I suppose field linguists have the same problem. In the last few hundred years of documenting rare & endangered languages, most languages started out with significant regional variations, which have been rapidly disappearing, and sometimes the whole language. So reference grammars of any language would be a snapshot in time of actual use, or in the case of conlangs, a snapsot in time of a creators ideas at the moment.
I think Volapuk had only one native speaker. If it hadn't played such a significant role in the history of Esperanto, I think everyone would have forgotten about it by now.