Quoted from Pin_Guy:U20 is the column driver, you typically don't see a ground switch fault with its failure; however this can happen if one of the column pulses out of U20 is sending a constant low level AND you have a closed switch in that column which would constantly drive the associated row low resulting in a ground short on the row with the closed switch.
The problem with the solution DNO is providing is that with no connector on J206, 207, 208, 209, and 212; it's just not possible for U20 to have ANY influence on the switch matrix, and impossible for it to cause a ground short.
In order to attempt to clear up confusion with this information I did some testing, I went ahead and took a red mini-grabber jumper and connected my column 2 pulse right to ground; I chose column 2 as column contains the always closed switch #24 (this switch is just a blocking diode on the coin door interface board) and grounding this would permanently drive row 4 low and should result in a Ground short on column 4.
pasted_image (resized).png
What this did was simulate a U20 failure causing a constant low output on the column driver, and I was rewarded with the suspected ground fault at Row 4 that DNO was referring to.
pasted_image (resized).png
Now if I remove J207, the ground short will go away as the failed column pulse cant leave the board; however if I remove J209, I've disconnected the playfield, but I still have a ground short since the column 2 pulse is also going out J212 and the constant low on row 4 is still there because it comes in on J212 from switch 24.
The takeaway from this is with all 5 matrix connectors removed, if the ground short remains then its not U20 and its not switch related, if it goes away then its likely not a U18 problem.