I think the best argument for FSD would be safety. If at some point FSD could prevent a collision that would injure me or my family, it would be worth the full $8k several times over. But I don't know if it's demonstrated that yet over and above what Autopilot is capable of. At the moment I'm willing to think of the $2k as an opportunity cost payment for the opportunity to wait and see how the beta goes.
I don't have statistics to back this up - only anecdote. I've been in infrequent driving situations over the last couple of years where I felt that autopilot in my Model X was the better / safer driver than I was. All of these situations involved hard-to-see circumstances that might affect me more than average, but I believe would effect any/every driver to some important degree.
One involved daytime heavy rain + heavy water on the road driving, and dealing with the clouds / sheets of water being thrown up by other cars on the road. I was having trouble keeping track of the lane markers (freeway driving) while autopilot was rock solid.
The other, more common, involves nighttime driving on a state highway. Reasonably wide and well traveled state highway, but oncoming traffic would usually white out my vision down to the stripe on the outer edge of my lane. I drive this particular road reasonably often, and when the sun is down, I find myself relying more and more on my supervision of autopilot than on my own driving.
Those are both probably examples where the default driver assist in the car would do the lane keeping exactly as well, and thus are not necessarily arguments for paying for FSD (I don't know precisely what the default software will and won't do). These are, for me, reasons why I will pay for FSD regardless on the future Teslas we'll be buying in my household (I'm also more insensitive to the price than average as well - an argument in favor of not listening to me too much
).
Worth noting as well - it's also my observation that the car doesn't see as far down the road as I do. So the benefits aren't entirely one sided - I do a better job of judging overtake rates on the freeway and deciding when to get out of the left lane and when to stay (or vice versa). Make that ahead and behind - I see farther. Hopefully that will change with the big new update coming along.
