IPR’s automated spam filter

Since some of you have been writing us to wonder whether your comments are being taken down, I thought I would explain what is going on.

IPR is currently using an automated spam filtering system known as akismet. This system works most of the time, but every once in a while it tags a legitimate comment as spam or lets a spam comment through.

The rate of spam being sent to IPR has increased recently.

Our all time ratio since May 2008: 646,412 total spam to 137,805 non-spam comments
In the past year: 456,056 total spam to 37,129 non-spam comments
In the past 6 months: 391,324 total spam to 14,041 non-spam comments
In the past 60 days: 209,678 total spam to 4,695 non-spam comments
In September: 130,281 spam to 2,541 non-spam comments
So far in October: 22,785 spam, 604 non-spam
There have been several days when we got over 8,000 spam comments per day.

In the past, at least some of the time when I was not busy, I was able to keep up with looking through a couple of hundred spam comments a day and seeing if and when the system ensnared legitimate comments by mistake.

Now, with thousands of spam comments a day, that is simply not possible for myself or anyone else at IPR to do.

Thus, the only way we will know if you have a legitimate comment get mistakenly spam filtered is if you write us at [email protected] and let us know.

If you do that, there is a chance that someone at IPR will retrieve your comment (I do when I’m around; other people may or may not).

If you sign up to write for IPR you will be able to get your own comments out of spam without having to ask anyone to do it for you. You will also be able to edit your comments after you post them, another feature people have been asking for. And you will be able to post images and videos in the comments.

The only other alternative that I know of is to bring back the system whereby people had to register to post comments at IPR, which we had in 2008.

2009-01 1,807 spam vs 3,408 non-spam (first month we turned off comment registration)
2008-12 4 spam vs 6,099 non-spam
2008-11 49 spam vs 5,397 non-spam
2008-10 12 spam vs 3,999 non-spam
2008-09 12 spam vs 6,518 non-spam
2008-08 24 spam vs 3,391 non-spam
2008-07 16 spam vs 1,041 non-spam

One problem with comment registration was that sometimes people had difficulty registering (I had to try it several times myself before it worked).

If you have any ideas for improving our spam policies feel free to share them in the comments.

-paulie