idiomag

Most people agree that the online experience will increasingly involve personalisation, as content becomes almost infinite and therefore the process of filtering becomes paramount. Some companies (eg Digg) have approached this from a social perspective, relying on friend networks or crowd wisdom to help us find what we want. Others (eg Matchmine or indeed idiomag so far) have approached this from an algorithmic perspective, mapping user interests and developing systems that make the selection decision. Since both have strengths (the social approach is able to incorporate emotional and other tacit factors, whereas algorithms can lead to a greater degree of individual accuracy and scalability), it is likely that they will be almost always used in combination.

Since we are already developing a solid algorithmic approach, I have had a think about the social side of things. There are three key social filters that are helpful in my search for content. 

Firstly, I want experts to inform me. These are people I trust, because they have positions of credibility (Mossberg @ the WSJ), expertise based on experience and success (Paul Graham), or a reputation formed from a large following (Michael Arrington). They provide a first-level filter for more important coverage, and they provide a knowledgeable opinion on the subject as well as a brief factual overview. However, they don’t provide an all-encompassing view and they can’t provide all the relevant information on the topics I am most interested in.

Secondly, I want friends to help me form my own opinions on the issue. These are people I trust, like or am interested in, but they aren’t necessarily experts. Their reaction to coverage is important, as it fleshes out the story and provides a wider perspective. I also listen when these friends recommend content to me, as they know my interests or we share similar interests.

Thirdly, I want crowd wisdom to highlight coverage of interest to me. Beyond the bounds of my friends’ recommendations, I am interested in reading what everybody rates as good. Two examples are that when I want a funny video or quirky story, I see what the popular vote at Digg gives me, and when I want to buy a new laptop, the crowds’ opinion via eBuyer is more important to me than advertising, brand or even price. Crowd wisdom is not only useful for recommendations, it also creates an incredible depth of further information about any topic (Wikipedia, Blogsearch etc). This all-encompassing viewpoint allows me to explore topics of interest in more depth.

All three are important and provide different functions to ensure I receive a broad but interesting outlook, and can explore in-depth when required. A system that can effectively combine these filters in the areas in which they are strong, would be very powerful indeed.