For a while running this site was becoming a headache as the amount of comment spam I had to deal with began increasing day on day. At one point there were 100s a day coming in. Unlike email spam there's no truly reliable and effective solution to this problem. Although now I think I've arrived at a suitable compromise.

The most commonly used solution out there seems to be getting the user to prove they're a real person by typing the text they see in an obscured image in to the box below it. Personally I really don't like this approach. It annoy me no end whenever I have to use one. Not the principle of it as much as the way the image is so obscured it's sometimes impossible for me — a real person — to make sense of it.

If you have a blog and wat to encourage commenting from as many readers as possible, then you need to make it as painless as possible.

It's my choice to run this blog, so the burdens that come with it rest on my shoulders. Not yours. As such I've had to put in place ways of dealing with the tide of links posted to blogs without resorting to any kind of person-or-computer front-end logic. This started as a fully manual process of deletion but has since become a semi-automated process and 99% effective.

The "comment" form has two fields on it (well, obviously more than that, but...). One is "SaveOptions" and the other is called "Approved".

SaveOptions is computed and refers to an App Setting document which lists the "names" that have been banned. The majority of spam here used to come from people called "Cialis" and "Viagra". Maintaining this list and not even saving documents posted by them has cut out 75% of the problem.

The "Approved" field is also computed and equates to a boolean 1 or 0. One of the things this field does is trap comments to any blog entry older than 7 days. The vast majority of feedback on a blog entry comes either on the day it's posted or during the following couple of days. As soon as it's replaced by a more recent blog entry then the comments all but stop. On the flipside almost all spam is posted to old entries (maybe in the vain hope we either won't notice or don't care).

If the comment is approved it appears directly on the web page and I get an email telling me. If it's not approved for whatever reason then it won't appear directly on the web. Instead there's a message explaining that it's awaiting approval. The same email comes to me but it also highlights the fact it's unapproved and needs my attention.

This traps another 20% of the problem and at least means there's no chance of them getting any Google juice while I get round to deleting them. The fact that it doesn't appear on the web immediately should also act as a deterrent against bothering. Even if it did appear, all links in blog comments are rendered with the attribute rel="nofollow", which stops them getting the Google traffic they so crave anyway.

While this might trap the spam posts I still don't want them to be in store.nsf at all and so I make a point of clearing them all out. To do this I use a view in the Notes client called "everything". It's the view all databases should have, with columns including @Created, Remote_Addr, Name, Form etc. Here's what it looks like:

The reason I prefer to use the client-side view is mainly that deleting multiple documents in one go is much quicker than in the browser. There are patterns to spam. Most of it comes in spurts of 30 or so at once and happens in a matter of minutes — almost always overnight. This makes it a lot easier to deal with in one single swoop, especially using a client view. It's become one of my chores each morning when I wake up.

And a chore it can be. Although, having had plenty of time to get used to it and notice the pattern to how it happens, it's now just a sad fact of life on the web, but a lot easier to deal with using simple measures to block it and remove it. Whether it will ever be a fully automated system is another matter. One thing's for sure - I won't ever expect you to do it for me.