What is needed is more evidence based software engineering. Statistical methods applied to datasets correlating issue trackers with code ASTs to show us exactly which ways of coding are correlated with longer issue times, frequent bugs etc.
I sometimes wonder if there could be an optimal number of microservices. As far as I know no one has connected issue data to the number of microservices before. Maybe there‘s an optimal number like „8“ which leads to lower number of bugs and faster resolution times.