"Ian" wrote in message
news:PZNhi.19327$3j1.15905@newsfe6-gui.ntli.net...
>
> "Vanguard" wrote ...
>> "Stef" wrote ...
>>> Can anyone recommend some basic anti spam software that merges
>>> painlessly with OE and that doesn't require emails to be checked at
>>> a different location or only allowed though when the sender put a
>>> code in or anything like that
>>> I've happily used Norton anti spam for 12 months but the
>>> subscription had run out and they'll only upgrade to their full
>>> security suite which I don't want
>>
>> SpamPal
>>
>> It only tags suspect mail as spam. It is up to you to define
>> whatever rules you want to do whatever you want on this tagged mail.
>> SpamPal identifies. You choose what to do with it.
>
> SpamPal works for me. I use it together with the Bayesian plugin, I
> still get the occasional one through, maybe 1 a week out of 200-300
> spam messages.
That's why I like that the primary function of SpamPal is the DNS
blacklists of known spam sources. I don't use the aggressive lists
(although SpamCop is usually considered above normal for aggressiveness
but I do use that one). SPEWS is *not* a spam filter for personal use
but instead to guage the spamminess of a domain. SORBS is way, way too
slow to update their lists. I use the Spamhaus SBL+XBL (which include
Composite Blocking List (CBL), and blitzed.org), NJABL, ORDB, and
SpamCop blacklists. Bayesian filtering should ALWAYS be the *last*
mechanism used to detect spam since it is a guessing scheme based on
word weigthing over a historical sample set experienced by just one
particular user. There are variations where you download a "community"
driven database, like Outlook's junk filter; however, the weighting is
based on a sample set that may not reflect the particular junk that you
happen to get. There are also voting schemes, like Cloudmark, but then
spam gets through until enough [trusted] community users have voted that
the mail is spam so it will get through if it hits you while still new
(i.e., not voted on yet or not enough votes yet), which is also a
problem with the DNS blacklists (and why you need Bayesian or another
scheme as a second filter).
I also use the MXblocking plug-in because I don't want mails sent from
dynamically IP addressed hosts. Those are the hosts that are infected
with trojan mailers. If someone wants to operate their own mail server
then let them get a static IP address (and also maintain the PTR records
so a reverse lookup shows they list the valid mail server hosts at their
domain).
I used to use the HTML-Modify plugin but recent versions of e-mail
client have an option to disable linked images. The plug-in was getting
old and not updated by its author so the detection of old ploys for
malware were no longer valid within the plug-in or already handled by
firewalls, anti-virus, or e-mail/browser clients. Spammers quite using
HTML, anyway, and most of anything that I see that leaks past my
filtering is always text.
I still use the UserLogfile plug-in because that gives me a plain-text
version of any e-mails that got tagged as spam and may get [permanently]
deleted within my e-mail client's rules. Sometimes a false positive
still occurs (and why Bayesian isn't perfect or why DNSBLs may point to
someone who just got their IP re-leased but got a spammer's prior IP
address) and it helps to have the text version as backup. Unfortunately
the author didn't provide for auto-expiration of old saved plain-text
copies of spam-tagged e-mails so I wrote up a .bat script to do that.
The author used to have a link to it on his site. I probably could use
robocopy from the Resource Kit or other 3rd party software to do the
expiration.
Of course, I still leave the spam filtering option enabled on the mail
server. There is no reason to waste downloading the spam and the CPU
cycles disk space required to interrogate the e-mails to find the spam
if the server can already do that upstream of my host. I consider any
e-mail provider that does not provide an option to DISABLE their
anti-spam filtering as a rude, uneducated, and egocentric service
provider. I may not want their filtering if it generates lots of false
positives or blocks from domains of my friends, so I will rely on my own
spam filtering if their's sucks. Gmail is one of those rude providers.
I was getting an average of 120 messages per day. About 3 or 4 might
leak past the DNSBLs and Bayesian filtering. Those were brand new spam
that wouldn't yet be on the blacklists, were being sent by zombied hosts
sending through static IP addressed mail servers, and happen to use
content that wasn't weight yet or enough in the Bayesian database to
catch them. Considering the blast of commercials on television,
including cable, I consider these few to be more than acceptable,
especially since no one is obviously getting the same effect at spam
filtering that also visits Usenet.
The next anti-spam scheme that I'd like to add would be greylisting
(
http://en.wikipedia.org/wiki/Greylisting) but that has to be performed
at the receiving mail server, not by a user's local e-mail client.
Rejecting delivery for, say, an hour lets the DNSBLs get updated in time
to detect the mail came from a spam source plus it would eliminate the
zombied hosts since they don't bother to queue up and resend their spam.
Obviously users should still be able to employ whitelisting to
circumvent greylisting for known senders or trusted domains. If a
sending mail server isn't willing to retry sending mails due to a
first-time rejection then I don't want it.
Don't even get me started on the a-holes that are irresponsible uses of
challenge-response mail providers or clients (and challenges sent by
client can be detected from those originating from mail servers that
issue them). If someone tosses their C-R challenge "turd" in my Inbox
then I will either not reply to them (because it is someone that is
going to get hurt by my lack of response) or I will reply to them so
they get the response and end up seeing the spam that there were
attempting to use me to filter out of their Inbox. Read
http://spamlinks.net/filter-cr.htm#issues-harmful.
anti spam sw? 
Linear Mode
