The current campaign of Strategic Primer, my strategy game (which still needs more players) has required me to do a great deal of unanticipated research, and even with all that research still make entirely too much up out of thin air.
The game assumes a basically medieval level of technical and scientific knowledge, and there are a lot of details I don’t require the players to specify. For instance, I just ignore plows and scythes when considering farming, axes in woodcutting, shovels in digging and mining, etc.; otherwise, the list of advances the player knows about would be more than unwieldy even before the game begins.
But the other consequence of this assumed starting point is that the players can make lots of reasonable orders I hadn’t anticipated, or hadn’t thought they’d try yet, and I have to come up with results of those orders. Generally the first thing I try is searching for “medieval X efficiency” or “medieval X productivity,” where X is farming, mining, woodcutting, carpentry, or whatever else the player ordered. (I still can’t find any data on how productive mining was in the medieval era, or at least can’t find any data that puts it in terms of time and the number of workers, which is what I’m interested in.)
On the other hand, better to do this now in the campaign version and start the computer version with real data before I start the balancing process. And I’m learning all sorts of details that just might be tremendously useful in my writing. So if you run across some detail that I might find interesting or useful, let me know. (And if you’d like to join this campaign of the game, please contact me.)