(Half Japanese: Greatest Hits, Disc B: “Stripping for Cash”)
…more people read this story on professional poker in the Times than anything on the editorial page.
Dave Winer’s much-hyped new search app seems kind of useless. Unless I’m missing something, what it does is search a single weblog’s archives (in his demo, Winer’s own Scripting News) by keyword with Google and then reorder Google’s results by date. If you have access to the actual archived posts themselves, this could be done trivially and with better accuracy by directly searching them (after all, Google could miss a page or not yet have indexed recent posts; or, even worse, Google could mistake a sidebar text–which you wouldn’t want in your search results–for archive content). E.g., Moveable type users might search through their databases for entries containing a keyword (or set of keywords, or whatever) and then sort the result set by post date. This could be done with about 5 lines of php/perl/python, one line of SQL, and no Google API keys.
And this isn’t exactly new, either. The search box at the top of Moveably Type’s editing pages works in this way exactly, searching through posts for a term and sorting them by date.
So, I don’t get it. What’s the big deal? And why use Google? After all, Google’s biggest value-add is their results sorting, not their index of any specific site (plus having to deal with Google’s API keys seems to be a pain). Maybe Frontier’s built-in database (n.b., Winer uses Frontier for his webapps, including weblogs) makes this kind of search difficult to do and so doing it in this roundabout way seems revelatory to Dave?
How’s that for stop energy? Sorry.