The primary problem is funding. The USPTO does only a tiny fraction of searching for prior art when reviewing software patents. And god forbid they trust the applicants.
Several Apple patents are clearly shown to be prior art in videos from the 1970s Architecture Machine Group under Nicholas Negroponte, yet the USPTO appears to have ignored these. For example: page flipping as part of my Master's Thesis; The Spatial Data Management System (SDMS) 1978
Many other examples of current software patents can be seen to have appeared in the 1970s at MIT and other universities and research groups.
The USPTO gets pressed over and over by rich companies to award patents, and they "let the courts decide" with the results we see today.
The system is broken.