Land revenue granularity

When the domain is first secured, roll 3d3. The total rolled should be noted as the domain’s land revenue in gp per peasant family per month.

Does the above mean that you apply a single land revenue roll to an entire domain?

If so, what’s the logic behind it? It seems that using one roll for the domain (rather than a separate roll for each hex) means that you can clear three adjacent hexes and the resulting domain can have three completely different land revenue values depending on which hex you build your stronghold in, which doesn’t seem right to me. Just find the best hex in an area, build a stronghold there, and then expand without regard for whether you’re expanding into good or poor land, since they all inherit the value of your first hex.

If not, what’s the intended method for keeping track of how much of the domain’s population is in each hex and how much power does the domain’s ruler have for moving families from one hex to another? Are you allowed to just fill the best hex to capacity, then the next, and so on? If your starting hex is land revenue 7 and you later find a land revenue 9 hex, can you tell all your peasants to pack up and move there?

I usually roll land value per hex, but I could definitely see this becoming unviable for larger domains, at which point I’d probably just take an average of n 3d3 rolls (which will quickly tend toward the mean).

From what I recall reading in other threads on the topic (I am not an Autarch and this is not an official answer), the intent is that you roll land value when you don’t know what the land value is.

If, for example, the players scout one hex and get a 3, then go into an adjacent hex, it is perfectly normal and acceptable to declare that the land value of this adjacent hex is 1d4+2, or you could simply assign a number to it, or you could just use one of the many options to give it +/- 1 from the known value.

Similarly, if you’ve already decided that a particular hex is especially rich or barren, you can just slap a number on it.

Since close hexes will tend to have similar terrain types, they will tend to have similar land values, and so the land value applies to the whole domain. It’s a simplification, and if you want to, you can count per hex and track population per hex and so on and so forth but as you correctly point out, that’s a lot of work and I don’t think you get a whole lot of benefit out of it. I prefer simply adjusting land values such that you don’t get a totally random value for every different hex.