From the schematics it is clear that all these are connected to row 20 in the cable (I knew from earlier that it was the start key switch that was the problem, not the end).
I started by measuring between the diode kathode and pin 20 and got nothing. But the same was true for the working lines! Turns out there is an inaccuracy in the VS service manual - the diodes are before the key switches, not after, so it is the columns that are common to the diodes, not the rows.
Even if I tried pressing a key, my multimeter's connection checker didn't beep. I gave up and unscrewed the pcbs to get a closer look. It quickly became apparent where the error was. I also figured out why the connection checker didn't work: the switch pads have a tiny resistance. Also, I could see that both the top and bottom switches are in fact in the same place. Assumingly, the outer ring that makes the connection is pressed down slightly slanted, connecting two of the poles before the other two.
|What happened here?!|
Scraping away a bit of solder mask from the broken trace enabled me to solder a new wire to bypass the fault. It worked straight away.
|The bent spring in the middle black key|
|The bent spring|