Balance would be nice, yes, but I think the conservative approach is closer to correct, especially given the natural human bias toward believing sensational theories.

Maybe not closer to correct, but definitely less likely to admit errors. But sometimes the negative space around a particular thing becomes overwhelming. To me this is like circumstantial evidence—in general it’s weaker than physical evidence, but in high enough numbers it can serve.

That's an interesting thought. I wonder if you can quantify this belief? That Weibull (presumably) distribution would be an interesting and useful thing to know.