raymond at prolocation
Apr 7, 2010, 7:12 AM
Post #13 of 18
(2035 views)
Permalink
|
Hi!
>> Setup a blacklist blocking ANY ip and you are ranked #1 in this test.
>> Its of no use at all IMHO.
> Yes, certainly, and I guess it was a loaded question of me to ask,
> because it was almost too obvious that I thought I was missing
> something. I don't think it's _completely_ useless though, because the
> two leaders are for the most part reputable and don't just block ANY
> IP.
> Raymond, I remember your name from reading at surbl.org (actually, you
> must go to www.surbl.org, heh). Is this the list you recommend?
I dont recommend any list. Its up to anyone to pick and mix. Some list
work for the other, some dont. My personal experience with the cuda list
is not good. That doesnt mean it will work for others. That wasnt the
story at all. You asked if this comparison is a benchmark, it is, but
doesnt say anything about FP rating or effectiveness so its just a
number. A rather useless number also IMHO.
> I somewhat agree with the comment about barracuda being accountable,
> but mxtools is apparently running the business side of things at
> SURBL, and for large systems you need to pay, so I would assume they
> want you to come back to renew.
Not interested to talk about that. You asked about comparison, not about
pricing or surbl or mxtools or whatever. Please check my e-mail from thats
used on the list. I post on personal title here. If you wanna post or talk
to mxtools please try another list ;) Its offtopic.
> I'm sure searching could probably eventually find something, but does
> anyone know or have links that show the surbl and barracuda motivation
> behind their service? Of course it is to block spam, but what does
> barracuda expect to gain by making it free, for example, where surbl
Simple, to stop spam? For SURBL, the main reason was to help people out.
And we do that and have done that for allmost 8 years now.
We moved after giving away free service, and years of free time. Yes a
blacklist doesnt compile just like that... That didnt turn out for the
long term so we switched models. All is explained on different lists and
if you like to know, check there. We basicly wanted to secure service
level and existance for over time.
There is a lot of talks about Cuda, ask the people from Spamhaus or
any some of the open source vendor who's code is used on the Cuda devices,
on the SA list are enough people who can help you out with getting some
clue whats going on. I wont poke around in that fire, you get burned
easilly. So this is all you get ;)
> An occasional FP is almost equally as subjective as anything else, and
> a difficult metric in which to place any comparison value.
>
> Would a more detailed and thorough comparison be any use? IOW, number
> of changes per day, how many total entries, etc, or is everyone okay
> with how it is currently?
Independent testing like the VB tests tell me much more.
http://www.virusbtn.com/virusbulletin/archive/2010/03/vb201003-vbspam-comparative
And yes that more or less the commercial products, but it shows also how
lits like SURBL perform. But also ratings of the large vendors. And the
FP levels.
Even the SA corpus, where we have a open ticket about the ratings isnt
currently ok with rating. Score is calculated over time but domains
expire. Its hard... retesting doesnt give a full solution for that either.
See:
https://issues.apache.org/SpamAssassin/show_bug.cgi?id=6386
There is much more to do when rating things then just do some simple
counting. The scoring in SA is done pretty well (compliments!) but even
there are gaps and holes like pointed out.
So its certainly not something you can put aside in just a 5 mins talk.
bye,
Raymond.
|
