- Unsupervised Learning
- Posts
- New Anti-Spam Tactics
New Anti-Spam Tactics
So I’m trying out a new anti-spam combination:
If you have any problems with the CAPTCHA software, just reload the image or try the audio option. What I like most about the re-captcha system is that it doesn’t dump your post if you don’t get the CAPTCHA right. I tried another implementation a while back that did this, and got several emails threatening my safety.
Another cool thing about the re-captcha system is that every time you use it you’re helping a good cause. The re-captcha system is part of a project to digitize books, which is done using OCR software. Well, when the computer has a problem reading in a given piece of text it needs a human to help determine what the word is.
That’s what this system does — it’s using us to identify what the computer couldn’t do. And that just happens to be a good way to also keep those same words from being easily guessed by a computer. Or, to put it another way, if the words were easily guessed by computers (hence making a CAPTCHA text) they wouldn’t be in the system in the first place. Pretty cool.
Anyway, spam seemed to be picking up quite a bit recently and I needed to switch tactics due to the overhead of managing it. Let me know if this is unbearable, however, and I’ll adjust as necessary.