My understanding was based on the name of the columns.

I thought they would be self-evident, but on reflection, I probably should not have assumed as much. I have a bad habit of doing this. I just get wrapped up in the problem, and don't think about it from an outside perspective. I have updated the original post. Hopefully the further explanation is helpful.

To clarify, what I meant was . . .

Indeed, this appears to be a nicely normalized design. I can understand what you were thinking when looking at my bad examples. This isn't quite the situation I was discussing, though. I hinted at that when I said The data column's meaning changes depending on the value of the type column, but I can see how my post could have been misunderstood. Let me know if the update explains things better, or if there is still something missing.