divide all existing web-content
into units
(pix, videos, tweets, etc)
new units are added continually
units have authors
(but may be posted without attribution
or with false attribution)
each unit
hypothetically
has a potential value
for each surfer
(partially dependent
on the context of presentation:
i appreciate flickr minimalism more
when it's recommended by akiruna)
the goal is to efficiently find
as much highrated content
as you have time for
(lowering these standards
when you have extra time)
the primary filter
is recommender/linkblogger
subscribing to recommendations
from those we trust
(though recommenders
are usually more trustworthy
in some genres than others)
and a secondary filter
(fewer items per day)
is author
another filter is subject
subscribing to [tags]
eg on del.icio.us or flickr
and even worse is
popularity
aggregated recommendations
like digg or Favrd
linkbloggers
evaluate N items per day
and filter out M items
(M<<N)
.