Yep definitely limited. Me() can be made much faster—I actually re wrote a brms stan program once to do it. This area needs some methods research!

In particular it would be interesting to discuss how in a Bayesian model the posterior distribution of the scores is different than the prior estimates.