What I’ve observed is that when you have a complex game, you can’t always iterate over parts of it to make the whole game significantly better. Sometimes you need to throw out significant parts of the game and redo them, or possibly change the scope altogether. There’s no locally optimal place (for the specific mechanic) to get to from where the game is, that is also globally optimal (for the whole game). A good example of this is Into The Breach, which it turns out had a whole metagame layer that just didn’t work. It comprised 60% of the game. Once they threw it out, the design clicked.
This is a problem for the Paradox model of throwing a concept at the wall and iterating over it with patches and expansions. It worked for them in the past, but that may have been luck. Plus, those other games didn’t just come out as they are – they are whole iterations over older versions that made big revisions to core mechanics, but that doesn’t seem to be the Paradox shtick anymore.
Where will that be?