The guides are gradually building a database of approved links that correspond to various search terms. At the launch of Mahalo’s alpha version in May 2007, the engine was already equipped with results for the top 4,000 Internet searches, all manually chosen by the guides for their quality. For each search term, Mahalo provides a list of links that were ruled as mostly relevant by the guides, including the ‘top 7’ of the matches found. Furthermore, the engine separately displays a ‘Fast Facts” window concerning the query with some basic information on the requested term, and also provides users with a set of Google-generated search results.
Mahalo broadens its approved database not only with links selected by the guides, but also with those submitted by the users themselves. Similarly to the popular Digg website, users can submit links for various entries and those sites that were recommended by a sufficient number of users will be examined by the guides, who will add them to a set of Mahalo results for that query. The company keeps a close watch out for possible ‘link spamming’ and immediately bans users who are suspected of such activities. Mahalo says it takes the guides a little over two hours to create a single page with appropriate results for a new term – a process in which each potential site is checked against a list of Mahalo’s criteria.
The criteria for ‘inappropriate content’ in websites is simple – spam-sites, sites that have overbearing advertising, and sites that appear to participate in ‘phishing’ or any other type of malicious activity are all banned from the engine. Mahalo also omits websites that regurgitate information from other sources – with or without giving the appropriate credit, sites of unknown origin, and sites hosting ‘hate speech’ or adult content.
The ‘elite’ search engine does have some disadvantages – its database grows relatively slowly, and many search terms have yet to be processed by the guides and currently yield only Google-generated results. A human-driven search is less likely to produce relevant results as quickly and efficiently as its machine-driven counterpart, but the company believes that with the steady growth of its content pool, the search engine will become more and more practical. “Humans using machines can create much better search results than machines alone. Our ‘Guides’ use Google, Yahoo, Ask, MSN, Flickr, Delicious, and dozens of other services to hand-craft the cleanest, most organized, and spam-free search engine results pages available today” – they say.
The engine is currently ranked in the 69th spot in the US, but experts say it is growing faster than was expected, attracting more and more dedicated users who were not redirected to the website from other search engines but rather use Mahalo as their default resource.
TFOT has previously reported on the “StupidFilter”, an open-source filter software that is able to detect “rampant stupidity” of web-content in written English.