Do I Need a Web Recommendation Service?

Xydo is a recommendation startup I first discovered at Hoboken Tech Meetup. Since then I’ve partially trained GetGlue and Hunch to respond to my tastes (not successfully), perused Parse.ly’s recommendation app for filtering feeds, and gauged Google’s own Prediction APIs and Set suggestion tools (pretty good stuff).

So when I received the beta invite from Xydo, I was almost at the beginnings of an existential crisis: do I really need a web site to show me other URLs to look at? After all, I was heavily reliant on Google Reader to bring the feeds I like to my attention. I wasn’t sure whether I required additional content advice.

I would want Xydo and other such sites to be my web magazine 2.0, bringing both content that I absolutely need yet also uncannily anticipate what I may want.

I believe that automated content recommendation is a viable and humane approach to sifting through the terrabytes of Internet content. No human should have to be enslaved to a URL approval assembly line that takes as raw material thousand of RSS feeds. Well, maybe some web publications treat their interns that way.

Many social networks obviously have had great success simply spraying out posts and links. On more than one occasion I’ve become delightfully lost in a Twitter fog as I watched and reacted to random tweets. For a more structured Web surfing safari, I focus instead on my Google Reader page as it blasts out posts from sites that I had previously selected.

Xydo: both a raw RSS topic feed and user generated recommendation stream.

The recommendation services are a middle ground between crowdsourced content courtesy of Twitter, Delicious, FourSquare et al. and posts and articles from edited web publications retrieved using Google Reader and other RSS clients.

Recommendation startups have the potential to find their own niche in web publishing by aggregating and selectively disseminating raw feeds, aided perhaps by algorithms that automate the selection process, along with a community of engaged and active users who can take control of the editorial side as needed.

So that brings us to Xydo. You know the drill:register for a few topic areas (food, movies, technology) and then communicate your likes and dislikes to the crowd through the obligatory “recommend” button. In Xydo, a topic page displays the raw aggregated feed along with the stream of community recommendations.

So who is your editor at Xydo? Perhaps you can find soulmates who share your tastes, and they become your de facto Internet guide. There may be also some automated filtering of the content feeds. Since Xydo doesn’t force users to go through a training session, I doubt there’s much in the way of algorithm generated suggestions.

Xydo is in beta, so it is too early to judge the value of this particular recommendation service. My overall critique of Xydo and its cohorts is that they need decide on a goalpost.

If a recommendation site chooses to morph into a semi-automated web publication with interesting aggregated content, I’d prefer greater control be given to subscribers in customizing what’s delivered. I’ve already happened on a few quirky food blogs—thanks Xydo!— but I don’t currently have a way to dedicate a block of Xydo web real-estate to their posts. To some extent, I want to be my own editor and decide on specific content and how it will appear.

Or if Xydo leans towards more of a community network approach—say, MetaFilter, GetGlue, Aardvark— then it should have a forum in which likes and dislikes can be discussed. As of now, there’s little in the way of web infrastructure in Xydo to handle public conversations.

I’ll check back with Xydo in the coming months to see what they ultimately arrive at.

In truth, the whole recommendation model is in a kind of beta testing phase.

Or perhaps alpha.

Related articles

• Xydo
• Parse.ly’s P3 Platform (technoverseblog.com)
• The View From Hoboken (technoverseblog.com)
• Google’s Pretty Good Recommendation Service (technoverseblog.com)