mediajunk

For those of you who, like me, are very interested in search engines (you'd never have guessed, right?), it's worth checking out this month's edition of e-marketing news. There's a fascinating, lengthy interview with Yahoo's Search Manger, John Glick. In it, Glick describes how Yahoo went about taking all the knowledge it had gained through its acquisitions of erstwhile competitor search engines, and put it together to create a new technology, aimed at taking on Google. Glick also provides many insights into what Yahoo regards as spam, and the measures it takes to prevent it (these often differ from Google's). I don't buy Glick's attempts to explain Yahoo's new paid inclusion programme (called SiteMatch), which has all the usual fudges we've been getting from Yahoo. (See my post about how SiteMatch inevitably means stale listings.) Nor do I buy any of the claims that Yahoo's future ambitions -- personalisation and local search -- are motivated by the need to better serve searchers. Rather, they are aimed at better serving merchants (or, from Yahoo's point of view, its customers). The examples that Glick provides make this clear, as they smack of overtly commercial concern, not genuine concern for the searcher's needs: "For example, if you been looking at a lot of travel sites and you type China, then you may want information on the country. If you've just been looking through jewellery sites, or wedding venue sites and that sort of thing, you may be planning a wedding and maybe you're looking for China dishes! So it's taking that type of information. "And it's not just which product you're looking for. It's where you are in that product buying cycle ... A person who simply types in: iPod - maybe looking to figure out what they are, should they be getting into this digital music download thing and dropping their CD player for the new thing. Whereas, further into the process, they may be looking for who's got the best price on iPods. Or maybe it's about, does the store down the street have the new one in stock yet. It's taking a lot of that context." Still, lots of useful info here, plus some juicy jargon, including "hill-climbing genetic algorithms" (as opposed to "tree-parsing/gradient-boosting" ones), and "de-aliasing tables". More marginally valuable information to clutter up my small, overcrowded brain...