Both packages fit [generalized] linear (and nonlinear) mixed models. Many questions deal with both packages.
Is it useful to have all these separate tags?
The problem is that it seems that, e.g.,
[lmer] are often used almost interchangeably. And in general, such a deep tag nesting is a bit annoying; e.g., every Q with
[lmer] could/should also be tagged with
[lme4]+[mixed-models]+[r], i.e., this automatically eats up 4 out of 5 possible tags.
So one could consider mapping all functions to the respective packages. This would end up with
[lme4] with 906 Qs and
[nlme] with 294 Qs. Or perhaps even map everything together to
[nlme-lme4] (or something like that) with 1138 Qs.
Or is it a bad idea?