In our case I don’t believe it was solar bitflips but rather wildly changing ionospheric conditions. I was primarily pointing out how you can have a software-only fix for problems like this.

Without going too far into the weeds, the fact that the receivers in question were reporting high accuracy under uncertainty is definitely a software bug in the receivers from my perspective. There was a different receiver with a completely different chipset in it on-site too that was experiencing similar issues but was reporting low accuracy. Without going into too much detail, I’ve got pretty good reasons to believe it wasn’t spoofing.