I think an effective system would be using html5/flash that grabs an image and requires a user to do some movement with their mouse. i.e. select a house with a broken window, followed by a house with red windows and followed by a house with smoke coming out of it.
since it's flash/html5, I figure its going to be a lot harder to automate a way to break it.
since it's flash/html5, I figure its going to be a lot harder to automate a way to break it.