[John_Shaft]

Steadi, first of all I commend you on a great effort to defeat the (to me) tremendously growing problem of spam wrecking forums.

Originally Posted by steadicamop

I can't comment on this hack as I haven't installed or tested it - although I feel that a text based version could be defeated - this is why I created this one, using images that only real people can interpret.

Originally Posted by steadicamop

I don't see how a peice of software can read what an image is - yes it can understand text but how would it know that it's a picture of an aeroplane, or a car, or a person, or a banana .....

Ah, but there's the rub. _It doesn't have to_. A real person isn't at all needed to defeat it.

I was thinking of installing this (as I need a solution from somewhere) but I see an inherent problem with this that will surely allow it to be defeated with absolute ease, should it ever reach the popularity that it's worth spending a little time (and that's all it would take) to deal with it.

Quite simply, it doesn't need to be programmed to read the image or anything complicated like that. It has four pictures of which it has to click the right one to proceed. The easy way to beat it is to code the bot to just click on images, and go back and repeatedly click on images, until it reaches the accepted part of the page (birthdate or whatever). As there's only four images it's going to do it in a handfull of tries.

Whether they get around to coding that I don't know, but if they do it's instantly sunk as far as I can see.

The one way I can see to deal with bots doing that is to have the user type in the name of a single image (i.e. "cloud"). That's why captcha is a tougher problem (even though it's beaten atm) because there's a massive amount of inputs that need to be tried, rather than just "pick a number, 1 to 4".

Has this occured to you, or do you just believe they won't specifically target your hack, even if it gets popular?