Been underwater most of this week, and it looks to continue. But here are a few nuggets worth checking out:
Greg points to Amazon Zuggest, a Google Suggest/Amazon mashup. He also points to some interesting personal search work at Microsoft, courtesy Todd Bishop.
Rumors about Yahoo buying Flickr are getting to a boiling point, one source who pinged me (I can’t verify this) said it’s done deal at $20 million, $10 million now, $10 million on earn out. We’ll see….
Philipp reports that Mark Jen, famous for getting fired for blogging while at Google, is now at Plaxo.
Jon Udell is doing some interesting things with Google Maps. On Kuro5shin, some neat things with Google Print.
Technorati has launched “related tags” which is a cool idea. Weinberger reports.
Matt McAlister, who was employee #1 at the Standard, took the wraps off his personal blog and posts on “Why Google Isn’t What It Used to Be.”
SEW has a pointer to an article about a visit to Google India.
Via AdWeek, (SEW also notes this) Yahoo anticipates the end of its relationship with MSN (via Overture). We all knew this was coming.
Via Slashdot, a Threadwatch post noting that Google is cloaking its own pages, perhaps. This practice is banned by Google’s own TOS.
Regarding McAlister’s post: What a bunch of jargony B.S. Who the hell uses the term “digerati” anyway?
Comparing Google to the CIA??? Let us see what’s in their database??? I *know* what’s in Google’s database: Links! The content is on the Web, not google’s hard drives. (Don’t even think of pointing out the cached pages either.)
Wish he had the balls to enable comments. I’d post there. Alas.
I’ve spent too much time fighting comment spam in the past, so comments won’t be enabled until I get that sorted out. Sorry, dendrite.
I agree with you, that a semi-random collection of documents can’t provide enough insight into what someone is searching for. It surely won’t work for shared computers, and for personal computers where recipes and jokes are mixed in with research.
Even beyond that, any one person will be searching based on their needs at the time. A bioscientist researching caffiene will have a lot of research reports on their disk, and an algorthim would probably notice that. But if they search for “coffee”, there’s no way to tell if they want technical papers, local cafes, online roasteries or what.
I think “personalization” for search is best offered as suggestions for more precise queries, rather than as an intervening filter for search results.