Current USPTO patent examiner here. The most effective way to eliminate bad patents would be to give examiners more time, say double the time across the board to start. If an examiner can't find prior art in the little time they're given, and they have no other reasons to reject the application, they'll have to grant it. The amount of time was (basically) set in the 1970s based on data from the 1960s. There have been…
The whole process has a scaling issue build in, meaning you would need to scale both required time and people for the review process as time marches on.
The USPTO seems to be doing decently over the decades in terms of increasing the number of examiners, though lately I know they've had a lot of difficulty hiring.