Fighting spam

This is not the first post I've done on the topic of fighting spam, and most likely, it isn't going to be the last either.

When I started out working on MVC Forum, I knew I was going to have to deal with spam and spambots. I also very quickly decided on not trying to make some intelligent anti-spam code myself. There's already a lot of smart (and not so smart) options out there.

The first anti-spam add-on for MVC Forum to use an external service, was the Akismet add-on. Akismet is actually more of a anti-comment-spam service, not really meant for anti-forum-spam.

So yesterday I decided on taking a look at some of the other options that exist, and came across first Stop Forum Spam and later BotScout. I couldn't find any .Net API for these services, so I decided on writing my own, and in the spirit of things, put it up as an open source project on Codeplex. So far I'm almost done with the Stop Forum Spam .Net API.

Common for both Stop Forum Spam and BotScout is that they don't try to determine whether or not the content posted is spam, they only look at the poster - the username, e-mail address and the IP address of the person/bot posting. The spam bots and anti-spam services are in a constant battle, and trying to recognize spam by looking at the content simply doesn't seem like the right approach!

I have one last anti-spam or anti-bot measure I want to include in MVC Forum, a DNS black list lookup. Before adding that, I need to figure out how to let 3rd party developers (myself included) add anti-bot measures when a new user signs up for an account. What I want is to let add-on developers put in their own CAPTCHAs, "Are you a human" etc.

Categories: MVC Forum - Akismet - Alpha 5 - Stop Forum Spam - BotScout - Anti-spam

Posted by Steen F. Tøttrup on 19 February 2012 07:42. There are 2 comments.

Comments

  1. 29 March 2012 19:44 by Rezalas:

    One aspect of verification I've been considering is that if trivia questions stored inside the CAPTCHA. The idea would be that instead of having someone repeat data to you like a parrot, you have them answer an easily known or referencable question not easily answered by a bot (ie not math questions). For instance, "who is the host of Jeopardy?" can be found easily by a human, but the level of complexity required to analyze the question properly and produce an accurate result in an automated way is extreme (at least in my own view). What do you think?

  2. 17 April 2012 19:22 by Steen F. Tøttrup:

    That's a really good idea, the challenge here would be to compile a dictionary with questions and answers that all the visitors can answer.

Write new comment