Why I’m switching to Yahoo Search

[Disclaimer: Yahoo supports RDFa, which is a specification I've worked on. So, obviously, I'm excited. But hey, that doesn't mean I'm wrong.]

Yahoo recently announced SearchMonkey, and for the first time in 10 years, I have a reason to switch search engines, from Google to Yahoo (In fact, I just did that in Firefox.) Most web-savvy engineers know that online services succeed in big ways when they become platforms: when other developers can expand on the functionality in ways not foreseen by the original developers. Yahoo is the first to figure out how to do just that with a major search engine. With SearchMonkey, any developer gains the ability to provide custom ways of extracting and presenting page data within Yahoo search results.

Let’s be specific…

Let’s say I’m a recruiter who often searches for candidates by searching the web. I log in to Yahoo and add the “Resume” application. From now on, whenever I search, each search result that is a page at LinkedIn, CareerBuilder, or Monster.com, is presented in a new layout which immediately highlights the skills and latest experience of each person right on the Yahoo research page. Super cool.

And again, anyone can write this Resume application. The code is hosted by Yahoo, so there is zero operational cost to the developer. Users can pick and choose any number of Search Applications to customize their search experience. In other words, because SearchMonkey applications are low-cost and targeted, they can fill highly specific needs, maybe the needs of a few hundred people only. Developers get the power of the Yahoo search engine, with cheap customization down to tiny niches.

And the best part is that Yahoo has separated data extraction from data presentation. Specifically, a data extractor for LinkedIn can produce RDFa, and different presentation applications can use different portions of that RDFa. If the extractor for Monster.com produces the same RDFa, then the same presentation application can be used to display both Monster.com and LinkedIn data. And if LinkedIn and Monster.com produce RDFa natively, within their web sites, then there’s no need to build data extraction… the presentation application can work natively on the raw web pages.

And that’s going to be the major incentive for web publishers to join the data web community, where human-rendered markup is augmented with machine-readable labels, using technology like RDFa. This is quite exciting.