So, in your scenario, a Legendary Miller might work longer than a Dabbling one?
Sorry if I was unclear. No, the Legendary will still be able to do the job faster/better than a Proficient, just not noticeably so, and certainly not with enough gain in speed and/or quality to justify the time & effort of his working all the way up to Legendary. A lowly Dabbler, meanwhile, is going to waste time making mistakes the first few attempts (possibly even ruining a bag of grain or two), but getting the hang of it soon enough.
While having titles for skills levels not have a great importance for extensive fields (that is, fields easy to learn but with diminishing returns) might be seen as useless, in the future, such evidence of experience might give them political and social clout (for exemple, a Legendary Miller might be elected head of a Millers' guild).
Moreover, it will make a majority of the population work in the fields, which was the situation before the Agricultural Revolution of the 1780s.
Precisely. Of course, I'm not trying to force people to play their game in a certain way, but it just feels wrong for such a (generally) realistic game to do a major disservice to what's been all of humanity's predominant occupation ever since the dawn of civilization. I'm all for procedurally generated cultural differences between one civ & the next, but honestly there doesn't seem to be much room for variance here. Besides, it's difficult to take pride in your supposedly "productive, successful" fortress when every time you scroll through the Units list, 2/3rds of them are No Job or doing leisure activities.
And if differences in soil productivity are introduced, making sand nearly worthless while black soil very productive then we could have even more differences, with sites basing their economies on exporting food (see Ukraine) while some sites could have to devote themselves to craftwork in order to have more food (Switzerland).