Wrong.

When you change a song, you are injecting a new song into the code base. In theory that's all that should happen. But there is a lot more to it. In some cases pointers get changed and memory locations for things altered. That alone could cause this. Not to mention, earlier versions of pinball browser didn't calculate crc and you'd get a programming error on flash that you should ignore. It's possible there was a real programming error in this case.

Also due to the lack of crc, defaults weren't loaded causing the nvram to be out of sync with the code. That can also cause general memory corruption. My guess is that was the real culprit here.

This is definitely not a stern "bug". On non modified software update, the crc would be verified preventing bad code load. And the factory settings are restored on each update - which that alone was the solution to this problem.