The picture doesn't have anything to do with the topic today. I just wanted to get your attention.

Recently, I've noticed that several spam-busting solutions on other blogs were eating my completely legitimate comments, mostly because I was a "new commenter", which presumably meant that I didn't have enough "good karma" with the spam-solution in question. Normally this would be only a very minor inconvenience at most, but when the CAPTCHA image failed to load in the first place (due to server load, I imagine), I had to rely on the good graces of the blog owner to check through his/her moderated comments manually.

Now, considering the Akismet stats on the amount of spam out there (I don't use Akismet, although not due to anything other than laziness; the current spam solution for Moe Check is Spam Karma 2, for lack of anything better than default), I don't really blame the spam-busters for being overly paranoid. I mean, look at the graph: all that orange is spam, compared to the blue of legit comments, small and lonely. I've been told that this is normal, which makes it even worse, since we have apparently progressed to the point where only one out of ten comments, if that, is legit, and nobody believes that this is odd. I'm well aware that each spam comment is essentially free, or at least has insignificant and trivial cost to the spammer, but all the effort expended in trying to block spam is beginning to feel like trying to inhale a hurricane. It's depressing, and it's even more depressing that I appear to be one of the very few who even think that it's depressing instead of Life As Normal. Zetsuboushita, and all that.

Moe Check receives something along the lines of fifty spam comments a day (down from a hundred a day), which isn't anything compared to the really popular blogs out there, which I've heard get something like a thousand spam comments every day. This means that it's still (barely) possible for me to go through each and every one of the moderated comments every day, checking that no legit comments got accidentally eaten. I worry every night that I've missed something, and some unfortunate commenter out there is wondering why his or her comment hasn't shown up yet, and will never show up.

And if my spam-buster solution is messing up somehow and eating legit comments, it's not like I'd know until I checked. Even if it does happen, it's not like I know what to do to allow these legit comments without working even harder to moderate away all the spam which slips through.

I'm not even going to touch the issue of scrapers and splogs, which I don't have a clue on how to deal with.

In any case, based on observations on how Spam Karma 2 works, I've come up with a few ideas on how to make sure that your comment won't get flagged as spam, either by automation or manually:

• Use proper English. This is extremely important. If English isn't possible, I also accept comments in Chinese, Malay, and Japanese. Please check your spelling, grammar, and punctuation; the worse it is, the more likely I am to conclude that it is spam. On a related note, if you comment in a language I don't understand (ie anything other than the four listed: English, Chinese, Malay, Japanese), I can't know what it is, and I'll err on the side of paranoia.
• Make your comment relevant. "I agree" and "I dunno" and "Interesting", with no other references, are one of the favourite tools of spammers. At the very least, say something about the content of the post in question. Just quoting the title is almost a sure sign of a spammer.
• Don't list Geocities as your website host. This one is Spam Karma 2's problem: Geocities is very blacklisted, possibly because of the large number of ad sites hosted there. Listing a Geocities site pretty much guarantees that your comment will be marked as spam, through no fault of your own.
• The fewer URLs per comment, the better. Spam Karma 2 deducts points based on the number of URLs in a comment. Three is pretty much the maximum before Spam Karma 2 gets too suspicious. Five is right out.

It's a sort of Catch-22 situation, really: if my spam-buster is eating comments, I'd like to know, but for me to know, the quickest way is for someone to comment, and those affected can't comment due to the spam-buster eating said comments.

EDIT: In what is probably an extreme case of irony, I am locking pings on this post, because in the course of half an hour, I received no less than five pingbacks from various scrapers and splogs. These are pretty easy to weed out, since they invariably go "BlahBlahBlah wrote an interesting post today, here's an excerpt", and then get the blog name wrong.

If I get spam comments on this post anyway, I'll be locking those too.