Greetings. I'm Eric Litman, I'm an Internet entrepreneur, investor and executive, and I run a mobile ad and analytics company called 
Trusting Human-Powered Search
Published by Eric Litman on Wednesday, October 10th, 2007 2:22am
Jason Calacanis today writes that Mahalo’s challenge is breadth rather than depth. Fair enough, but to me currency, and by extension accuracy, are of even greater importance for their sustainability. In search, accuracy = trust = user retention1.
Google’s trust model is based on a broad acceptance that their crawlers do a good job of finding pages and that their indexers and ranking algorithms are likely to lead you to the most relevant results for a given search. Certainly there are flaws in the model and room for improvement, from which the category (of one?) of human-powered search draws its raison d’être, but it has been successful enough to attract an enormous user base. Yahoo!2., Ask.com, and other algorithmic search engines operate under a similar model.
What’s the trust model for a human-powered search engine (HPSE)? I’d argue that it’s strongly influenced by three factors:
- Quality and context of search results.
Not only does the cruft need to be filtered, but results need to be segmented and described on a results page to clearly identify their value. - Authority.
I’m open to comment on this one, but my belief is that for a given HSPE results page to be useful it needs to be a canonical source for the most salient points of a given topic. I don’t want to feel like I need to dig through Google’s results in addition to what an HPSE provides me. - Currency.
People need a transparent view into both when a link was added and when a human last verified that it remains current to trust that content is relevant to them at the time of their search. Note that neither Mahalo nor its older, distant cousin About.com do this today.
There’s a fourth factor I’ll toss out for discussion, and that’s the personal brand of the reviewers themselves. This is much less of an issue today while the model is still proving itself than it will be two or three years from now when people have learned to value the quality of work of the individuals crafting the content. After all, if HPSEs’ primary selling point is that they’re “human-powered”, isn’t it reasonable to think people are eventually going to pay attention to the “humans” powering them?
About.com lost much of its early luster because savvy folks have learned not to trust it. Much of its content is junk, frankly, either because it was hastily written to fill a void or the data represented there are out of date. Mahalo doesn’t really have this problem yet given its newness, but I already find myself questioning the currency of some of the links I find there, just as I do with About.com. Other HPSE’s will have to combat this issue as well.
What influences your trust in search results? Can/do you trust people over algorithms, or vice versa?
2. Yahoo! combines a directory with their search results, although the integration these days has been de-emphasized. Interestingly, Mahalo is essentially an evolution of what were the origins of Yahoo!. Think of it as Yellow Pages meets Lonely Planet, but with fewer plumbers’ cracks and less patchouli.
jonny goldstein on Wednesday, January 2nd, 2008
In the end, the search engine that gives me the most useful results earns my trust. Now, through inertia, I’ve been using Google for the last many years. Maybe there is something better out there, but for now Google is so useful that I have not felt the need to fish around for anything better, if indeed their is anything better.
Gerald Buckley on Wednesday, April 16th, 2008
What if you were in a VERY vertical, commodity space… the “content” was very volatile (short shelf life) and the need/expectation to get “it” right (context + authority) were critically important. For the sake of argument, let’s say it’s the securities market. Pretty important to have timely and correct data. Or, let’s say your FareCast or Zillow or any number of very vertically oriented information silos…. What would it take to be relegated to Ask.com-like oblivion?
I’m curious to know as that’s a real risk my own early stage startup might face… Bet your bippee I want to hedge against being irrelevant. Will be interested in your take.