’s location and the stories and home pages are edited automatically every minute.” [from Press Release]

4. Don’t Confuse Yourself with a Search Engine: For some reason, Keegy creates and maintains its own index of content blogs, fundamentally limiting the depth and breadth of Keegy-served content to its own crawl. Why not tie Keegy in with a browser extension and let users roam the net? I imagine that would create a much more valuable and interesting clickstream, and keeping only reasonably correlated data (across a given geographic user body) would eliminate “outlier” content.

In a nutshell, Keegy seems like a half-implemented slice of a larger project. Err, fortunately Keegy is “…closing seed capital stage on November 21th. and starting their first round of investment next October.” [from Press Release] Well, good luck to you. In a crowded market of social discovery services, I don’t think Keegy has the spark to go big.