Multi-pronged
attack on spam
Sify AntiSpam Appliance
- SAS using a slight variation of Bayesian Filtering
technique to filter around 3 million mails/day
from spam. SAS is able achieve 99.8% effectiveness
in dealing with spam. Let us examine what all
the techniques available for an individual user
to can spam.
Solution
for Individual Users :
- Use filtering capability of Mail Clients The
technique used in most of these new clients
like Mozilla is Bayesian filtering. This technique
uses mathematics and involves you to train the
software to recognize undesirable content based
on personal reading habits. If you use a Linux
Desktop, you will have the added advantage of
using powerful Unix program like procmail, a
recipe-based set of scripts that will route,
reject, forward, and modify your mail based
on criteria you specify. But the varieties of
the spam are so many, it is impossible to create
a filter for each and every type of Spam. Also,
it is very easy to bypass filters as filters
look for some fixed words or phrases. Most spammers
now cleverly alter the subject and body in ways,
no matter what filtering mechanism is built,
to find a way around it.
- Bayesian Filtering : This technique is increasingly
becoming popular. This involves training the
software on understanding what is spam and what
is not spam. Once the training is over, filtering
happens on statistical precedents. There are
a number of filters available that use the Bayesian
technique. Sify uses a product called "Sify
AntiSpam Appliance - SAS" using a slight variation
of this technique to filter around 3 million
mails/day from spam. SAS is able to achieve
99.8% effectiveness in dealing with spam. Again
if you are using an Open Source system like
Linux or any UNIX flavored OS, you have a wide
number of choices.
- Third-Party Black-hole Lists : These blacklists
work by publishing IP addresses that the spammers
use. This is a very clumsy way of blocking mails
and catches only a small percentage of Spam.
By the time recipients have got flooded with
spam, the spammers are already using a new set
of IP addresses.
- Whitelist Strategy : This is accept mails
only from known "good senders". When an unknown
sender sends you a mail, your system can send
a mail back asking for confirmation from the
sender. The mail gets delivered only when the
sender confirms. Most of the spam is generated
by automated program which cannot respond back
to such confirmations.
- Deleting : It is not that easy. What you may
not realize is that in order to deliver that
message to you, other people paid for it. According
to estimates, the cost of spam is estimated
at US $8 billion/year. All this results in higher
ISP fees, in lost business, delayed mails, lost
mails, website crash, etc.
- Complain about spam you get : Spammers want
you to Just Hit Delete if you somehow, inexplicably,
aren't interested in their advertisements. Because
if you complain to their Internet service provider,
you can get their access canceled, and they
know it. SpamCop is a convenient tool for sending
a spam complaint.
- Legislation : The US recently enacted an inept
'can spam' act that requires people to opt-out
of spam and, predictably, has had no impact
at all on the problem.
- Imposing cost : It seems to be the most promising
approach if it can be implemented. The reason
spam is becoming ubiquitous is that it is free.
It costs almost as little to send 10 million
messages as it costs to send 10. This way, even
the tiniest response rate makes it fantastically
profitable.
Realistically,
any solution to spam has to make it expensive
for the spammer. The key question is how to do
that.
Next Issue: Dealing
with Spam - Administrator Level.
Manvendra Bhanguli is Vice President, Software
and systems, Sify. He can be contacted at esbmarketing@sifycorp.com
Author: Manvendra Bhangui |
