> One of the things I'm trying to work out is a mechanism like that found in > Black and White. Essentially the agent functions in the background crawling > the web for things you might like. You can then positively or negatively > reinforce it's findings. By keeping a catalog of page analyses, and a > positive/negative rating for "found" pages an agent could be "trained" to > bring back things you're likely to find interesting. That's pretty kickin' - which is precisely what I said to myself, outloud, upon reading that. Not a bad idea, either. While I don't have a constant connection, I do have a shell login to my website, and could always establish it there - thereby generating the 'clippings' page you mention later so I can access it when I *do* logon. My default home page is already my website, why not make it /clips? I like it. The pos/neg option is one I've explored in the past (and continue to, when I happen across such projects) with rudimentary AI systems. VERY rudimentary. ;) I believe a game my brother has, Creatures (1, 2, 3) does similar - you're breeding furry little, er, creatures, and can try to 'teach' them by showing them something, 'saying' the name in the input window, then spanking or petting them based upon their reactions. Sometimes they're stubborn, and go running off to munch on the poison weeds and taunt the grendels anyways, but hey.
> 2> As our interaction becomes less direct, so to does our ability to verify > the source and truth of a web asset. As if it weren't bad enough as it is :/ > 4> Acquisition of a persons mediation agent profile could reveal a lot of > information about them. And what marketing firm wouldn't LOVE that? *encryptencrypt* -- To unsubscribe from this list, please send an email to [EMAIL PROTECTED] with "unsubscribe" in the subject, without the quotes.
