On Thursday, July 1, 2004, 11:47:37 AM, Don Newcomer wrote:
> Hat's off to the folks who brought out the new SURBL checks! Here's my top
> 15 rule hits over the past 20 hours and look where my 4 URIBLs come in:
> 18832 - HTML_MESSAGE (0.100) - 50_scores.cf
> 10296 - BAYES_99 (5.400) - 50_scores.cf
> 9810 - OB_URI_RBL (4.0) - surbl.cf
> 9403 - MIME_HTML_ONLY (0.320) - 50_scores.cf
> 8367 - WS_URI_RBL (3.0) - surbl.cf
> 7922 - CLICK_BELOW (0.100) - 50_scores.cf
> 5220 - HTML_LINK_CLICK_HERE (0.100) - 50_scores.cf
> 5102 - SPAMCOP_URI_RBL (3.0) - surbl.cf
> 4470 - MIME_MISSING_BOUNDARY (1.838) - 50_scores.cf
> 4401 - MY_SHRT_IMG (0.848) - coding_html.cf
> 4285 - MK_BAD_HTML_05 (0.3) - coding_html.cf
> 4118 - NO_REAL_NAME (0.160) - 50_scores.cf
> 4111 - AB_URI_RBL (5.0) - surbl.cf
> 3887 - SARE_FROM_SPAM_WORD3 (0.100) - 70_sare_header.cf
> 3678 - MIME_HTML_NO_CHARSET (0.561) - 50_scores.cf
Thanks much for the data and the compliments Don! I'm
forwarding your results to the SURBL discussion list.
It's interesting to see how well ob is detecting spams. My hat
is off in thanks to the OutBlaze folks for providing the data.
Still looking for anyone's spam detection rates and false
positive rates with all the lists:
sc.surbl.org - SpamCop spamvertised sites
ws.surbl.org - sa-blacklist, BigEvil and other data
ob.surbl.org - OutBlaze spamvertised sites
ab.surbl.org - AbuseButler spamvertised sites
ds.surbl.org (beta, 6dos data)
Jeff C.
--
Jeff Chan
mailto:jeffc@surbl.org
http://www.surbl.org/
>-----Original Message-----
>From: Jeff Chan [mailto:jeffc@surbl.org]
>Sent: Thursday, July 01, 2004 3:39 PM
>To: 'SURBL Discussion list'
>Subject: Re: [SURBL-Discuss] uptilt.com and their customers
>probably not
>s pammers
>
>
>On Thursday, July 1, 2004, 6:51:32 AM, Chris Santerre wrote:
>> -1 here
>
>> I think private is better. Public reports of FPs I am +1.
>
>OK To me FPs and whitelisting are the same thing, but it
>sounds like if we did this you're saying it should be called
>"FP reporting" and not "whitelist reporting". Am I interpreting
>what you're saying correctly?
>
Yup. (I can be confusing sometimes.) Saying "Tell us who should be
whitelisted." vs. "Did someone get marked incorrectly?" are different, yet
the outcome is the same....sort of.
If we never listed ebay, then no one will ever report ebay as an fp.
CON: No proactive submissions. We may miss a popular domain.
PRO: We don't get tons of whitelist submissions. Taking all our time
checking them to see if they are legit instead of checking domains the
should be added.
--Chris
-1 here
I think private is better. Public reports of FPs I am +1.
--Chris
>-----Original Message-----
>From: Jeff Chan [mailto:jeffc@surbl.org]
>Sent: Thursday, July 01, 2004 7:51 AM
>To: SURBL Discussion list
>Subject: Re: [SURBL-Discuss] uptilt.com and their customers
>probably not
>spammers
>
>
>On Thursday, July 1, 2004, 4:40:54 AM, David Hooton wrote:
>> On Thu, 1 Jul 2004 02:33:33 -0700, Jeff Chan <jeffc(a)surbl.org> wrote:
>>> the relative lack of inclusion across multiple SURBL
>>> data sources leads me to think that uptilt.com is probably
>>> not a spamhaus. Therefore I've used that list of uptilt.com
>>> domains and customer domains as a whitelist. That means they
>>> will not be included in SURBLs.
>
>> Thanks for your super diligence, we all appreciate it, so do
>our clients :)
>
>Thanks for your kind words David. As you can see I like to avoid
>false positives. :-)
>
>
>I'm strongly tempted to make a public form for submitting
>whitelist entries, fully logged, rate-limited and reviewed,
>of course. It could help with the FPs and add more public
>visibility to the whitelisting process.
>
>Jeff C.
>
>_______________________________________________
>Discuss mailing list
>Discuss(a)lists.surbl.org
>http://lists.surbl.org/mailman/listinfo/discuss
>
Here's some additional info on some of the recent
sa-blacklist/ws.surbl.org/6dos goings on, copied from a
message I sent to the SpamAssassin-users list:
> 1477667 Jun 21 18:48 /etc/spamassassin/RulesDuJour/blacklist.cf.20040623-0106
> 421286 Jun 21 18:49 /etc/spamassassin/RulesDuJour/blacklist-uri.cf.20040623-0106
> 1459329 Jun 23 00:03 /etc/spamassassin/RulesDuJour/blacklist.cf.20040624-1602
> 415544 Jun 23 00:04 /etc/spamassassin/RulesDuJour/blacklist-uri.cf.20040624-1602
> 1484137 Jun 24 15:48 /etc/spamassassin/RulesDuJour/blacklist.cf.20040627-0301
> 422228 Jun 24 15:49 /etc/spamassassin/RulesDuJour/blacklist-uri.cf.20040627-0301
> 1559813 Jun 27 02:17 /etc/spamassassin/RulesDuJour/blacklist.cf.20040628-1544
> 443922 Jun 27 02:18 /etc/spamassassin/RulesDuJour/blacklist-uri.cf.20040628-1544
> 5432965 Jun 28 15:25 /etc/spamassassin/RulesDuJour/blacklist.cf.20040628-1558
> 1544207 Jun 28 15:28 /etc/spamassassin/blacklist-uri.cf
> 7070231 Jun 28 15:54 /etc/spamassassin/blacklist.cf
I think I can help explain why sa-blacklist went from 1.5 MB
to 5.5 MB in size suddenly. Chris Santerre added a fairly large
set of records from 6dos (6 degrees of spam) around that time in
order to get the records into ws.surbl.org and sa-blacklist.
Chris, Bill and I then discussed this and decided to take them
back out of sa-blacklist and therefore ws.surbl.org, and put
the 6dos entries into its own SURBL instead.
However Bill's server experienced a hard disk problem around
the same time so the entries have not come out of sa-blacklist
yet. But they will come out once Chris gets access to Bill's
server again. Until then, backing off to an earlier version
of sa-blacklist makes perfect sense and it's what we've done
for ws.surbl.org.
When Chris gets in again, he will get the 6dos entries off
sa-blacklist, it will come back down in size, and I'll restore
live feeds of ws.surbl.org from the sa-blacklist data instead
of freezing it at the older version, as it is now.
Hopefully this makes some sense. If not I'll glady try to
answer any questions or comments anyone has, though I'm not
the original source of the changes.
Jeff C.
--
Jeff Chan
mailto:jeffc@surbl.org
http://www.surbl.org/
They got on my list due to the tonyrobbins.com incedent.
http://tinyurl.com/28xpc
I that case I believe a "marketing company" was hired, who spammed for
tonyrobbins.com.
They are at least off-white in my mind.
--Chris
>-----Original Message-----
>From: Jeff Chan [mailto:jeffc@surbl.org]
>Sent: Thursday, July 01, 2004 5:34 AM
>To: SURBL Discuss
>Subject: [SURBL-Discuss] uptilt.com and their customers probably not
>spammers
>
>
>I made a list of some of uptilt.com (emaillabs.com) customers':
>
> http://www.emaillabs.com/clientlist.html
>
>likely domains:
>
> http://spamcheck.freeapp.net/whitelists/uptilt
>
>and checked those domains against all existing SURBLs.
>The only matches were:
>
> uptilt.com
>
>on ws.surbl.org, due to Chris' earlier manual listing, and:
>
> digitalimpact.com
> uptilt.com
> wordbiz.com
>
>on ds.surbl.org (6dos) which is a pretty low hit rate for
>nearly a hundred domains against a fairly aggressive list
>*if uptilt.com were a spamhaus*. It's not totally conclusive,
>but the relative lack of inclusion across multiple SURBL
>data sources leads me to think that uptilt.com is probably
>not a spamhaus. Therefore I've used that list of uptilt.com
>domains and customer domains as a whitelist. That means they
>will not be included in SURBLs.
>
>Jeff C.
>--
>Jeff Chan
>mailto:jeffc@surbl.org
>http://www.surbl.org/
>
>_______________________________________________
>Discuss mailing list
>Discuss(a)lists.surbl.org
>http://lists.surbl.org/mailman/listinfo/discuss
>
Yesterday I activated ds.surbl.org in our local server for very few domains,
so far, I got 54 hits, NO false positives.
Regards.
--
Mariano Absatz
El Baby
----------------------------------------------------------
Asking if computers can think is like asking if submarines can swim.
This is a forwarded message
From: Jeff Chan <jeffc(a)surbl.org>
To: William Stearns <wstearns(a)pobox.com>
Date: Wednesday, June 30, 2004, 8:45:52 AM
Subject: : Re: procmail exited w/ EX_TEMPFAIL and stucked spamd processes
===8<==============Original message text===============
On Wednesday, June 30, 2004, 7:38:42 AM, William Stearns wrote:
>> From: Damon McMahon <inst_karma(a)hotmail.com>
>> Are you by chance using Bill Stearn's sa-blacklist?
>>
>> Same thing happened on my mail server yesterday - it almost ground to a
>> halt, stalled procmail processes all over the place, sendmail errors
>> identical to yours - very ugly.
>>
>> Eventually traced it down to the fact that the sa-blacklist.current had
>> inadvertently quintupled (5x) in size overnight, from 1.5 to almost 7 MB in
>> size. I use daily auto-updates of this list so hadn't noticed until the
>> side-effects appeared.
> Am I missing something? Please take a look at
> http://www.stearns.org/sa-blacklist/sa-blacklist.current.domains
> - 23,752 domains, just
> before Chris started adding 6dos. I relinked the last good one when I
> brought the server back up.
For generating ws.surbl.org, that's perfect, and I've re-enabled
my getting of sa-blacklist.current.domains from your server.
When Chris can get the 6dos data disabled, and you link back to
an un-6dosed but otherwise current version, I think we will be
fully functional again in terms of the ws.surbl.org generation
process. I suppose you could rename the 6dos blacklist to
disable it also.
> If you're using my list, what version are you using? 200406281446
> should be a safe one to use.
I was mainly referring to the folks using sa-blacklist as a cf
file, which one of the versions apparently did get several times
larger than before, presumably due to the inclusion of the 6dos
data.
Jeff C.
--
Jeff Chan
mailto:jeffc@surbl.org
http://www.surbl.org/
===8<===========End of original message text===========
>-----Original Message-----
>From: Jeff Chan [mailto:jeffc@surbl.org]
>Sent: Wednesday, June 30, 2004 4:40 AM
>To: 'SURBL Discussion list'
>Subject: Re: [SURBL-Discuss] Remove overlapping rules due to SURBL?
>
>
>On Wednesday, June 30, 2004, 12:46:50 AM, Martin Lyberg wrote:
>> This is the lists i use at the moment:
>
>> # sc.surbl.org - SpamCop message-body URI domains
>> # ws.surbl.org - sa-blacklist domains as a SURBL
>> # be.surbl.org - BigEvil and MidEvil domains
>> # ob.surbl.org - OutBlaze spamvertised sites
>> # ab.surbl.org - AbuseButler spamvertised sites
>
>> I wonder if any of my following rulesets is overlapping the
>SURBL-lists and
>> should be removed?
>
>> I have the following rulesets:
>
>> 70_sare_adult.cf
>> 70_sare_random.cf
>> 70_sare_specific.cf
>> 72_sare_bml_post25x.cf
>> antidrug.cf
>> backhair.cf
>> bigevil.cf
>> chickenpox.cf
>> evilnumbers.cf
>> tripwire.cf
>> weeds.cf
>
>Hi Martin,
>On behalf of everyone contributing to the SURBL poject, thanks
>for your kind words. Glad you're finding SURBLs useful.
>
>Chris or one of the SARE guys will know a lot more about
>the specific SARE rules, but I know that the domains in
>bigevil.cf are in be.surbl.org, so you may be able to
>get rid of bigevil.cf. For that matter the records in
>be.surbl.org are now in ws.surbl.org, so you can get rid
>of be.surbl.org also.
>
>Chris & Co are probably creating a heavily wildcarded ruleset
>that you may want to use in future in addition to SURBLs.
>
>1. Get rid of bigevil.cf, it's mostly in be.surbl.org
>2. Get rid of be.surbl.org, it's in ws.surbl.org
>
Just a followup. Yup everything Jeff covered is spot on! However I see your
not running any of the 70_SARE_HTML rulesets. You should look into those.
They don't overlap either. And they are very good.
-_Chris
>-----Original Message-----
>From: Jeff Chan [mailto:jeffc@surbl.org]
>Sent: Wednesday, June 30, 2004 2:06 AM
>To: SURBL Discuss
>Subject: [SURBL-Discuss] Fwd: Re: blacklist brings system to halt
>
>
>Here's some additional info on some of the recent
>sa-blacklist/ws.surbl.org/6dos goings on, copied from a
>message I sent to the SpamAssassin-users list:
>
*snip*
>
>I think I can help explain why sa-blacklist went from 1.5 MB
>to 5.5 MB in size suddenly. Chris Santerre added a fairly large
>set of records from 6dos (6 degrees of spam) around that time in
>order to get the records into ws.surbl.org and sa-blacklist.
>Chris, Bill and I then discussed this and decided to take them
>back out of sa-blacklist and therefore ws.surbl.org, and put
>the 6dos entries into its own SURBL instead.
Yeah....ummm.....I made a little boo boo :)
This is what happens with 3 weeks off of hockey and only 3 hours sleep the
night before. I have been LART'd and it won't happen again.
"Bad little monkey! Bad!"
--Chris