On Thu, Oct 19, 2023 at 09:05:07PM +0200, Michael Koch wrote:
> > IIRC each regex give -10 so multiple matches give more.
> 
> So far, the search strings contain only combinations of keywords, for example "concretevictoria".

> False positives are almost impossible.

strings that are certain spam, you can just add that twice for double the points lost


> 
> I could add shorter keywords, for example "concrete" and "victoria".
> Then we would get multiple matches, but also a few false positives.
> Especially for people from Canada. Keywords would be:
> "ottawa", "halifax", "victoria", "moncton", "nanaimo", "vancouver",
> "concrete", "excavating", "fabricators", "lawncare", "landscapedesign", "retainingwalls"
> Shall I do that?

no


> 
> I agree with you that the problem is not trivial.
> 
> > I think the captcha works as intended. It adds cost to the spammer
> I doubt its economic for spammers to solve a captcha for having some spam
> up for a few hours.
> 
> It's not only for a few hours. We have the spammer's links in our archives forever.
> http://ffmpeg.org/pipermail/ffmpeg-trac/2023-October/067523.html
> Search engines can find the links, and that's what the spammer probably wants.

should i add /pipermail/ffmpeg-trac to robots.txt ?
is there anyone who googles in these ffmpeg trac archives ?

thx

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

During times of universal deceit, telling the truth becomes a
revolutionary act. -- George Orwell