[Foundation-l] Letter to the community on Controversial Content

Kim Bruning kim at bruning.xs4all.nl
Mon Oct 10 22:08:30 UTC 2011


On Mon, Oct 10, 2011 at 04:52:48PM -0400, Risker wrote:
> 
> Given the number of people who insist that any categorization system seems
> to be vulnerable, I'd like to hear the reasons why the current system, which
> is obviously necessary in order for people to find types of images, does not
> have the same effect.  I'm not trying to be provocative here, but I am
> rather concerned that this does not seem to have been discussed.


Been discussed to death, raised from the dead, chopped up with a chainsaw,reresurrected, taken out
with a sawn-off-shotgun, stood back up missing an arm...  "they just keep on coming!"


The current category system is not as vulnerable to being abused because it is not a prejudicial labelling
system.

In straight english:

Computers are sort of stupid. They can't infer intent.

A. If we want a computer program to offer something to be blocked, it needs a label that essentially says "This Is
Something People Might Want To Block"

B. A computer program cannot really safely determine what to do with "licking" or "exposed breasts" (especially as
are different norms on what is appropriate in different parts of the world)


Our current category system conforms to B. We would need some sort of mapping to A to make a category based filter
work.

Social problem: Mapping B to A is evil, according to ALA. ;-)

sincerely,
	Kim Bruning

Patient: "Doctor Doctor, it hurts when I map B to A!"
Doctor: "So Don't Do That Then"



-- 




More information about the wikimedia-l mailing list