How search engines are getting smarter

Search gets personal

In May 2008, SEO expert Mike Grehan wrote: "Search will gradually become more of a personalised experience. Personalisation and digital asset optimisation will end 1999-style ranking reports, as search engine results will be based on blended results from end-user specifics, such as geographic location, time of day, previous searching history, and peer group preference".

Marissa Mayer said it in less words in her blog post on the future of search: "Search engines of the future will be better in part because they will understand more about you, the individual user." Google's SearchWiki service is a step in this direction. Currently, SearchWiki enables the user to move results higher up the rankings, remove unwanted results, add missing results and leave public comments for other web users to view.

"SearchWiki is something that we're really interested in to allow users to really personalise their searches, and to give feedback to tell us what they think about the search ranking that we're doing," says Galler.

Are we going full circle, back to the days of Yahoo when results were compiled manually? No, says Galler, although people can improve on computer results. "The internet is growing very fast and will be growing much faster in the future. The question really is, 'Can this massive growth be covered by just humans, or is it more efficient to have a computational process?'

"Humans can improve specific aspects of that – that's partly what we do with SearchWiki. We don't need a human to go out and investigate something and come back and list it. What we do is, he enters the search and we provide a result. If he thinks he has a better site that we haven't ranked, he can include it. So having a computational back-end that scales with the internet allows you to generate a basis, and then humans can go in and say 'I like this site, I don't like that site, I can leave a comment on one or rank one higher'. So right now we do this for personal use, and I think that's the most important."

Stoddart agrees that, while computer-generated results are the most accurate, we shouldn't rule out a human element. "It might be a way of helping the algorithm learn quicker, but technology is an amazing thing. We take a different approach: we have a very algorithmic-based machine learning system. We've always said we believe the algorithmic way is the right way, because a computer will pick things up faster than a human will, especially when it comes to picking up patterns."

While search engines learn ever more detailed information about our search habits, which they can then use to refine future results, should they also incorporate the knowledge and preferences of our friends in the results?

It's massive privacy minefield, says Goddart. "There's a privacy issue there which we'd have to work through and say: what do you want your friends knowing about what you're doing, where you shop and where you don't shop?

"When will people be ready to share this information in the right way? People could jump into this and get into very scary situations very quickly. We're very socially aware and we'll do it when we have the right protocols and policies in place to make sure people get the best experience possible."


First published in .net magazine, Issue 185