So I do a search on google for Yann Tiersen. I want to legally download one of his songs, and it’s not on iTunes or Buymusic. I can enter “music download” and end up with a million news articles about pirated music. I can enter “Yann Tiersen” and end up with a million fansites. But I don’t want that. I want to download a specific song.
Can we create a search engine based on the Dewey decimal system? One that categorizes sites and uses specific search terms, i.e. to buy music you type “Music Yann Tiersen Purchase” while to go to find fan sites you search for “music Yann Tiersen fansites” Or perhaps something similar to the tag searches one can perform on www.flickr.com. I’m also thinking we should include a little Wiki think in there, because we need some way to prevent people from mislabeling their sites to bring people in. Is it plausible? Can anyone think of any pitfalls? Also, can any computer programmer types get on top of implementing it, because I want my Yann Tiersen download, dammit!
… Plus, even if you did have a good classification, you’d need a building full of people doing nothing but looking at and classifying web sites and they’d still get behind on all the new web sites. For those of you young 'uns, that’s what yahoo and some other search engines tried to do in the early days (when the internet used stone tablets and dinosaur couriers). They couldn’t keep up even with the paleozoic proto-internet. Google’s robot searchers go a lot faster, and can actually index most of the current internet before it changes too much.
The other problem with classifying web sites is that sometimes it can’t be specific enough. For instance, you might not be interested in all celebrity gossip web sites, just those that mention the latest incident with Paris Hilton, the Oakland Raiders, and the slaughtered camels. In that case, some kind of more detailed altavista/google -like indexing is necessary.
I don’t think the system I propose will completely replace Google and their ilk. After all, if you want the newspaper article in question you can just Google “Paris Hilton Raiders camels” and get what you’re looking for. Google’s great strength and great weakness is that it is basically a keyword search. I’m envisioning a websearch that acts as a subject and author search.
Further, I don’t think it would be necessary to hire a building full of web page researchers. I’m not interested in cataloguing every website out there, only the most pertinent ones. We could use a system similar to wikipedia, in which anyone can submit a link to a website (note that I wish to deal in websites, not individual pages).