Re: Syndicate: bureau automation - oops

[Amy Alexander <plagiari@plagiarist.org>]

On Mon, 20 Aug 2001, m e t a wrote:

> a spider could be programmed - in pearl
> & it would be subscribed 2 syndicate.
>
i have some experience with this:
http://thebot.org
http://netsong.org

> it could crawl the web - or the syndicate archives

i found it's easier to write a spider that stays on one server
(e.g syndicate archives.)
when i was working on the spider for above,
i had to write in lots of behaviors for situations like a) it hits a dead
end for links (must back up and follow another branch) b) it keeps chasing
ic) as a non-frames, no acrobat
browser webpages constantly try to direct it to netscape, adobe, or the
bottomless pit of microsoft's site. (i programmed my bots to just never
go to those sites, or it would be stuck there forever, evidence that these
are the true center of the web... )

and d) something to be aware of: what is ultimately revealed by the bots
in theBot and netsong, what i shamelessly call "the narrative of the web"
is that no matter where they start, spiders fairly soon drift to the
population centers of the web, the endless stretches of web-centric
corporate marketspeak pages; i.e. the bulk of the web is selling the web.
the path to this "final destination" can be interesting, but maybe not
what you had in mind in this case.

on the other hand, a bot that sticks with one website can be more
manageable. you can start out with the example bot here, then modify to
taste:

http://www.oreilly.com/openbook/webclient/ch06.html#30995

though of course it depends on the goal of the spider. a spider that is
absorbed in only syndicate becomes ego-centric if there is not a
particular reason for it to avoid venturing out....

> collect / process / digest
> and mail us its output - once a day / week / year.
>

i'm curious what you have in mind -  collect/process/digest in what way?
what do you think would be interesting/useful? (i've
often thought of having the netsong bot sing the nettime archives, but i
think this would amuse only me.) i can think of some amusing and/or
potentially thought-provoking algorithms that could be applied to mailing
list archives, actually tried something similar years ago with usenet...
but again, this idea was/is for entertainment/artistic purposes but not
particularly functional in the traditional sense. would be of course
groovy to be both.


> or - it could mail its output 2 another spider - programmed differently.
>

how can a spider follow another spider's trails and still be a spider?
(i.e. a spider follows links but the links would seem to now be
a list...)

not denying the possibility; just asking....


-- 
plagiarist.org
Recontextualizing script-kiddyism as net-art for over 1/20 of a century.


-----Syndicate mailinglist--------------------
Syndicate network for media culture and media art
information and archive: http://www.v2.nl/syndicate
to post to the Syndicate list: <syndicate@eg-r.isp-eg.de>
to unsubscribe, write to <majordomo@eg-r.isp-eg.de>, in
the body of the msg: unsubscribe syndicate your@email.adress