yesterday i was quite surprised to find a washington post article linking to a related new york times article. and today, i was even more surprised when the text and links were removed (see above, the highlighted text used to link).
i was so amazed with the discovery, i did some research, which i’m not nearly as excited about now. anyway, the washington post is using a service from inform.com. the new york times has a similar offering: sphere. below i discuss both offerings, which the newspapers are positioning as means of competing with new news aggregators: yahoo, google, digg, etc. i’m not sure either is successful.
inform
inform.com began in 2004 and has recreated itself a couple times since. the service being employed by the washington post, “Inform Publisher Services”, began in mid-2006. here’s a summary from business 2.0:
Inform sucks in all of a Websites articles and then automagically categorizes and tags each one. So it knows that an article about Lebanon should also show up in a search about Hezbollah even if the word “Hezbollah” never appears in the article. [Inform] is not a believer in human tagging. All of this is done by machine. Once Inform’s software tags and categorizes each article, it can serve up much more relevant search results. It can also create hotlinks to a search-results page for any word inside an article that is also a popular topic, like “George Bush” or “Google.” Each new article a reader clicks on is another advertising opportunity.
one of the first newspapers to use the service was The Oklahoman. they create links within an article which lead to an internal search results page. personally, i do not find this very useful, as with firefox i can already right click on a highlighted term and perform a google search, which i perceive to have more relevant (or less biased) results. nonetheless, this may be of some value to a less technically savvy audience who are accustomed to following links, although i think the user expects to land on a specific page as a result of following such a link, rather than arriving at a search page.
having said that, i like how the washington post implemented the inform results in their article. instead of placing links within the article’s context, they have created a closing paragraph that lists three related links. yesterday’s links were presented at the end of the article. doing sothis is would be effective for several reasons:
- the links lead directly to a given article, skipping the search results page
- the links are manually controlled, and are absolutely relevant
- readers are accustomed to finding related links at an articles conclusion
one possible criticism of this technique is that a reader may miss the links if they do not complete the article, however, i would argue that if the reader has not completed an article they are not likely to want additional information, either becuase the topic is not compelling or they are already well informed on the subject.
the linking was not in fact some new editorial policy, nor does it seem that these links came from inform. above, i have circled the inform gathered related articles link, which then takes you to a results page. i would never have seen this had i not been scrutinizing this article’s presentation so carefully. that entire box of links is a blur to me, although i do like the descriptive text. couple that with an icon (see below) and it could work.
sphere
sphere is a technorati-like product created by sphere and adaptive path. the times has implemented it as an icon, labelled “sphere it!”, at the end of a given article. clicking on the icon reveals a collection of related links, both internal and external. this is presented in a layer that floats over the given article. while i like the presentation, the implementation is somewhat flawed – unless i know what “sphere it!” means, or to use a google term, “i’m feeling lucky”, i probably won’t click on the icon, and therefore won’t see these results. as is, sphere becomes just another web 2.0 icon lost in the shuffle. some descriptove text, like “see articles related to this one with sphere”, or a sampling of the sphere results, followed by “get more related articles from sphere”, would be much more effective.
