I'm glad to see that I'm not the only one who doesn't quite yet get an immediate eureka about the new Google APIs-- searching in particular. Of course there are the non-web crossovers, like searching in AIM via googlematic, but this mostly makes me yawn. Yes, its fun and geeky, but yawn. This is not to say that the Google search API itself makes me yawn.

What makes me yawn is anything that's just an alternative direct user interface on the service. Search from my IDE while I program? Eh, that's okay, but I could do that by just spawning a browser with a cooked URL and not have to re-engineer a UI do display the results. Display some results of a canned search in my weblog? Eh, that's cute, but I could do that with some simple HTML scraping and SSI, if I really really wanted to. Yeah, I know the web service makes that somuch easier, but the thing it makes easier isn't something I was really interested in the first place. Maybe I'm not interested because I don't get it yet, or maybe it really is just a novel triviality.

No, what will make my overhead lightbulb spark up are applications which involve indirection. That is, some application which makes searches to answer some other question of mine. Search results used to spawn further churning. Or, search results as the result of churning. Google's suggestions are intriguing: Auto-monitor the web for new information on a subject; Glean market research insights and trends over time; Invent a catchy online game. But, these sound disappointingly close to a corp-speak shrug.

Not that this is unexpected or a bad thing or a statement of derision. Their Alpha Geeks made the service available, and now its up to the world Alpha Geeks to turn it into magic. I'm just waiting and thinking though... the AG's are churning out all permutations of language bindings, alternative interfaces, and weavings of the service into other apps. This is the first stage of play. I don't know that I'll feel like playing much yet. So I'll watch, and maybe tinker a bit, but mostly be thinking about what the next stage of play will become.


Archived Comments

  • tracerlock? http://tracerlock.com/
  • Okay... Just trying to think how one might do this with the Google search API. Forget for the moment that the "demo" API is limited to 10 search results. So I make a search and download all 186,000 or however many results come up for my terms. I store these. A specified number of minutes later I perform the search again and download the 186,123 results that come up for my search. I run a diff between these and have my changes. Seems like downloading those hundreds of thousands of search results, a periodic times, possibly by dozens to hundreds of users of the app, will be nasty and undesirable. I suppose some pooling of terms and centrallizing the searching into another app would maybe help cushion this... but something seems itchy and wrong about this to me.