Spam filters
I decided to build this page because spam
continues to be a growing problem. If you include your real email
address
when you post to public newsgroups, if you send more than a minimum
number
of emails or give out your address to more than just a few select
people,
or if you have your email address published on people-finder services
or
web pages you know what I mean.
Programs available:
- I tested Spam Killer back when it was free.
- The filter controls in this program are better than the
standard ones
in
my Netscape Messenger, allowing greater control over what is filtered.
It also comes standard with a large selection of filters, plus you can
get updates from McAfee on these filters making it a good tool for
ongoing
filtering.
- That's the good news; the bad news is you have to check Spam
Killer,
then
check your mail: now two places to check mail instead of just one.
- It also isn't perfect, lots of spam still gets through - and
some mail
that I wanted was getting stopped as spam.
- Ultra Spam Filter is software that works with most email
programs.
It's a separate program like Spam Killer.
- I tested a program called Spam
Inspector and found it to be about the best program around for
automaticaly
filtering spam. Instead of being a separate program it integrates right
into your email program. Program updates and filter updates are free
and
automatic. There are versions for Outlook, Outlook Express and AOL. The latest news on this product is the
company's been bought by Microsoft. The
filter rules it uses seem to be about 60% effective, plus it has a
couple
of features I like:
- It has a Bounce button. This sends an error message to the
spammer
which
hopefully will convince them to take you off their list.
- It has a pair of buttons called "Not Spam" and "Is Spam". Both
of these
provide feedback to the program's filter maintenance team including a
copy
of the email message in question so that they can further tweak their
filters.
- Another program that is similar to Spam Inspector is called I
Hate
Spam,
a very appropriate name in my opinion. I haven't tried it myself so I
can't
tell you how well the filters work but it integrates into the
Outlook/Outlook
Express email program like Spam Inspector, has a Bounce button but
lacks
the Not Spam and Is Spam buttons.
- SpamCop only works with Outlook.
- It works by retrieving your POP mail on their server and
filtering,
then
you get the mail that's left from your POP address on their system.
This
requires a change to the setup of your email program to point to the
SpamCop
mail server.
- Management of filtered messages requires logging onto a web
page on
their
server instead of messages being moved to your Deleted Items folder as
with Spam Inspector.
- It has one feature that is similar to Spam Inspector: the
feedback from
the users on identified spam messages is used to update their filters.
- Despammed.com works like SpamCop.
- It's free but ads are appended to your emails.
- Mozilla, the open-source version of Netscape, now has
automatic
filtering built into the email program by the user identifying emails
as
spam or not. Presumably a future version of Netscape will include this.
It's a free product.
- This uses a standalone filter set so there is no feedback to a
central
server anywhere, so
no hope of benefitting from other's filtering efforts. It
starts with no filters at all. You train it by clicking the Junk button
while you have a spam message open that it didn't identify as such and
by clicking the Not Junk button while you have a message open that's
not spam that it thought was spam. The way it automatically filters is
smart enough it's pretty effective in only a short
time, a couple weeks to a month is typical from what I have seen.
- I compared Mozilla to
Spam Inspector by having two computers get the same mail from the same
account for a few months. (One computer was set to remove the messages
from the server and the other one wasn't so it was the very same email
they were both receiving.) I found Spam Inspector was about 60%
effective from the beginning as it comes with a preloaded set of filter
rules. Mozilla was 0% effective at the start because it had no rules.
Over a period of a few weeks Mozilla learned and got to about 90 to 95%
accurate - and Spam Inspector was still only 60% accurate. I think the
reason SI is less accurate is because they're building their filters to
work best for everyone who uses their product. When you train the
filters yourself it can be very accurately tuned to your definition of
spam. The downside is you have to do it yourself, whereas with SI the
people who made the product take care of maintaining the filters for
you. If you feel you must stay with Outlook or Outlook Express then I'd
recommend this product to you but if you are willing to switch then
Mozilla is what I recommend.
- The type of automatic filtering Mozilla uses is called Bayesian;
read
more about it and efforts by Paul
Graham using it to block spam.
Services available:
- There are online filtering services that catch and delete the
spam on
your
server before it ever gets to you. This eliminates the two-step mail
check
issue with Spam Killer and similar programs, but for most of these it's
a monthly fee for the service so more costly. Many ISPs are now adding
spam and virus filtering to the mail servers, often at now extra cost.
I see this as a further indication of how much the spam problem is
growing.
My email and web hosting is through Blacksun,
and they have added virus scanning and spam filtering to their
accounts.
Nothing is perfect in the field of spam filtering but I have noticed a
dramatic reduction in received spam since this was added. (If you're
looking
for a good hosting service, I highly recommend them, and no, I don't
get
any monetary consideration from them for recommending them.)
- Atqui requires that a user who is attempting to send you
an
email
message verify themselves in an extra web page step. Spamming programs
are completely blocked by this approach.
- Works with any email program as no changes are made on the
user's
computer.
- Annual or bi-annual fee.
- Herbivore uses any mail client and existing accounts.
- Uses feedback system to see if a particular message has been
bulk sent
to lots of users
- Doesn't appear to have content filtering.
- Annual fee.
What I did before on my Netscape mail
program was just manually build
filters. I was always adding to this
because
spammers are always coming up with new wording for their spam partly
intentionally
to avoid filters and partly to find new ways to say what they're saying
in an effort to hook new people. I stopped updating these filters for
awhile, however the percentage and total number of spam messages I
receive has continued to rise so I have resorted to manually filtering
again. The Thunderbird crew says this is redundant when using their
automatic Baeysian filters, however I achieve a higher accuracy in
dealing with spam in my own filters than the automatic filters in
Thunderbird achieve, and since I receive between 7,000 and 10,000 emails total per
month I need all the accuracy I can get just to keep from being
overwhelmed. My manual spam filters also mark the message as read and
move it directly to my trash folder so I don't have to do a thing to
it, as opposed to always having to scroll through the spam folder, then
manually mark them all as read, select them all and move them all to
the trash folder with my delete key. I'm typically seeing at least 150
messages in my spam filter overnight, and sometimes it's twice that
these days. Every time I scroll through the messages in my spam folder
I'm looking for false positives (messages that aren't spam) and for
patterns that I can manually filter on. I generally add at least one
new entry to my existing spam filters per day. On only rare occasions
now I also add a filter rule to spam that has come into my inbox.
I discovered how well my manual filters work
recently when I was waiting for an email. My Thunderbird automatically
checks for and downloads messages every 10 minutes, but I was
impatient. I kept clicking my Get Mail button every 30 seconds or so.
The interesting thing is, every time I did that I would see Thunderbird
say it was downloading 10 or more messages. When it was done, there
were no new messages in my inbox or some other folder, nor were there
any in my spam folder. They were all spam being filtered by my manual
filters! I'd guess I see about one spam message in my inbox each day,
so that's about 30 per month. That means my manual filters and the
Thunderbird filters combined give me a spam filtering accuracy of
better than 99%.
- The single most effective filter I used to use was a NotMe
filter,
that is, if my email address isn't in the the To: or CC: fields then
it's
bulk email and junk it. I say 'used to use' because spammers seem to be
more and more using addressing methods that include my address -
including
some I've received that have my email address not only in the To: but
also
in the From: fields. This filter is general enough instead of deleting
the message I move it into my Junk folder. I still have
to deal with it, but my expectation is that it's spam when I look in
this
folder so it doesn't bug me as much as spam in my inbox. This filter I
have put at the end of everything else so it is the last rule invoked.
- Build filters to specifically keep group mailing list messages
I
want to receive. While I was building these filters I built various
folders
for the mailing list messages to go into, so the filters move them to
those
folders automatically. This means when I receive new mail lots of times
the messages get automatically moved to various folders from these
filters,
making it easy for me to prioritize my reading emails: worthwhile for
busy
people like me in any case. I also have various filters not for spam
but
for other messages I receive regularly such as ones coming from certain
email addresses. These also get placed into various folders for
organizing.
- My spam filters go first in my list so they are acted
upon first. Messages filtered here are
automatically marked as read and moved to the trash folder.
- I have a filter to delete incoming messages with a date older
than 30 days. For whatever reason, some bulk email messages have
very old
dates on them. I have even seen one or two with the impossible date of
1969! Perhaps the people sending the bulk email didn't set up their
software
correctly. I could set this rule to be much more current, since after
all
email usually is delivered anywhere in the world in seconds or perhaps
as much as an hour (with all the spam filtering and antivirus checking
that mail servers do these days) and I get my email daily, but I felt
someone
whom I wanted to receive email from might have a slow system clock or a
messed up computer configuration or something
so decided this would be the safest way. I used to have to manually
update the date this filter used every once in awhile, however starting
around version 2.0 of Thunderbird there is a choice to set it for a
certain age. I don't have to do anything with this one now. This one is
at the top of my filter list. I'd like to see Thunderbird also have the
ability to select on messages that come in with a similarly impossible
future date on them. I have found these are also, on all the ones I've
checked, 100% spam.
- I found the best is to build filters mostly based on the subject.
Build the subject filters specific enough to catch the spam but not
delete
messages you want to read. This is a compromise because spam subject
lines
are always changing. If your filter is too broad it could delete valid
non-spam messages too.
- Examples I have built filters for are: "Accept credit cards",
"Attention
all Gamblers", "Best Cigars Ever", "Credit When You Need It", "double
your
money in 30 days", "Earn Money", "fire your boss", "Get into that
summer
look", "high mortgage rate", "let us pay your bills", "Modeling Jobs
for
You", "pay less interest", "Quit your job", "Ready to Save Money?",
"Start
Earning Today", "This really worked for me", "viagra", "work at home"
and
"Y O U are a W I N N E R here !"
- I also have some filters based on sender's address. Outlook
Express
has a convenient function called Block Sender, but unfortunately very
few
spammers use the same email address twice so this is of limited use.
Spammers
also frequently use non-spammer's email addresses (I've had mine used
by
spammers, as evidenced by all the error reports I received from their
messages)
so it's better not to block some possible valid future messages from
the
real people associated with that address.
- Examples I filter on are: "Airline Tickets", "Be Dept
Free",
"Credit
Repair", "Great Deals", "HeyThisIs4u@","Investor Insights",
"makemoneyathome@",
"news@newafrica" and "smartbuys@"
- I had one filter for invalid addresses, such as ones
with no @
in them: an obviously fraudulent address meant not to provide the
ability
to communicate back to the sender. If the From: address doesn't contain
an @ it's deleted. Update to this filter: I discovered this
filter
also automatically deletes error reports from my mail server, which has
a return address of "Mail Delivery Subsystem". I've updated the filter
to delete if no @ AND it's not from my mail server. (It just serves to
show how you must take care when deleting with filters and also to use
them sparingly.)
- Finally, I had a few carefully chosen filters when certain
phrases are
contained in the body of the message. These are the filters
most
likely to delete valid messages on you, so if you decide to have any be
careful how you make them. (I have also found this filter to not work
now anyway: with Thunderbird 1.5 they added the ability to move each
message to a temporary file in order to allow antivirus programs to
catch a virus in an individual message rather than quarantine thw e
whoel inbox. For some reason this means filters on message body no
longer work. Hopefully this will work again in some future version of
Thunderbird, but I'm using version 2.0 right now and it' not working
yet.)
- Examples are: "COPY ANY DVD MOVIE" and "READ THIS E-MAIL TO
THE END!"
The way I build my filters was to look at spam
messages
that get through to my inbox and decide what if anything I can filter
on
that would be unique to spam and not to regular emails I might receive.
Then I copy/pasted that portion to my filters. Feel free to use the
examples
I've given above, just copy the portion within the quotation marks (but
not the quotation marks themselves) and paste them into your own
filters.
Unfortunately, I am always having to add more
filters.
It takes time away from other things I'd rather be doing. It's really a
matter of choice, to spend your time to make your own filters rather
than use Mozilla or spend your money paying someone else to write a
program or do the filters for me.
It's your choice how you deal with spam you receive. I don't have much
hope things will get better in the future with spam. Spamming has a
very low response
rate, but it's so incredibly cheap to send out literally millions of
emails
they will keep doing it. Even a response rate of .1% on a million
emails
sent is still 1,000 responses, for example. There are efforts in the
works
to make it illegal to spam, and indeed there are some laws on the books
in various places for just that. Unfortunately, it's costly to pursue
spammers
as they intentionally make it difficult to find them because of these
laws
and because they don't want to receive masses of emails from irate
people.
I expect there will be progress in tracking down and prosecuting
spammers
but they will never be eliminated.
An idea I had was that more incentive is needed
for the law enforcement agencies to pursue this more vigorously. My
idea
was that if spammers were required to pay for the spam they send
out, that would mean they'd have to be registered and thus easily found
if they weren't playing by the rules. If they had to pay to send, those
who didn't pay at all would be pursued for taxes not paid. I expect
this
would be an effective incentive: it's all about money, as they say. The
spammers would pay for their own enforcement. The down side is, as has
been the trend in taxing of other forms, it would spread. I don't think
it would take long for taxing of other non-spamming to be created, such
as regular emails ordinary people like you and I send, and that would
mean
I'd be paying money for spam control, this time without a choice.
Update, November, 2007
The number of spam emails have continued to rise
since I last updated this page. Pretty much everyone with an email
address is now receiving far more spam than legitimate email, and the
increase appears to have no end in sight. Most email servers filter
spam before the user even downloads it, and even Outlook Express has
built-in filtering. Some places in the world have anti-spam laws in
place, and there has actually been some court action. Unfortunately,
though, that seems to have at best only incrementally reduced the
meteroic rise in the number of spam messages.
I continue to be amazed by how many spam messages
are so obviously misleading. Before I do business with a company, I
check them out to convice myself they are legitmate. I would never do
business with a company online or otherwise that doesn't at least
appear to be completely honest and operating in good faith. I can't
imagine anyone else being any different. I will also never deal with a
company that mangles their subject line so badly I have a hard time
deciphering what it means, or a company that sends out messages that
are composed of a paragraph or two of random words strung together and
a link at the bottom. I'll never do business with a company who uses my
email address as their from address. I think I'm far from unique on
these issues either. The question then is, how can the spammers
ever hope to have anyone at all respond? Could it be that they are so
focused on getting past the filters arrayed against them that they lose
sight of the fact that no one will ever respond unless they appear
legitimate? In spite of my inablilty to find the logic in spam there
does appear to contintue to be people who respond, people who send in
their money only to receive junk products at best, people who keep the
spammers rolling in money and rolling out yet more spam.
I still have my manual filters and update them on
occasion to augment the Mozilla (now Thunderbird) Baeysian filtering.
Baeysian filtering is supposed to give accuracy at least as good as 95%
and I'm inclined to agree it's around there. That still means I have to
manually scan through the messages Thunderbird has marked as possible
spam in case something was put in there that isn't spam. My process for
this is to scroll through, deal with any that aren't spam, then mark
all read and delete them from the folder. That way each time I go to
check what's in the folder I know it's all new. Tbird does have an
option to automaically purge messages after a set period of time, and
if your spam volume is low enough that may be fine. For me, though, I
get probably 95% spam and 5% legit email: not small volumes so this
works best for me right now. Here is an example of my email. The other
day my Tbird stopped automatically retreiving emails for a period of
perhaps 4 hours. (Tbird had an automatic update window pop up and it
stopped everything until I came back to the computer and answered it.)
Tbird then started downloading emails. 683 of them. The accumulated
backlog of just 4 hours. When it was finished, I had 118 messages in my
spam folder (all correctly spam), 8 messages in my inbox and 4 or 5
scattered through various other folders (I have a lot of filters
besides my spam filters, that keep me organized by moving new emails
from various people or about certain subjecta into various folders).
The other 80% or so of the emails were all caught by my manual spam
filter rules, marked as read and deleted. I didn't have to do a thing
with them.
Before this incident I didn't realize how well my
manual filters still were working and wasn't doing a lot to keep them
up. I've since watched a little more closely as I go through my spam
folder. I usually look for patterns, something that is obviously never
going to be in a legitimate email, and add probably one new manual
filter rule per week. Yes, Tbird is already catching them, but fewer
messages I have to scroll through each day in the spam folder is still
time not spent manually dealing with this electronic plague called
spam. Actually, even if I'm breaking even - even if the time spent
updating my manual filters equals the time saved by having them - I
still feel like I'm Doing Something.
If you would like more information on how to configure mail filters,
here is a basic set of instructions I found on Verizon's web site:
http://www.gte.net/announcements/spam.asp