Wired magazine has made a startling discovery; that people are better than algorithms when it comes to sifting through all the data on the internet and producing relevant search results.
Thank you, Rick Romero. Do I even have to give the “google search” example? You’ve all done a google search and produced 8890000852 results. And you’ve all gotten pretty good about varying the search language or using operands in order to produce more relevant 1st page searches. This is search 101 and most humans who have access to the internet know how to do it.
But, there are tough searches — searches that just don’t make any sense to you. Searches that only another human with experience, or a different way of looking at the topic, will be able to do: the deep searches. I’m not sure that an algorithm will ever be able to do these properly. Recognizing this, you might turn to a corporation and pay someone to help you, as the article suggest. Or, you know, maybe you could talk to a librarian. They have them at these place, called “libraries.” Most towns still have libraries don’t they?
Apart from Mahalo — which claims to be a human powered search engine — that article has very little to do with “search” and the business of search but focuses on two companies that I would call human-powered recommender systems. brijit.com is sort of like a news aggregator/recommender system/Cliff’s notes and squidoo is like a combination of wikipedia/recommender system/advertising aggregator. The problem with both of these sites is that they do too many things and none of them well. For example, places like Fark and Digg are better news aggregators simply because they are very popular. And both brijit and squidoo’s recommender systems suffer from what all recommender systems suffer from: they take forever to build up a useful relationship with their client. Yes, even if they were just algorithms they are building relationships.
Recommendations are one of the most difficult things to get right because most humans are wary of another human’s advice. Movie reviews, book reviews, restaurant reviews, web site reviews, article suggestions, think about it… if you’re like me, you build a relationship with the reviewer before you start even THINKING about seeing the movie that he suggests. That’s why friends make such great recommenders! Because you know them: you know that Todd has a thing for sci-fi films and so you weigh his recommendation to see Battlestar Galactica accordingly. And it took a long time to build that level of understanding.
Now, you could go on brijit and build a relationship (of sorts) with the submitters there. Read a few hundred of their posts, learn their biases, and maybe after a few months of daily reading, you might find their Cliff’s Notes style news aggregation useful. I just can’t see it. Instead of doing that, I’d recommend that you spend that time building a relationship with your local librarian. They’re infinitely more useful.
The one thing I do have a tinge of fear over is how popular and widely accepted it is that all answers can be found on the internet. I work in a library so I hear it all the time, but this article doesn’t even mention that there are a ton of 24/7 live human search options available through your local public or academic libraries. Not to mention the tons of resources that aren’t available on the internet that libraries can help you get. If you’re really interested in finding out that bit of information. Further, the article doesn’t mention that recommendations are problematic even though we all inherently know how problematic recommendations are.
It’s almost as if Wired magazine has a biased toward internet resources.
From Tacoma, WA, this is Rick Romero reporting.
The UWTLIT Weblog 