[Commons-l] Direct transfer of properly licensed images from wikipedias to commons?

[Magnus Manske]

On 10/17/07, samuli at samulilintula.net <samuli at samulilintula.net> wrote:
> > On 10/16/07, Samuli Lintula <samuli at samulilintula.net> wrote:
> >> On Tue, 16 Oct 2007 22:15:26 +0300, Magnus Manske
> >> <magnusmanske at googlemail.com> wrote:
> >>
> >> > OK, state of the union:
> >> > * WatchFlickr now adds {{flickrreview}} again
> >> > * CommonsHelper has a "direct upload" switch (off by default)
> >> > * Direct upload by CommonsHelper will add {{BotMoveToCommons}} with
> >> > the appropriate source language, so things will end in subcategories
> >> > of [[Category:Files moved to Commons requiring review]]
> >> >
> >> > An example of how it looks:
> >> > http://commons.wikimedia.org/w/index.php?title=Image%3ARosemary_Clooneys_home_in_Augusta.jpg
> >>
> >> I'm a bit lost in this conversation. Did any human review the license
> >> information and such before or right after the image was transferred to
> >> Commons?
> >
> > Well,
> > * /presumably/ it was checked when it was uploaded in the original
> > wikipedia
> > * /presumably/ the guy who transfers it checked (how else does he know
> > the name?)
> > * /presumably/ the guy who transfers it checks the end result
> > * /presumably/ the good people at commons check it when it is transfered
> >
> > IMHO that's an improvement over the "normal" upload to commons ("oh, I
> > found a pretty image on the web, I'll upload it to commons!";-)...
> >
> > If the original wikipedia image has a valid license tag, how should
> > one check anyway? The obvious criteria are:
> > * Image is too large to be a thumbnail stolen from some webpage
> > * Image does not contain borders, logos, or "(c)" texts
> > * Image either has a source given (eg., nasa.gov), or
> > * looks amateurish enough to be made by the original uploader ;-)
> >
> > Yay for reasonable doubt!
>
> Do I detect a bit of a mocking attitude here?

Just a little, maybe :-)

> I don't see any reason for
> that. Why would we want to lose a chance to check images that are being
> uploaded on Commons? It is very good practice that we check if source is
> provided, if the source seems plausible, if permission is provided, if
> there is anything dubious about the image etc. We should try to check
> these things with fresh uploads and transfers from other projects.

Absolutely. And there's nothing to stop anyone from checking the
uploads through the bot account. They are no different from "normal"
uploads, except many (most?) of them will already have been checked on
a wikipedia, as well as by the "transferer", and they all have an
{{Information}} template.

> If we are to presume something, I presume that local projects have at
> least as many copyvios or sourceless/permissionless images as Commons.
> That is, within the range of 10-30 %.

Probably. But by that reasoning, even if we were to blindly transfer
every image from, say, en.wikipedia that has {{GFDL}} in it, 70-90% of
the uploads through the bot would be acceptable for commons. Is that
the case for "normal" uploads? I'd think that the upload rate of
copyvios on commons through "normal" means will increase with the
popularity of the project.

I'm not saying that all images on en.wikipedia that are tagged
{{GFDL}} are good for commons. All I'm saying is that most of it is,
and that the average copyvio rate will likely be lower that what we
get through "normal" uploads. /Especially/ for en.wikipedia, where it
is (relatively) easy to call something "fair use" instead of GFDL, and
be done with it. Those won't even make it through CommonsHelper in the
first place.

Cheers,
Magnus