When I do the further tests, I will do a quick check on each nation (egads this is going to be tedious
) and do a check on lvl 0 sites, to get an accurate reading of their gem income on turn 1, and compare that to income listed on turn 21.
I do agree that the impact of luck is going to be less (or less welcome!) in smaller games. Not sure I agree so much on the long games though, the point in the long games, is that the luck is hopefully going to help you build up and operate more efficiently in the early/mid game. Yes the luck scale, and associated events impact less on a turn for turn basis later in the game, but if you got enough gems by then to get 1 more wish, or to get a couple more artifacts before other players, or to build out your first SC with top shelf gear rather than "what we had kicking around in the lab", then luck did its part to give you a better late game.
That is, IF it IS doing its part. With 1 test run, the overall variability of the events is seeming to have too wide a margin. The disparity between the L3 and Misf3 nations in gold in that first test is ~6000g apiece, which at that point is 300/turn, and is undeniably huge (L3 nations had 40% more gold than Misf3), but the Misf 3 nations had ~30 more gems apiece, even after losing territories. I'm not saying that's a fair trade, but when you're trading 6000g for 30 gems AND 240 design points, it starts looking like something one could work with.
At any rate, my deeper testing will involve running 7 games of 21 nations in each (yes I am insane), each game will be one luck scale at a time, and within each game, will be 3 nations set to each of the possible magic scales. So, 3 nations at each combination of Luck +/-3 and Magic +/-3. It will still be far from conclusive data, but it will give a broader picture.
If I'm ever making games, I will make sure this sort of testing can be automated, to pinpoint balance issues with minimal effort, but I'm looking for a lottery ticket for that to happen.