This night I found that the level for the fiddle needed to be much higher than any of the other instruments, by about 20db, in order to sound well matched in the hall. It still sounded pretty much fine, but I remember it didn't sound as crisp as the mandolin. I thought "that's very strange, but I guess this fiddle's just peculiar" and moved on.

At the break someone noticed that the clip-on mic the fiddler was using [2] was set up wrong. As a cardiod microphone it mostly picks up what you point it at and it was on backwards, pointing out into space. Turning it around it sounded better, because it was now picking up almost entirely "fiddle" instead of "a little bit of fiddle and a lot of random stage noise and other instruments". Correctly picking up just the fiddle also meant that its fader no longer needed to be up way higher than all the others.

If I'd paid more attention to my sense of confusion I might have gotten the fiddle sounding better earlier. While I think I was right at first to put it aside as less important than some other things I still needed to set up, I should have come back to it later.

On the other hand, calibrating your sense of confusion is hard. If you're too attentive to it you can spend too much time tracking down weird coincidences; you don't want it to become like a warning detector that's always going off making you learn to ignore it. While I don't know whether mine currently has the right threshold, I'm going to try paying more attention to it, especially when running sound.

[1] The meter on the board shows the energy of the signal, but we perceive the same amount of sound energy at, say, 100hz to be quieter than at 500hz. There are conversion tables, and meters really ought do this conversion and show the appropriately weighted level. While this would be expensive in a traditional analog board, the new digital ones could do this easily.

[2] I think it was an Audio Technica PRO 35, a common mic for dance fiddlers.