>>56390864I imagine those chatbots still typed perfectly when acting indignant. The primitive subhumans actually get less coherent when upset. Mind you, I wouldn't call it a more human reaction, when it's more of a primitive ape reaction. Our most famous example on this board hasn't posted in almost a week, but you might remember how many mistakes he makes when badgered.
The part about questions was just to illustrate how effective that method is, compared to just repeating the letters captcha gives you. Bots and subhumans can solve captchas, but they'll need to be trained specifically to answer questions. Even something like "how many letters does the word 'dog' have" is more effective than captcha. And more importantly, you might find yourself getting better users that way, since how a site works can influence how its users behave.