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

[Andreas Kolbe]

> From: Tobias Oelgarte <tobias.oelgarte at googlemail.com>
> > Someone on Meta has pointed out that Commons seems to list sexual image results for search terms like cucumber, electric toothbrushes or pearl necklace way higher than a corresponding Google search. See http://lists.wikimedia.org/pipermail/commons-l/2011-October/006290.html
>
> > Andreas
> This might just be coincidence for special cases. I'm sure if you search 
> long enough you will find opposite examples as well. 


Tobias,

If you can find counterexamples, I'll gladly look at them. These were the only three we checked this afternoon, and the difference was striking. 

Here is another search, "underwater":

http://commons.wikimedia.org/w/index.php?title=Special%3ASearch&search=underwater&fulltext=Search


The third search result in Commons is a bondage image:

http://commons.wikimedia.org/wiki/File:Underwater_bondage.jpg


On Google, with safe search off, the same image is the 58th result:

http://www.google.co.uk/search?gcx=w&q=underwater+site:commons.wikimedia.org&um=1&ie=UTF-8&hl=en&tbm=isch&source=og&sa=N&tab=wi&biw=1095&bih=638




> But wouldn't it run 
> against the intention of a search engine to rate down content by 
> "possibly offensive"? If you search for a cucumber you should expect to 
> find one. If the description is correct, you should find the most 
> suitable images first. But that should be based on the rating algorithm 
> that works on the description, not on the fact that content is/might 
> be/could be controversial.

> Implementing such a restriction for a search engine (by default) would 
> go against any principal and would be discrimination of content. We 
> should not do this.



You are not being realistic. If someone searches for "cucumber", "toothbrush" or "necklace" on Commons, they will not generally be looking for sexual images, and it is no use saying, "Well, you looked for a cucumber, and here you have one. Stuck up a woman's vagina."

Similarly, users entering "jumping ball" in the search field are unlikely to be looking for this image:


http://commons.wikimedia.org/wiki/File:Jumping_ball_01.jpg


Yet that is the first one the Commons search for "jumping ball" displays:


http://commons.wikimedia.org/w/index.php?title=Special%3ASearch&search=jumping+ball&fulltext=Search


We are offering an image service, and the principle of least astonishment should apply. By having these images come at the top of our search results, we are alienating at least part of our readers who were simply looking for an image of a toothbrush, cucumber, or whatever. 

On the other hand, if these images don't show up among our top results, we are not alienating users who look for images of the penetrative use of cucumbers or toothbrushes, because they can easily narrow their search if that is the image they're after.

Are you really saying that this is how Commons should work, bringing up sexual images for the most innocuous searches, and that this is how you would design the user experience for Commons users? 

Andreas