Jason Calacanis, co-founder of Weblogs and Entrepreneur in Residence at Sequoia Capital, recently announced the public alpha of his new company Mahalo, a human powered search engine. I'm not entirely convinced it should be called a search engine since it functions more as a well edited directory, but the top results pages are remarkably more informative than those from Google. I'm glad they're addressing the problem of search engine optimization running amok in certain areas such as hotels and travel since I find those searches incredibly frustrating.

The aspect I found most striking about the announcement, though, was the human component. In an era of improved automation and algorithms the decision to rely on people power appears ridiculous. However, there are indications that human involvement in previously automated tasks may become a powerful differentiator for high-tech companies. The key depends on exactly how cleverly the companies can harness the brains of humans.

One of the most intelligent applications of human effort I've seen comes from reCAPTCHA, out of Carnegie Mellon's School of Computer Science. They're using the CAPTCHA concept (verifying users are not robots by asking them to read small bits of garbled text) to get people to digitize books. With over 60m CAPTCHAs read each day there's a tremendous capacity to get work done. The beauty of this is they've figured out a way to get people to complete two unrelated tasks concurrently without the user even knowing.

Big players like Google are savvy to the concept too. Many have suggested that GOOG-411, a free local voice search service, was simply an effort to build a huge training data set of people speaking (and their privacy policy confirms this). eBay's acquisition of StumbleUpon on Thursday for $75m could also be perceived as a foray into harnessing human work. The 2.4m users with StumbledUpon toolbars are implicitly categorizing themselves and the websites they visit.

In thinking more about Mahalo, I'm sceptical about how well they've harnessed the efforts of people writing custom search results. While the output is great, they have not found a way to implicitly harness human efforts as the other examples above. However, with $16m of venture funding from top investors and claims that this is a five year project I wouldn't write them off.

Going forward it will be interesting to watch for more technology companies that rely on humans. There will be opportunities both for leveraging their output and creating tools to capture their input. Consider a company like 82ASK which is a service to answer questions texted from mobile phones. With technology developed for routing questions to either a database or humans I wonder if they might someday be perceived as an infrastructure company?