It's really a matter of engineering judgment. The problem with using the non-rigid plate is that the FEM analysis tends to show a local stress riser in the bearing pressure right at the corner of the column.
Realistically, there will be some spread of the pressure through the thickness of the plate. But, the FEM analysis doesn't directly show that. Therefore, you are forced to eyeball the bearing pressure contours and "accept" them even though the program is showing a localized failure.
This is probably the most common frustration experienced by RISABase users.
Hopefully, the next release of the program will come up with a method for automating the spread of these forces in that localized region.