spam filters

http://www.paulgraham.com/wfks.html

This article talks about something that came up early in the quarter regarding spam. Why do spammers continue spamming when there have been so many filters created and efforts to put them to a stop? this article investigates this issue and the statistics that spammers do this because they are successful some of the time. Will there be a day which spammers completely give up sending spam? The writer of this article does not believe so.

Here is a more detailed mathematical look at spam filtering.

http://en.wikipedia.org/wiki/Bayesian_spam_filtering

A lot of the filters use Bayesian statistics to determine what is spam and what is not. both articles indicate how easy and how simple it is to effectively implement a spam filter. however, with the amount of spam in the billions, even 1% of spam that gets through is enough to be annoying.

Law makers have long been considering implementing e-mail to become a paid-service, much like how the snail-mail is these days.

It would deter spammers to send junk mail since it would have to come at a monetary cost. However, it would also angry the normal users of these services since it is free.

Perhaps if mail service providers could cap a limit of how many mails you could send /week or /month before sending mail, that could be a possible solution.

Another solution would to add something like one of those person-validator programs. This would perhaps stop unauthentic e-mails from sending out from a spam-bot or program that generates spam e-mails.

It is definitely not realistic to try to make people pay per email you send. That would be horrible and there's no way people would agree to it in this day and age. The person-validator programs seem to work the best as far as spam filters go. But no matter what way you use to try to prevent spam, the spammers are going to get through your filter. So... basically, just dont give out your email to non-legit things and get the best spam filter you can. Otherwise, I'm not sure if there's that much you can do about it.

@Eric Chang

For those propositions to work, we have to assume that the spammers are honest, legitimate people who can be trusted to pay for every email they send. However, spam is mainly sent through botnets. botnets are everyday computers infected with a virus which can take over a computer and send email or do processing calculations. An example of a legitimate botnet is Stanford's folding@home program which distributes simulating protein folding over thousands of computers. Illegitimate botnets are used to send spam, run DDOS attacks on websites and other illegal acts. Since the user does not know their computer is infected with a botnet virus, we can assume neither does the ISP or whoever is monitoring these email caps. therefore, the pay to email and email limit schemes would hinder legitimate operations while doing no harm to illegitimate ones

http://www.icsi.berkeley.edu/cgi-bin/pubs/publication.pl?ID=002358

Here's a link to Spamalytics, an interesting paper written by a team who infiltrated a botnet system in order to determine the answer to your exact question: WHY would anyone bother to send spam, when it's hard to believe it could be profitable?

The answer is that it really IS profitable, and if it weren't, nobody would do it. This particular paper explains how they intercepted spam from the infrastructure they "hacked" (obviously without the knowledge of the spam campaign's real owner), and inserted their own tag onto one of the botnets, to test how the system really worked by tracking their own mole, so to speak. They also set up their own website for the click-through ads in the spam to direct to, so they could track how many people really do click on spam links. For their fake pharmaceuticals website (which was emailed to about 350 million email accounts and actually reached about 10 million), 10 thousand people actually clicked on the link! Only 28 made a purchase (as if that weren't embarrassing enough), but the profit from those 28 people still is enough to pay for and justify the effort of the entire system.

As for paying to send emails, the amount it would need to cost to actually deter email spammers would never go over well with the 99.9% of the rest of us who aren't spammers, and wouldn't buy anything from a spam ad. I don't think it'll happen.

It can also be said that spammer continuously tries to find ways to send spam. Simple spam filters such as bayesian statistics look at the probability of words to determine whether if an email is spam or not spam. This is why sometimes important emails get send to the spam box. Some spammers even try to spam by using a picture containing the words to stop if from being filtered out. But more filter to detect that are there. And of course, because of this same reason, some spam do get through. For instance, if a spam only contains phone numbers, it has a chance of getting through. And I totally agree that making people pay per email is unreasonable because it would hinder the world. Our social networks maintained by the emails would breakdown big time.

As in the above statements, spam will always find out the way to avoid the filter most in case of bayesian. If network is independent, e.g., intranet. then it will be possible to filter the spam 100% since there is no spammer.

However, I believe the main reason for spamming is that there is not many sunken cost to start with and maintain. Once you have connection and computer and appropriate program which almost everyone can get at comparably cheaper price than other business, there is no sunken cost for sending the e-mail. The more you send, there is more profit. (theoretically) Also, there aren't that many personnel effort in the market for this business. Therefore, spamming will go forever if there isn't any law which prevent from spamming, e.g,. fee cost such that. Since spam cost is almost 0, there will be always market for spam without market value is 0.