> They win a grant, but the actual research work is then done by postdoctoral researchers and grad students. They do the actual work, write it up, and pretty much everything else.
In this scenario, which does happen but is not the only way it works, the convention is that the student or postdoc is first author and the advisor is final author on the paper. That seems fair to me.
Order of authorship is quite dependent on discipline - in comp sci and engineering the convention is as you say. In other fields they prefer alphabetical order or other orders, as I understand.
Some journals have started requiring specific disclosures of activities that each individual carried out.
Interestingly for some of the professors, some publications' ethics codes don't consider merely obtaining funding as meeting the requirements for authorship of a paper. Of course they'll claim other activities, but a quick review of the papers and input they claim to have made will likely give rise to probable doubt. There's a few papers looking at "prolific authorship" in different disciplines.
In this scenario, which does happen but is not the only way it works, the convention is that the student or postdoc is first author and the advisor is final author on the paper. That seems fair to me.