[prev in list] [next in list] [prev in thread] [next in thread] 

List:       kde-look
Subject:    Re: Idea for tool preview on mouseover...
From:       Dave Leigh <dave.leigh () cratchit ! org>
Date:       2002-03-10 5:01:25
[Download RAW message or body]

On Saturday 09 March 2002 13:38, Sean Pecor wrote:

> First I'll preface my argument that I'm NOT a proponent of the capability I
> suggest. I was merely brainstorming in my original response. However, every
> contrary argument I have so far seen is operating on the assumption that
> the retrieval technology is stupidly written. Here is how the preemptive
> retrieval system I'm proposing might work:

The problem I see is that making the thing more "intelligent" doesn't 
necessarily improve it. For one thing, what you propose doesn't match the 
typical user's browsing habits (and I'm not even talking about just my habits 
here). To make it truly useful such a feature would have to be able to 
differentiate "interesting" content from "uninteresting" content. That's what 
users do when they browse.

Take a typical "portal" site. Try Netscape.com or whatever page is provided 
by ISP "X".  You'll find plenty of links all over that page that have nothing 
to do with the "interesting" content which you'll rarely if ever visit. Quite 
likely a lot of them are located in a header at the top of the page. Once in 
a blue moon you might want to edit your preferences or visit whatever 
self-serving links they've put up there, but most of the time your eye never 
even strays there. You bypass that AND the sidebars AND the banners AND even 
the "poster" ads in the middle of the bloody page and go straight to the 
content in the white space.

News sites are typically constructed the same way. and your eye treats it the 
same way. You learn to filter the noise. But the "top down" approach to 
cacheing the links is guaranteed to get the noise FIRST. Not nice.

The point is you can't get the machine to determine what's "interesting." You 
can't even leave it up to the webmasters to mark it with metadata, because to 
them it's the revenue stream (i.e. the ADVERTISEMENTS) that are 
"interesting." They'd mark the ads to generate false hits on the adverts. 
Which begs the question what happens when the ad companies find out that 
they're paying out for ads that aren't even displayed in large quantities due 
to this technology? This is different from blocking an ad. 

These aren't the only objections I have, but I think it's more than enough 
for now... I'm beginning to feel like I'm clubbing a baby seal.

I'm with Dre... if bandwidth is cheap, get a faster connection. Then the 
cacheing is moot. 

-- 
dave.leigh@cratchit.org
http://www.cratchit.org

Have you seen the latest Japanese camera?  Apparently it is so fast it can
photograph an American with his mouth shut!

[prev in list] [next in list] [prev in thread] [next in thread]