Maybe a social endorsement scheme could be an alternative to an invasive proof system.
problem is, how do you figure out someone's a bot? Ah well maybe someone will make an AI tool for that, we humans are too busy doing groceries and putting round cubes into square holes.
I bet we could get a major university (eg. Stanford) to help fund the initial deployment. I think we should call it a really big silly number, like "Gazillion" to emphasize how much it knows. Obviously this won't have ads, and it should make an explicit point to not be evil....
Hmm. I think I've heard this before.