Isn't GWRBI the GWRBI of metrics?
Seriously though, the ignoring of defense bothers me because they don't have to ignore it. They have metrics that show the percentage chance a batted ball is converted into an out. They could use this to allocate the WPA change on a play between pitcher and fielder. Say the batter's out on a batted ball that's an out 95% of the time, a pretty routine play. 95% of the change can be allocated to the pitcher, and the balance to the fielder. On the flip side, if the fielder makes an outstanding play on a ball that's an out only 5% of the time, then only 5% gets allocated to the pitcher and most of the credit goes to the fielder. As it is, all the credit goes to the pitcher either way.
Not sure why they don't do it. Maybe it isn't a huge difference in large samples, maybe the underlying % chance metrics aren't reliable themselves, maybe no one sees it as worth the time since it's not predictive. But it would be a better, more interesting statistic if they did.