During a recent discussion with Eran Sandler, findory came up and it got me thinking. If you haven’t seen it, findory is a pretty cool news site that shows you some news and then learns what you like over time by seeing which items you click on. As it gets to know you, it recommends more relevant items to you. It does this quite well and I find the site interesting. I’m a big believer in the ability of technology to intelligently help you find things you wouldn’t have found otherwise and I think findory’s founder Greg Linden feels the same way.

The question then is why did findory end up riding into the sunset? What’s missing? Why aren’t people knocking down doors to allow recommendation technology that’s really quite good to simplify their lives?

I think the answer lies in an inherent lack of trust between the user and the system. If I’m to trust a computer with the task of figuring out what news I should be reading, then I better learn to trust that computer, and fast. If I don’t, then I’ll eventually stray and visit other information sources in order to make sure that I’m not missing out on anything.

When Findory’s traffic growth slipped for the first time in 2006, Greg Linden blogged about it. The second comment on that blog post said the following:

One feature I would request is the ability to view my favorites by recency instead of (and as well as) by relevance ranking. I often want to scan all the article summaries from my fav’s to make sure Im not missing out on something.

If I had to guess, I’d say that this guy isn’t using findory anymore, but he’s using a feed reader of some sort.

To further clarify the problem, lets imagine a different version of Amazon.com. What if Amazon showed you 50 random products when you first arrived and nothing else. No browsing by category. No search. By looking at what you choose to view from that initial sample, it learns what you like and what you don’t like. Over time it starts to recommend more products to you. Every time you go to Amazon you get a cool list of 50 products that follow your interests. In my case, I’d probably see a lot of technology books, gadgets and movies. That would be handy.

But what happens when I want to do some serious shopping? What happens if I try to shop for something I’m not typically interested in? Am I going to make this my main destination? No, I’m not. I’m going to go somewhere where I have the ability to browse by topic, to search, to dig deep, to make sure that I’m not missing the perfect product for my needs. There’s no way that I’m going to trust the personalization technology to show me the perfect 50 items. It just can’t possibly do the job well enough. That’s why Amazon’s recommendations only supplement the browseable nature of their inventory. It has to be that way.

Now while the stakes aren’t nearly as high when it comes to consuming news online, the same underlying issues apply. I value my time quite highly, but I also place value on my ability to be informed accurately and completely.

Until recommendation technology becomes at least an order of magnitude better than it is today, it will have to provide the user with the ability to “prove it”. In order for me to determine if the recommendation technology is trustworthy, I must be able to explore the possibilities myself and compare. And, If I have to leave your site to make that determination, then I’m probably not going to come back.

This entry was posted on Monday, April 23rd, 2007 at 1:46 pm and is filed under Musings. You can follow any responses to this entry through the RSS 2.0 feed. You can leave a response, or trackback from your own site.