locksmith data appears to be very noisy

[solb]

This is on 362318f atop master. We're probably going to have to examine the data more closely and figure out what could be causing the variance.