On 16/10/2007, Magnus Manske <magnusmanske(a)googlemail.com> wrote:
My watchflickr tool [1] includes an option to upload
an image with a
suitable CC license to commons using a "bot" account [2]. So far, I
have received no complaints about bad uploads, and from its gallery it
seems OK as well (except some duplicate uploads).
It's a winner. I've been using it on placeholder images on en:wp.
WatchFlickr has about one hit in fifty, but that one is worth the
effort. Often the image found is crappy and needs cropping and
tweaking to be any good, but often enough it's bot-uploadable.
Now, anyone can upload an image to flickr, and release
it under
CC-BY(-SA). Same as wikipedia, right? Except that wikipedia uploads
are probably screened much more thoroughly for cases that are clearly
not under the given license.
The one thing I find problematic with the auto-uploading bot is that
it accepts the tag on Flickr as valid. I've found far too many cases
where the tag on Flickr is clearly invalid and the Flickr uploader's
just been sloppy.
That is: have the robot leave the image as requiring human validation,
and that should be cleared up.
My CommonsHelper tool [3] eases the transfer of images
from wikipedia
to the commons, and has been used a whooping 93435 times this year.
Assuming that every use results in an upload on commons, over 330
images /per day/ enter commons this way, a not unimportant proportion
of the 5000 uploads per day, especially considering that it will only
take images that have a commons-compatible license.
However, users still have to save the image on their own computer,
then upload them under their own user account, which is annoying and
time-consuming.
CommonsHelper does have the functionality to do direct uploads via the
aforementioned bot account, however, that has been deactivated since
forever, due to concerns.
I think a checking step like Flickr checking would be suitable.
I would like to propose the reactivation of that
feature. Concernes
about unsuitable uploads through the bot account are superflous, IMHO,
since images are screened thrice this way:
1. On the wikipedia where the image was originally uploaded
2. By the CommonsHelper (e.g. it will reject "fair use" images from en)
3. On commons, by the usual suspects :-)
Tag it as needing step 3 and it'll be fine until then.
All in all, I'd estimate that there's between
0.5 and 1 million images
on the wikipedias that would be suitable for commons. You can see how
"save locally, then upload manually" annoyance can scale up :-)
Indeed.
Providing you add a tag for "could a human check this please" then
anything that makes it easier would be marvellous, thank you!
- d.
