Kaiser censorware study and health sites
Here, very roughly, is what is going on with the Kaiser Family Foundation censorware study and health sites.
It's not complicated. One just has to realized that censorware is not a "filter", but basically a collection of blacklists.
Suppose one has several censorware blacklists:
| Blacklist | Accurate items | Inaccurate items |
| Sex | X1 % accurate | Y1 % inaccurate (including Z1 % health sites) |
| Drugs | X2 % accurate | Y2 % inaccurate (including Z2 % health sites) |
| Rock and Roll | X3 % accurate | Y3 % inaccurate (including Z3 % health sites) |
If you blacklist all of
Sex and Drugs and
Rock and Roll sites, then the blacklist
has the combined accurate items, plus combined also
ALL of the inaccurate items, that is
(Y1 % + Y2 % + Y3 %) inaccurate including
(Z1 % + Z2 % + Z3 %) health sites (neglecting any overlap).
The more blacklists that are used, the more wildly inaccurate bans are seen. The fewer blacklists that are used, the fewer wildly inaccurate bans are seen.
It's that simple.
Note the report does not address structural, architectural issues of censorware, such as the across-the-board banning of "loophole" sites (caches, anonymizers, and translation sites, etc.)
By Seth Finkelstein |
posted in censorware
|
on December 12, 2002 04:31 PM
(Infothought permalink)
| Followups
