Spammer at work.
One of the cool things about blogs—in comparison to regular Web sites—is the opportunity users are given to comment on the material. Instead of merely reading the marvelous things you have to say, readers can react to your ideas by sharing their own opinions and insights. When you see a trail of insightful comments on your blog, you know you've done something right—something that provokes discussion on the topic at hand.
Alas, your thoughtful readers aren't the only ones commenting on your blog. The spammers—human and otherwise—are out there too, and they get trickier all the time. One of the biggest challenges I face as a blogger is that of de-spamming comments that may or may not be legitimate.
SEO (Search Engine Optimization) is the primary reason you find spam in your blog comments. Web developers know that their search engine rankings will rise if they can get more sites to link to theirs. One of the easiest ways for them to do that is to put links to their sites in comments on blogs.
The search engines know this too, as do the software developers at Movable Type, Word Press, etc. In 2005 Google recommended adding rel="nofollow" to spam links to keep them from affecting the rankings, yet this doesn't alter spammer behavior and must be used judiciously in order to not penalize legitimate commenters. As one side comes up with a new way to deter the spammers, the other side comes up with a way to work around the technology. It's an ongoing battle.
Here at Case, SpamLookup and MT-Blacklist are installed to reduce the amount of spam that gets posted to your blog. If you go to the comments area on your blog administration page you'll be able to see the comments that were blocked so that you can either de-spam them or approve them.
These tools do a pretty good job. I find that about 90% of the time I can tell which comments should be deleted by just seeing the first line. But other comments sneak by and need a human appraisal.
This is the tricky part. I immediately get rid of any comments that begin with just a link, start with a nonsense phrase or are obviously pushing pornography, performance enhancement drugs or exciting new investment opportunities. The others I actually have to read. I used to approve anything that seemed vaguely on topic and appeared to be written by a human. But now the robots are getting more clever, some include key words and my name in comments that otherwise offer little value. Others may be written by humans, but appear rushed and unfocused, as though written by workers in a cube farm—who surf the net looking for new places to post their links. Perhaps some of the SEO firms employ such teams.
I could just go ahead and approve any comment that seems vaguely on point. I don't get so many of them that this would pose a challenge to my readers, but eventually I might. Thirty harmless comments could obfuscate the one or two thoughtful comments posted on a given entry.
I also want to preserve the credibility of this blog. If I approve every comment, with every link to anyone's Web site, then I serve to weaken my own authority. Links included in vague comments shouldn't be given the same "stamp of approval" as links in specific comments or links that I include in my entry.
So what should I do? I have a new batch of comments waiting to be approved or de-spammed and some of them fall into something of a gray area. I have to come up with a new way to determine who stays or goes, and if they stay, whether they get to keep their link. To start with I'm going to test the following checklist.
If a comment passes all checks I'll approve it. If it fails most checks I'll delete it, and if it falls somewhere in the middle I might just remove the link or add rel="nofollow". I'll test this out for the next few weeks and see how it goes.
What about you? How are you handling spam on your blogs? If you've come up with any best practices for how to handle this issue, we'd love to hear about them.
* I wonder how many ice fishing comments this will provoke.
Sorry, the comment form is closed at this time.