libravatar-fans team mailing list archive

[Olly Betts <olly@xxxxxxxxxx>]

On Mon, Jun 21, 2010 at 01:34:11PM +1200, Francois Marier wrote:
> On 2010-06-21 at 09:42:21, Brett Wilkins wrote:
> > Should we really be worried about users uploading porn and other
> > offensive images to libravatar?
> 
> In general I would say yes, because we are hosting other people's content. I
> suppose we could be banned by Internet filters if we had too much
> objectionable content.

Some images may be illegal - for example, Germany has laws against display
Nazi imagery, so an avatar consisting of a swastika is problematic, apparently
even if it's explicitly anti-Nazi (e.g. a crossed out swastika):

http://www.religionnewsblog.com/16139/german-nazi-symbols-law-may-need-change

Another example: insulting Ataturk is a crime in Turkey which has resulted in
Youtube being blocked there:

http://www.guardian.co.uk/world/2007/mar/07/turkey

I suspect that there are many similar issues in various jurisdictions.

There's also the issue of using copyrighted images in/for avatars, which seems
fairly common.  I doubt you get many attempts at DMCA takedowns for avatar
images though.

> > If people want to, they will try and we can't really stop them from doing
> > that. The notes in the TODO list appear to imply screening images for
> > offensive content, which in my mind seems to be a task that is doomed to
> > fail if any one site gets too big.
> 
> It might be a big task, but maybe we should at least have the ability to do
> so? I guess I was thinking of something like StatusNet where users can click
> a "Report this" or "Flag this user" button. So it's not moderation, but
> rather the ability to respond to things that are flagged as inappropriate.
> 
> Also, with images, I think it's a bit easier because you could just bang a
> whole bunch of 80x80 images on a single page and then click on the ones that
> need to be disabled.

It seems there are two aspects here - one is what the software should support
and the other is what the instance of the software running on libavatar.org
should do.  The latter is more of an concern in that you don't want the site
to be blocked as pornographic, or to be arrested next time you visit Germany
or Turkey or ...  So you need both support in the software, and a process
(and some way to man it) for the site.

And for the site you'd want a policy on what is and isn't OK.

> > Why not take a more passive approach, and allow the internet to screen the
> > images for us? There is no better source of effort in this regard, after
> > all. I'm thinking of some kind of reporting functionality.
> 
> I agree with a passive approach and a reporting button or form. But at the
> same time if a Libravatar server owner wants to browse all images, it would
> be nice to be able to do it. For example, if a school uses Libravatar, they
> will want to make sure not of the students have objectionable images.
> 
> Also, given that images will be shown on other sites, out of context
> (i.e. unless you look at the URL, you won't know it comes from Libravatar)
> it's a big hard to have people flag images they find are inappropriate...

How about a URL which flags an avatar as offensive?  Then as well as linking
to the avatar, sites using libravatars can add a "flag as offensive" link or
button next to it.

So perhaps:

<img src="http://libravatar.org/<hash>">
<a href="http://libravatar.org/flag/<hash>">Flag as offensive</a>

And the link takes the visitor to a page which explains what they are about to
do with a captcha or similar to prevent spambots and other abuse.

Or perhaps just make the avatar a link to an "about this avatar" page on
libravatar.org, where you can flag it.

For federated images, the offensiveness report could either be forwarded or
reported locally and used to feed a federation blacklist.  Forwarding seems
a reasonable default.

Cheers,
    Olly