Use different method for determining model uncertainty
Currently, we use a method that is called "jackknife" but in reality is somewhere in between what a true jackknife setup would be (removing one trace and re-inverting, repeat for next trace) and what a bootstrap setup would be (randomly grab a certain number of traces w/ replacement and re-invert, which means that traces could appear more than once). We could consider variations on these as alternatives to our current setup.