I'm not unsympathetic to this view point, but you are talking about a tiny, tiny fraction of the patents in the system. I'd WAG it at on one-thousandth of a percent, tops (while acknowledging the difficulty of measuring this).

So if the vast majority of the work being done is fundamentally different, does it make sense to set policy based on the outliers? The work of the patent office is almost entirely focused on "common ideas".

NB: I don't know the answer to this, just pointing out the difficulty.