headers
Jason Lunz | 1 Dec 2006 05:52

Re: "Captcha" when posting?

larsi <at> gnus.org said:
> What do all y'all think about adding one of those image thingies that
> show a word, and people have to type in what the word is to be able to
> post?

Norse people say "all y'all"?

Jason
Permalink | Reply | 3 comments |
headers
Lars Magne Ingebrigtsen | 1 Dec 2006 06:59
Face
Picon
Favicon
Gravatar

Xapian-assisted spam reporting for fun and glee

It just kinda occurred to me that I could use nnweb to find spam to
report.

With the following settings in .gnus.el:

(setq spam-install-hooks nil)
(require 'spam-report)
(setq spam-report-gmane-use-article-number nil)

I typed `G w gmane RET OutOfDebt RET'

and got a buffer with about 100 matches.  I narrowed the buffer to the
From field that seemed to fit the pattern (to avoid messages that were
taking about the spam (like this message :-)), leaned on `#' for a bit
to process-mark all the messages, and then `M-x spam-report-gmane-spam'.

Presto!

I did the same with "Symbol: CNPM" (which is a string found in the
latest pump'n'dump stock spam), and got 800 (!) messages.  Repeat and
rinse.

Not only is it pretty easy to do it this way, it's really fast, due to
Olly's fantastic Xapian work.

One caveat, though: We still haven't set up the thing that's supposed
to ping Xapian whenever a message is cancelled, so if you search for
"OutOfDebt" now, you'll still find the messages via nnweb, whether
they're cancelled or not.  Reporting them more than once doesn't harm,
though, but it's a nuisance.
(Continue reading)

Permalink | Reply | 52 comments |
headers
Adam Sjøgren | 1 Dec 2006 07:57
X-Face
Picon
Favicon
Gravatar

Re: "Captcha" when posting?

On Fri, 1 Dec 2006 04:52:17 +0000 (UTC), Jason wrote:

> larsi <at> gnus.org said:
>> What do all y'all think about adding one of those image thingies that
>> show a word, and people have to type in what the word is to be able to
>> post?

> Norse people say "all y'all"?

It a rodeo up there.

  Best regards,

    Adam

--

-- 
 "Limonès, c'est nous"                                        Adam Sjøgren
                                                         asjo <at> koldfront.dk
Permalink | Reply | 2 comments |
headers
Lars Magne Ingebrigtsen | 1 Dec 2006 08:29
Face
Picon
Favicon
Gravatar

Re: Xapian-assisted spam reporting for fun and glee

Lars Magne Ingebrigtsen <larsi <at> gnus.org> writes:

> Presto!

Hm.  But there seems like there's a faulty mapping between nnweb
headers and articles sometimes.  I need to investigate...

--

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi <at> gnus.org * Lars Magne Ingebrigtsen
Permalink | Reply | 2 comments |
headers
Lars Magne Ingebrigtsen | 1 Dec 2006 08:33
Face
Picon
Favicon
Gravatar

Re: Xapian-assisted spam reporting for fun and glee

Lars Magne Ingebrigtsen <larsi <at> gnus.org> writes:

> Hm.  But there seems like there's a faulty mapping between nnweb
> headers and articles sometimes.  I need to investigate...

Actually, it looks like http://search.gmane.org/nov.php is outputting
the wrong Subject headers or something...

--

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi <at> gnus.org * Lars Magne Ingebrigtsen
Permalink | Reply | 1 comment |
headers
Lars Magne Ingebrigtsen | 1 Dec 2006 08:57
Face
Picon
Favicon
Gravatar

Re: Xapian-assisted spam reporting for fun and glee

Lars Magne Ingebrigtsen <larsi <at> gnus.org> writes:

> Actually, it looks like http://search.gmane.org/nov.php is outputting
> the wrong Subject headers or something...

Nope; it was just a mis-mapping in nnweb.  So I'd suggest that if
people want to use this for spam reports, they should do a CVS update
on their Gnusae first.  :-)

--

-- 
(domestic pets only, the antidote for overdose, milk.)
  larsi <at> gnus.org * Lars Magne Ingebrigtsen
Permalink | Reply | 0 comments |
headers
Matej Cepl | 1 Dec 2006 10:02
Picon
Favicon
Gravatar

gmane.comp.editors.vim.outliner has different email address

Probably because of mistake was gmane.comp.editors.vim.outliner registered
with its old address. New (and stable) address is
vimoutliner <at> vimoutliner.org
(see http://lists.vimoutliner.org/mailman/listinfo/vimoutliner). It is
Mailman-based list.

Best,

Matěj
--

-- 
http://www.ceplovi.cz/matej/blog/, Jabber: ceplma<at>jabber.cz
GPG Finger: 89EF 4BC6 288A BF43 1BAB  25C3 E09F EF25 D964 84AC

This message has been composed of recycled electrons. None of
these electrons has been harmed or injured in the creation and
transmission of this message but they have been shamelessly
exploited for this use.
Permalink | Reply | 2 comments |
headers
Steinar Bang | 1 Dec 2006 11:43
Picon
Picon
Favicon

Re: "Captcha" when posting?

>>>>> asjo <at> koldfront.dk (Adam Sjøgren):

> On Fri, 1 Dec 2006 04:52:17 +0000 (UTC), Jason wrote:

>> Norse people say "all y'all"?

> It a rodeo up there.

It's the Texan influence through the oil industry.
Permalink | Reply | 1 comment |
headers
Jouni K. Seppänen | 1 Dec 2006 14:11
Picon
Picon
Picon
Favicon
Gravatar

Re: Xapian-assisted spam reporting for fun and glee

Lars Magne Ingebrigtsen <larsi <at> gnus.org> writes:

> I narrowed the buffer to the From field that seemed to fit the
> pattern (to avoid messages that were taking about the spam (like
> this message :-)),

This didn't work quite perfectly in all cases; check your spam report
approval logs for (a small number of) rejected reports. Some of the
messages discussed a piece of spam received by the mailing list,
e.g. warning people not to reply to the "remove" address, and in the
process quoted pieces of the spam. In at least one case, someone's
signature quoted a discussion on spam, and at least one message was
about debugging some mail software (Horde?) where the debug output
included parts of spams. There was also some mailing list (Cocoon
documentation or something like that) where the project's wiki sends
diffs of changed pages, and when someone had removed spam links from
the wiki, the diff included bad words.

On some mailing lists, there is already pretty good spam filtering,
and hitting them with this kind of semi-automatic reporting is likely
to cause false positives. I wonder if it would be feasible to check
somewhere in the process the historical spamminess rate of the mailing
list, and be more careful with less spammy groups. The check could be
a part of the automatic reporting process, or there could be some
hints to report approvers.

One indicator of non-spamminess that I imagine would currently work is
if the message has an In-Reply-To or References header with a
message-id pointing to a previous message on the mailing list. Since
you already do threading in the web interface, you probably have the
(Continue reading)

Permalink | Reply | 46 comments |
headers
Lars Magne Ingebrigtsen | 1 Dec 2006 14:20
Face
Picon
Favicon
Gravatar

Re: Xapian-assisted spam reporting for fun and glee

jks <at> iki.fi (Jouni K. Seppänen) writes:

> This didn't work quite perfectly in all cases; check your spam report
> approval logs for (a small number of) rejected reports.

The main problem was that nnweb was buggy -- it showed the
summary/from data from one article, while displaying a different
article.  The submission function the submitted that article.  :-/
So what articles I submitted was actually quite random.  

I've fixed that bug now.

> On some mailing lists, there is already pretty good spam filtering,
> and hitting them with this kind of semi-automatic reporting is likely
> to cause false positives.

Especially on the gmane.mail.spam.* groups.  How do you deny groups
from a search result on Xapian?  Just say -gmane.mail.spam.*.  Hm.
Nope.  

> I wonder if it would be feasible to check somewhere in the process
> the historical spamminess rate of the mailing list, and be more
> careful with less spammy groups.

Yes, that would be helpful.

> One indicator of non-spamminess that I imagine would currently work is
> if the message has an In-Reply-To or References header with a
> message-id pointing to a previous message on the mailing list. Since
> you already do threading in the web interface, you probably have the
(Continue reading)

Permalink | Reply | 45 comments |

Gmane