Carl Beeth Web - Sifting the web

We read the magazines whose editors select the sort of articles that interest us. There is a need for that kind of editorial selection on the web. How could it be achieved?

he other day I was reading a long editorial in WIRED about how push technologies would somehow save the web from itself. Although I agree with WIRED that push is important, I certainly don't think the web needs saving and even less that push technologies would be the one to achieve that. It seems to me that the WIRED editors believe the web would somehow be better if it looked and behaved more like TV. They whine at the fact that it is so difficult to find interesting things among all the junk on the web, and somehow they believe the junk will disappear if it is spoon fed with push.

The single most important improvement the web could get is editorial technology.

The problems of the web are not about push or pull. In my opinion the single most important improvement the web could get is in editorial technologies. When we read a magazine, an editor has taken decisions on what stories are worth reading, and we choose to read the magazines where we feel the editor is doing a good job. On the web we don't have to use a person to do the editing; we can use agent technologies to customize our selection according to our tastes.

Take for example the FireFly agent system that will recommend web sites based on other web sites you liked by making pattern matchings on what other users liked. What I would like to see is a system where this kind of service would be embedded with your search engine so that when you do a search on "color psychology", not only will your search engine return the web sites that contain the words color and psychology but it would order them according to your agent profile, thereby making it more likely that the site will be relevant to you. I know this would be a massive task to achieve. I don't pretend to have any solutions but here are some thoughts on the concept as seen from the user's perspective.

Example of how an integrated rating system could look.

First of all we would need an easier and more transparent way of rating the sites we like. Maybe it could be integrated into the browser interface, or maybe it could be handled by a small floating window using java to monitor the URLs of the other windows.

Secondly we should have the possibility to rate single pages. As an example I like the WebMonkeys site a lot. I think they have many articles that are brilliant, but that a few border on crap. If we limit the rating system to whole web sites we lose those distinctions. A major issue would be how to handle this potentially enormous database of ratings. This leads to the question of how we identify a web site.

One other point is, what do we rate? Should it be how we liked it, how interesting, how relevant, how true or how funny it was? I guess it doesn't really matter, as long as you are more or less coherent in your system. Your agent will hook you up with others using a similar type of rating.

One nice option would be to have the ability to vary the importance of different editors/raters. Let's say that I have a colleague whose opinions I trust. I could ask my agent to take special consideration for his ratings. One might even imagine some famous editors/raters selling you special access to their ratings.

All in all I don't think this kind of system would be impossible. It could make it be much easier to find the memes relevant to you. Now if we could also adapt the concept to the push technologies we would have a really interesting product.