How I reduced spam at home - Part 3

In the previous parts of this discussion, I had talked about various steps that I took to reduce spam that I was getting. I made note that they weren't necessarily for the average user, but for someone that was slightly more technically savvy.

In my case, one of the biggest things to consider was that I was on varying speed connections. If you have 120 messages and 100 of them are spam and 20 are messages you need to get to, you have a problem.
This problem on a fast connection is not that big a deal, you download them all and then sort it out on your machine.
But if you are on a slow connection, then downloading 100 extra messages costs you extra time/money.

So on a fast connection, you can focus more on blocking spam via your e-mail client on your machine.
On a slower connection you want to focus more on blocking the spam on the server so that you are downloading at a better signal to noise ratio and saving you time/money.

Blocking on the client
To block on the client side, you just need to install something that will go through your inbox and filter out the spam as accurately as possible. I will review these options shortly in another write-up. Examples of this are SpamBayes, Cloudmark's SpamNet, Norton's AntiSpam solution, and others.
You might want to consider what techniques would work the best, and usually some combination of them is ideal. In the end, the less you have to think about it, the better that gets the more spam you get.

Blocking on the server
Not everyone can block what is on their server... not everyone even fully grasps what the server is or where it is. This write-up is geared towards those that do know what a server is, what it is, and they have access to it.
In my case, I have a dedicated server at Pair.com.
I installed SpamAssassin on it and configured it to be fairly strict. I then added all of my known friendly e-mail addresses to the whitelist - which guaranteed that they would get through. Then I set the trigger value down so that it was a very picky filter - it saves out the "caught spam" for two days which I can review to see if it is getting anything it shouldn't be - if it is, then I add that address to the whitelist (you can add entire domains to the whitelist as well - which is fine if it is your company - bad if you add "yahoo.com").
I had RBL checking disabled for speed and had it doing its built in rules and then learning using Bayesian analysis as well.

What you really end up wanting to do is blocking on the server with fairly light rules (bring those 100 spam messages down to 25 messages that might be spam, might not be) and then do the rest on the client side.
That way you can limit the amount of false positives that occur (so your uncle sending you jokes doesn't get blocked) and yet still significantly reduce the volume of mail that you have to download.

Next up we will look into what the average user can do (mostly all client side unless your ISP does some spam scanning on the server side for you), as well as solutions for businesses of varying sizes.

There will also be upcoming discussions of how the various tools work - what methods they use, and how effective they are (if at all).

Posted by Eric at March 11, 2004 09:44 AM | TrackBack