SpamSolution.org

	SPAM
		Current legislative and private attempts to stop spam are either ineffective, or involve unacceptable tradeoffs. The key to solving the spam problem is recognizing the importance of e-mail authentication, the granting of permissions, and making its circumvention computationally infeasible.

Contact | Our Solution | FAQ | Software | About Us

October 30, 2003

Spam is being defined too narrowly

One of our criticisms about the whole spam problem is that it is usually defined as unsolicited commercial e-mail + porn. This is too narrow a definition. The most repugnant form of spam is now more correctly characterized as scams. PCWorld discusses this new form of spam--which Dan and I had predicted would take place back in February, 2003. You can read PCWorld's article here. Note they talk about digital signatures as a way to prevent it, but then say it's too complicated to make work. Why aren't they demanding software companies make it easier?

October 27, 2003

The insurmountable hurdles that filters face

In preparing a proposal for the SPAM 2004 conference, Trevor found an impossibility proof against spam filters, of any kind, for unrestricted e-mail. You can read the outline for his proposed talk here. For suitably restricted e-mail it appears that effective spam filters are computationally infeasible (which means they are effectively impossible over our lifetime, unless you had the luck of being struck by lighting several million times in a row). The key to this result is a combination of the large set of possible messages and the fact that we don't know the defining characteristics of spam. By searching for the characteristics of spam over such a large search space, the subset of all messages that signifies spam messages becomes impossible to cover. Oh, and this ignores the fact that what may be spam today, may not be spam tomorrow (i.e. the subset keeps changing). The paper discusses the hurdles that current filters face, the theoretical hurdles they face, and the future hurdles they will face. Together, it is easily concluded that filters face insurmountable hurdles.

October 27, 2003

Software

It has been a long time since I've updated you on the software situation. As you probably can tell, not much has been done on it for about a month now. I had a weak prototype working, but then had a hard crash that caused me to lose everything. I have since built up a more secure file system (ala RAID) so that this never happens again. Other personal situations have caused me to focus on them in the meantime, and a couple of papers that are promising keep edging this project out, but I hope to start back on it very soon.

October 27, 2003

News Coverage

The Pittsburgh Business Times interviewed us and their story on us can be read online here. The interviewer had a hard time believing that we were not attempting to make money. The piece is mostly about finding local businesses that have formed in response to the spam problem. Trevor is quoted extensively throughout the article, but very spotty as per our conversation. Overall the coverage is a wash; what is good isn't that good, and what's bad isn't that bad. The print version has our picture in it.

September 30. 2003

Spam filters

Apparently, people are beginning to notice serious problems with spam filters. PCWorld relates the problem of false positives in this story here. Dan has two messages that pose a problem for spam filters. One is a message from a college student writing home to her parents and the other is a scam on getting you to send money. The messages are so similar that spam filters tend to either accept both or reject both. You can read them here. Take a look at this site. People are comming to the same conclusion we are when it comes to Bayesian spam filters--they are defeatable. What's amazing is that they propose that no one respond, read, or bounce any spam messages. The answer to spam is to further restrict what we can do with these messages! Can somone PLEASE notice that authentication is the real problem here? Note: our published article on a real spam solution will only be available here starting tomorrow.

September 8. 2003

Implementation

Some of you have asked how reasonable it is for individuals to get web space for the challenge webforms like we have set up (you can visit one of them throurgh a link on our contact page). There is a free service offered by Lycos on their angelfire website. It even has a script all written out for you so you can build your own form. The challenge is missing, but some more investigating could be done to determine whether or not that is implementable yet. It seems that a perl script is being used, and there is a cgi-bin directory available, so if some one could investigate this and report back here that would be great.

September 8. 2003

Solution

We have moved the outline of our solution to the following page. It is also listed above in the link section.

September 8. 2003

Spam filters

We are continually amazed by the interest in spam filters. An indication of how this only extends the war (but does not end it), we present the following link to filterbusters. This site is dedicated to helping advertisers defeat spam filters. At the moment it isn't very sophisticated. Trevor surmises some more on filters and provides a recipe for defeating Bayesian filters here. It's pretty rudimentary right now, but as he has time, he will update it and try to automate its success (meaning even less work for spammers--because he believes this path is not a solution to the problem).

September 1, 2003

FirstMonday.org

The founders of SpamSolution.org, Trevor Tompkins and Dan Handley, recently published a paper in the September issue of the online journal FirstMonday. In it, they argue that Spam can be controlled and even eliminated with a relatively simple solution--contrary to what Bill Gates and other leading computer industry leaders have been saying. You can read the article here.

August 31, 2003

FAQ

August 31, 2003

Software For Users

Trevor has started work on a software plugin for Outlook that will make it easier to manage keys as outlined in the FirstMonday paper. There are plans for a Eudora version, but until the Outlook version is done, that remains on hold. Proficient programmers who would like to help can contact Trevor. The Outlook version is being programed in Delphi, but other programs could be in C or C++ if there are programmers who would like to join the task and prefer those languages.