Have fun trying to stay behind. I don't understand people's panic over using or not using radar. The important point is that the systems (AP, FSD etc) should work well, whatever they use. Today none of those systems (radar or vision only) works well, they each have problems. Since Tesla will probably not spend energy developing and enhancing two separate code stacks to do the same thing, staying on radar means you're stuck with a bad implementation that will never evolve. Jumping on the vision stack at least means you get enhancements when they are developed. Maybe eventually it will work well, who knows. At least you're moving forward.
I'm surprised they waited this long. Probably because vision was tougher than they anticipated and they didn't want us to regress too much. Vision is probably at a point close to radar now so they switch.