[Tfug] ANTI-SPAM

[Leo Przybylski]

I did some more research on relays.osirusoft.com's software.

I started an evalutation of this software. It appears that it doesn't 
have to deny open relays. It can just 'tag' with headers the SPAM mail. 
With headers, the MTA can:
* Put the mail in a private SPAM folder
* Put the mail in a seperate SPAM mail spool.
* Post the SPAM to a private/public newsfeed

relays.osirusoft.com also contains a blacklist db from spamcop spamhaus 
and spews, so you don't have to build your own bl db.

Based-on my previous email, I believe it is possible to contact these 
open-relays to let them know when they are being used for spam delivery 
in order to limit the amount of SPAM traffic.

I believe I am going to set up the evaluation to store the mail in a 
privately accessible newsfeed. That way, if I don't look at it in a 
certain amount of time (in which it is probably not important to me), it 
will eventually expire and disappear into oblivion.

SPAM filtration systems are not 100% accurate and a lot has to do with 
tuning scripts. It seems even the case with open-relay filtration. Is 
there anyone getting 100% out there with any other SPAM filtering? I've 
read testimonials that say relays.osirusoft.com's rbl is accurate 90% of 
the time.

 From the article that Sam linked, I wonder if it is wise to deny SPAM 
at all. Seems that if you block spam with filtration of open-relay mail 
or otherwise, you still have a good chance of throwing out legitimate 
mail, so you are stuck having to let it all through anyway.

-Leo Przybylski

Sam Hart wrote:
> My biggest concerns have always been those raised in articles such as 
> this:
> http://theory.whirlycott.com/~phil/antispam/rbl-bad/rbl-bad.html
> 
> I'm not 100% convinced that simply blocking open-relays will solve the 
> solution any better. In fact, in the article above, the author mentions 
> this as I recall.
> 
> I'm aware that spam "costs" more to the recipient than to the sender, and 
> I'm aware of the fact that a per-mail filtration system like I employ will 
> cost you even more (cost in terms of processing power used)... But I still 
> am of the opinion that it's much better/less discriminatory than simply 
> using RBLs or blocking open relays.
> 
> 
>>  It has the right idea: to get as SPAM at the source. Everybody and his
>>mother can install a "spam filter" on their machine, but that doesn't change
>>the fact that SPAM robs everybody of bandwidth and personal time that
>>could be used for something more productive.
>>
>>  There is such a thing a "junk" snail mail and unsolicited commercial
>>telephone calls. They will never be as prolific as SPAM because there
>>is a media cost for each message. In contrast the price is the same
>>weather you send 100 SPAM's or 1,000,000 SPAM's.
>>
>>                                                     Cheers,
>>                                                     Erich
>>
>>
>>Sam Hart wrote:
>>
>>
>>>>Has anyone ever used this? http://relays.osirusoft.com/
>>>>
>>>
>>>Phew... that is one poorly designed page (I couldn't make heads or tails 
>>>out of it ;-)
>>>
>>>I personally haven't used it (or even heard of it), but FWIW, I've had a 
>>>great deal of luck with SpamAssassin + Vipul's Razor. My results can 
>>>actually be found online here:
>>>http://geekcomix.com/snh/spam.shtml
>>>
>>
>>>from there, you can see that since July 2002, I've gotten ~4800 items of 
>>
>>>spam, and only 83 items got past my filters.
>>>
>>>Also, I've been tabulating these results since July 2002, and I will make
>>>a graph of them, and probably do some numerical analysis (to see if my
>>>rate of spam has increased, decreased, and by how much ;-) So, anyone
>>>interested should check back in July 2003 for the results ;-)
>>>
>>>Anyway, appropriate URLs:
>>>http://spamassassin.taint.org/
>>>http://razor.sourceforge.net/
>>>
>>>begin  quote: On 03-02-28, Leo Przybylski wrote:
>>>
>>>
>>>>I am currently evaluating it. I wonder if anyone has any experience with 
>>>>it. I have a friend that swears by it, but I really need more/better 
>>>>testimonials.
> 
> 

-- 

-Leo Przybylski (leop at engr dot arizona dot edu)
http://foopan.nulldomain.net
http://grow.arizona.edu