From reading the BEAST book and seeing various comments, I understand that the use of outgroups is not recommended, being kind of inconsistent with Bayesian phylogenetics. On the other hand, I see that outgroups are used very often, even in published papers about BEAST by the creators, where example datasets with outgroups are used for validation.
So, is there an updated perspective on outgroups? Or maybe is it just recommended that if you use one, you should also run the same analysis without one to see whether it has an effect? Just looking for some philosophical perspective on this.