[Foundation-l] Image filtering without undermining the category system

[Kim Bruning]

On Tue, Oct 11, 2011 at 11:20:15PM +0100, Thomas Morton wrote:
> On 11 October 2011 21:51, Kim Bruning <kim at bruning.xs4all.nl> wrote:
> 
> > On Tue, Oct 11, 2011 at 09:55:46PM +0100, WereSpielChequers wrote:
> > > OK in a spirit of compromise I have designed an Image filter which should
> > > meet most of the needs that people have expressed and resolve most of the
> > > objections that I'm aware of. Just as importantly it should actually
> > work.
> > > http://meta.wikimedia.org/wiki/User:WereSpielChequers/filter
> >
> >
> > Hmm, how would it fare against a marblecake attack? ;-)
> >
> > http://musicmachinery.com/2009/04/15/inside-the-precision-hack/
> >
> >
> I agree on the one hand that anything is potentially gameable but:
> 
> a) Wikipedia is notoriously gameable and yet, fingers crossed, we have not
> had a mass Anon. attack. 

Actually, we've had all kinds of attacks. Anonymous is essentially our
friend though. The folks going after us are <name suppressed to protect
the guilty>.

Thanks to our anti-gaming policy (aka IAR), they don't often succeed }:-)>

> b) The Time "hack" was rudimentary at every step - no matter how the media
> (or this blog) portray it. The root cause of the hack was a technical
> ineptness at a several levels in the Time poll which allowed it to be
> maninpulated on a number of levels.

Quite so. This means that we should learn from their ineptness, rather
than -say- copying it ;-)

> 
> On the face of it any such system might be gameable; but no specific
> implementation details have been laid out (beyond the basic framework). So
> the concern "it might be hacked to force a certain result" is one of the
> most easily addressed :)

General rule of thumb: If you leak value preferences between users, with
no intermediate (soft) security, your system might be game-able. 

> More problematic with your blithe dismissal

I didn't dismiss anything, 
I asked how it would fare against a marblecake attack!


> is that the proposed implementation is inherently not all that
> gameable. This is because, as described, working out the filter for
> any individual is a P=NP (travelling salesman) problem.

And this is an interesting and constructive answer to that question. \o/

:-)

sincerely,
	Kim Bruning
--