Aside: this paper demonstrates an interesting use of Bayesian regression to get the full posterior distribution over plausible decay rates rather than just a single point estimate. See the last few pages.

Is this pretty normal in these sorts of papers?