Thoughts on Mahalo
I got an invite to try the Mahalo extension the other day. Mahalo is a human powered search engine, hand picking the best results for the worlds’ top 10000 queries. I installed the thing and so far I can’t say I am impressed.
• Web search is all about long tail. Google is so powerful that is teaches us to make more precise queries. More words = better results. But more words also makes it less likely the query will belong to the top 10′000. I wonder how many queries of my Google search history would be covered by Mahalo but I expect a very very limited number.
• Aren’t the top 10′000 searches the ones where a human search engine is the less needed? Aren’t these searches the ones for which Google has the most material to make good rankings? What’s the point of having a human search engines on terms like “nfl” or “mtv”, surely some of the most popular queries in the world?
• The Mahalo extension runs the query to Mahalo, then when there are no results (99.9% of the time) it goes to Google. It goes with the principle that Google is a complement of Mahalo, but all signs point toward the exact opposite. Mahalo is, at best, a complement to their “friends at Google”, because of the simple fact their processes don’t scale.
• Isn’t Mahalo feeding Google? If I am the guy doing pagerank in Mountain View, I use Mahalo to make the Google rankings better, piggybacking the human search engine to refine the machine rankings.
• What’s new compared to Yahoo or About.com? That’s the most puzzling aspect maybe. What’s new?
• How can you pretend to be a serious search engine when you don’t even cover the world’s most searched term ;) ?
I really wonder how this thing will play out. Jason Calacanis is a bright guy, he might have a master plan we can’t see yet. But it’s hard to see Mahalo becoming a major player at this point and with this model.
