Bjoern has pointed out a flaw in that some filterers might get trigger happy, but that can be resolved by giving people the option when they decide to click on an image A not to filter that image in future and B to disregard everything else filtered out by the person who thought that image problematic. We could also slightly complicate the system by splitting option 3 into a cautious and a very cautious button - very cautious filters anything that any other filterer thought problematic, and cautious ignores filterers who have often been ignored by other filterers.
I would have thought that a Bot trawling all images to see which have been objected to by somebody would probably be blocked as a denial of service attack, afterall how many readers actually read more than 100,000 articles a year?
Re Stephen Bain's point re Flickr, I raised Flickr in a previous thread as proof that whether or not this is theoretically possible it has been done in practice. Fae then criticised the way Flickr operates its filters, hence my design which I hope would work and I believe would avoid the problems we would have in using the Flickr approach.
Re Andrew's point re readers on blocked IPs, we have ways of creating accounts for people who are caught up by IP range blocks. If the overlap between readers wanting to create an account in order to filter images and readers caught up in IP range blocks becomes excessive then we could probably create a filter only account for them.
I'm uncomfortable about a session cookie based system for IP readers, many of our readers are in Internet Cafes and I'm not sure if PCs in those sorts of environments get rebooted and the session cookies wiped between customers.
WereSpielChequers
Message: 6 Date: Thu, 22 Sep 2011 13:44:09 +0200 From: Bjoern Hoehrmann derhoermi@gmx.net Subject: Re: [Foundation-l] A possible solution for the image filter To: Wikimedia Foundation Mailing List foundation-l@lists.wikimedia.org Message-ID: 1c7m771u12n25l5tdkdcafdo1kvf49sm79@hive.bjoern.hoehrmann.de Content-Type: text/plain; charset=ISO-8859-1
- WereSpielChequers wrote:
For obvious reasons we don't want a system that creates a publicly
available
set of filters that net nannies of various descriptions could use to stop other people from seeing things that they deemed inappropriate.
This cannot be prevented. You just need a bot that emulates a reader who has the desired filter settings enabled and then load all the images or articles or whatever and check what is blocked and then you have a list.
- Hide all images and just show caption and description. (recommended
for users with slow internet connections)
(I note that it's trivial to blur images on the client side and reveal them on hover or tapping or whatever input method would be appropriate.)
- Show all images except ones that I or another editor have decided
not
to see again
This will not work unless you introduce some process to block editors who put too much on their filter list for some definition of "too much". -- Bj?rn H?hrmann ? mailto:bjoern@hoehrmann.de ? http://bjoern.hoehrmann.de Am Badedeich 7 ? Telefon: +49(0)160/4415681 ? http://www.bjoernsworld.de 25899 Dageb?ll ? PGP Pub. KeyID: 0xA4357E78 ? http://www.websitedev.de/
Message: 9 Date: Thu, 22 Sep 2011 22:08:45 +1000 From: Stephen Bain stephen.bain@gmail.com Subject: Re: [Foundation-l] A possible solution for the image filter To: Wikimedia Foundation Mailing List foundation-l@lists.wikimedia.org Message-ID: <CAO5b2fv=85w2wJa9nsFFGFoXicHgL_1BEq+gL8QtR2jPMTh1Wg@mail.gmail.com
Content-Type: text/plain; charset=ISO-8859-1
On Thu, Sep 22, 2011 at 9:19 PM, WereSpielChequers werespielchequers@gmail.com wrote:
One of the objections is that we don't want a Flickr style system which involves images being deleted, accounts being suspended and the burden of filtering being put on the uploader.
When have any of those things been part of the proposal?
-- Stephen Bain stephen.bain@gmail.com
Message: 10 Date: Thu, 22 Sep 2011 13:16:16 +0100 From: Andrew Gray andrew.gray@dunelm.org.uk Subject: Re: [Foundation-l] A possible solution for the image filter To: Wikimedia Foundation Mailing List foundation-l@lists.wikimedia.org Message-ID: <CAE4f==fnQ2R59zUkwDMrQ_kA8t8Xubyzcw=bwcBaTMet_NkudA@mail.gmail.com
Content-Type: text/plain; charset=ISO-8859-1
On 22 September 2011 12:23, David Gerard dgerard@gmail.com wrote:
On 22 September 2011 12:19, WereSpielChequers werespielchequers@gmail.com wrote:
So is there a simpler way to do this, is there some flaw in this that
would
prevent it working, or is this the flying unicorn option?
I believe it was envisioned as working for anonymous casual readers as
well.
There *should* be some way to at least have the no-images option for anonymous readers without ruining caching ...
Cookies? It would work on at least a per-session basis, I'd think.
One issue here is that if we make it registered-user-only we need to work out how this interacts with account creation - and IP blocks. It clearly will cause problems if people *want* to turn on the filter, go to create an account, and discover one of our famed cryptic block messages telling them they can't...
--
- Andrew Gray
? andrew.gray@dunelm.org.uk
foundation-l mailing list foundation-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/foundation-l
End of foundation-l Digest, Vol 90, Issue 135
On 22 September 2011 14:46, WereSpielChequers werespielchequers@gmail.com wrote:
I'm uncomfortable about a session cookie based system for IP readers, many of our readers are in Internet Cafes and I'm not sure if PCs in those sorts of environments get rebooted and the session cookies wiped between customers.
It varies depending on the specific location - some effectively reboot and wipe the profile between users, some merely kick one user out of the seat and put a new one in without even closing the browser tabs. Same with domestic one-computer-per-household situations!
On the other hand, the proposed implementation is relatively transparently reversible - concealed images are shown as a placeholder with a trivial "click to display again" option - and this sort of "legacy filtering" should be fairly easy for a user to switch back on or off. It's not perfect, but it's probably no *less* clunky than requiring people to sign in (and the associated forgetting-to-sign-out...)
wikimedia-l@lists.wikimedia.org