Spam Filtering: January 2007 Archives

I mentioned previously how blogs and forums (fora) are converging. I've been running the excellent VBSeo mod on a couple of forums for the last few months and the SEO results have been more than satisfactory, however there's no such thing as a free lunch :) The downside is that the forum is being hit with trackbacks from splogs all the time. In the last 24 hours it got really silly, with splogs hitting it every couple of minutes. I've had to disable trackbacks until I can come up with a sane solution for filtering them. Yes they do go into a moderation queue, but the number of them is such that moderating that queue could become a full time job!

Rather than try to reinvent the wheel and in order to keep in tune with the Open Source / Technical flavour of Skycon.... I've decided to give the talk on email filtering using MailScanner... Since that sounds very uninspired I'll see if I can come up with a sexier title. How about: Mine's big enough! - Filtering 50,000 users' mail using open source technology It doesn't have a great ring to it... OR Phishers and phuckers Neither exactly scream "inspired!"

Sometimes the cure can be a lot worse than the disease. In the case of WikiPedia one of the issues was the amount of spam that was being added. While it was all being removed as quickly as it was added the editors have decided to add the "nofollow" tag to all outbound links. If you would like to see the discussion on their mailing list here it is. Some of the posts are interesting and informative. Others sound like they were written by petulant children. What does this mean? Basically the "nofollow" tag, which I've mentioned in the past, is used to signal to search engine robots, such as Yahoo's Slurp or Googlebot, that they should ignore the outbound link ie. not follow it for spidering. In practical terms it means that any links from WikiPedia pages to your site(s) will lose their SEO value. While I can appreciate how annoying spam links can be, this sort of Draconian solution is totally uncalled for. (In case you care, which if you are interested in SEO you probably should, all outbound links from comments etc., here use "rel=external") The net effect of this move is that WikiPedia will become a one way street. The "neighbour" effect will be all but removed. If a site is an authoritative source of information on a subject then people will link to it. It's only reasonable that an outbound link from that site should be given the weight that it deserves. So maybe we should all add "nofollow" to our links to them? :) At least that's what Marketing Pilgrim is suggesting and I'd be inclined to agree.. though I generally don't link much to WikiPedia anyway. (via)

Laura has put a gun to my head. I have to decide on a topic for my talk at Skycon and I've got about 24 hours to do it. For once I won't be the only representative from Blacknight presenting!! Paul, in a possible moment of madness, also agreed to give a talk. As he's a hardcore geek he'll probably be right at home, whereas I'm a little more wary. What should I give a talk on?? Bearing in mind that it's quite a technical conference, judging by the timetable, what would people recommend? I was contemplating doing something on email filtering. (The last talk I gave in UL was on SEO) Any thoughts?

Reverse DNS (PTR records) is extremely useful. However when your office PC(s) are infected with viruses and basically spamming the entire country it might be quite embarassing. A well known Irish company ran into this issue earlier this week when a PC or PCs in their office started sending out emails to hundreds of Irish companies (both real and imagined). Unfortunately for the aforementioned company their office IP includes their company name in its reverse DNS entry, so anyone who checks their mail logs could see not only which DSL provider's network was responsible, but also the actual company responsible. In the days of broadband you need to take responsibility for traffic coming from your PCs ie. patch them properly and use up to date anti-virus software!!

Phishers are getting dumber every day. Irish banks don't send emails to their clients that often and they definitely don't send them on Saturdays... The phisher who took advantage of an insecure server should really have waited until Monday to send out his phish And maybe they need to get some lessons in design as well... I love the way the email was put together using FrontPage!

Whether it's a forum, a blog or a web directory, it's going to get hit by spamvertising scum. The problem is how often.... The level of junk hitting any dynamic website is just silly. Even if you sanitise the form input, so that's it doesn't do you any actual harm you still have to wade through the emails and other rubbish that is generated. At least some people are still able to see the funny side of spam and comment spam. Blog spam and how to deal with it is hardly worth going over again.. I think it's been done to death by me and everyone else :) Forum spam... It depends on the software that you are using.. There is supposedly an Akismet implementation for most of the forum software these days.. I've never had the chance to use it, so I'm relying on SpamHaus instead. Any registration requests from IPs listed in XBL get blocked and I get a short PM telling me Web directories are a completely different kettle of fish... In the case of some I've almost given up moderating the links at the moment and am seriously considering replacing the backend completely... However most of the links directory software available these days does offer some level of builtin captcha. There are some very strong arguments against the use of captcha, but when you are managing a website you sometimes have to give way to pragmatism. If it wasn't for captcha etc., managing link directories would be unbearable!