How recommendations can isolate people

It’s curious to me that Findory is fading on us, perhaps slipping into oblivion. It’s not the only startup with a good idea to see the lights dimming through the window of opportunity. And it’s certainly not for a lack of insight or technical chops.


I love the idea of automatic learning and recommendations. But the application of that concept makes more sense in the context of an experience as opposed to an experience itself. Last.fm understands this and represents it more effectively than just about anything I’ve seen on the Internet yet.

Scott Karp faults Findory for failing to capture the serendipity of learning what you didn’t realize you needed to know…and, particularly, who else doesn’t know it yet.

“Perhaps the good old fashioned niche is as personalized as we need to get. TechMeme and Digg are highly niche sites, and thus are tailored to the specific interests of their users without getting so personalized as to break the bonds of community.”

Succeeding as a startup in online media is also part first-mover. It’s part aesthetics. It’s part who you know and who you adopt into your friends and family.

It’s also part editorial voice. I think Findory is an interesting mirror into my interests, but I’ve found that I’m not an interesting filter through which to view the world. There are probably few individuals who are. Similarly, what I liked most about Megite was seeing other people’s personalized news, not so much my own.

But there are many communities and concepts that filter the world of information in ways that are interesting and also matter in a particular context. The filter (or media brand) represents the community and gives meaning to the information. This is publishing 101 stuff, and it applies online as much as it does off.

Succeeding in online media is also part fashion. Findory never reached a critical mass of coolness. It’s hard to say what difference the fashion sense of the engineers and founders at startups makes in their ability to succeed, but decision-making at any company is driven at least in part by intuition, moreso at startups. Without keen intuition for market forces or at least your customers, your efforts will be out of synch with the world you live in.

Related to all this, I suspect the personalized My New York Times and similar publisher offerings are going to fail or at least prove lackluster for similar reasons. None of them are first-mover offerings. None of them are pretty or engaging. None of them provide a community filter that makes information relevant and interesting. They lose the editorial voice of their parent. They’re too hard to setup. And they just aren’t cool.

On the other hand, publishers can take the behavior data their readers are generating on the site and apply that in a way that surfaces interesting information or in fact connects them to similar people directly. TechMeme is a pureplay in this space, but it’s applicable at any media site. Greg Linden probably has great software to make that possible for all publishers’ sites.

I’m also curious to see how publishers pick up on socially engaging widgets like MyBlogLog and the new ways to build community within your domain. I like how Scott framed up the issue in this quote:

“Despite all the hype about the “user in control,” purely personalized news may be too much control, a slippery slope that leads to solipsism. The proverbial “water cooler” is symbolic of our fundamental need to share the news, to validate our experiences by sharing them with others. How can there be “conversation” if we’re all talking about something different?”

Nice.