It's "solved" in that we have a way forward to reduce the errors down to 0.00001% (a number I just made up). Throwing more compute/time/money at these problems seems to reduce that error number.