There are many arguments to and fro about realism, complexity and so on. Without taking sides on these discussions, here is what I think makes this game "unnecessarily" complicated.
Having read dozens of AARs (yes they are really fun to read), EVERY good player that starts a game in 1936 playing say, Germany, immediately starts to plan around not what he will need in 1939 (Poland), now what he will need in 1940 (France), but what he will need in 1941 (Russia). This includes players who are the utmost supporters of realism in every aspect of the game. And this planning is by far the most complex part of the game.
Now I ask you: does anyone really believe that the German leadership was planning in 1936 how many Panzer divisions they would need in 1941 and how many brigades of medium tanks and engineers each division would need? And how many militia and military police they would need to occupy France and Poland, and which German and Polish provinces would need improved infrastructure to allow supplies for the invasion of The Soviet Union? I have nothing against those who enjoy this, but I l believe that such calculations needlessly complicate the game not to mention being highly unrealistic. This is called "gaming the system" where a player concentrates on beating the program instead of the enemy.
It is not obvious how to get around this. One way would be to automate the whole process of research and/or production to yield approximate historical levels, but this would not take away the knowledge that the player already has about what the status of production was in 1940 and other years. Perhaps adding a random factor to insert some unpredictability would help, but I am not sure.
Maybe total realism both historical and military is impossible, and maybe having a game allowing more historical variety would not be any more unrealistic than a game than ensures historical results by gaming the system. And maybe the former would be simpler. I don't know the answer.