but it worked completely different. as an example: here is basically how the development of a plane worked:
the "Generalluftzeugmeister" in the "Reichsluftfahrtministerium" Ernst Udet and his staff did set the specifications they envisioned for a new plane and those were made more or less public. then all important companies (and often also unimportant and unrelated companies) started developing prototypes. those were tested by the Reichsluftfahrtministerium and the best were developed further. after a while the best one was picked up and the production was started often by other companies too because that was the only way to provide the necessary numbers.
if you accept that it is abstracted and that it is necessary that all countries have the same models the current system is fairly "realistic". just imagine you are ordering you industry to provide you with a better engine or whatever you need at that moment.
speer was only coordinating the efforts better but development was still mostly in the hand of the companies itself (expect the wunderwaffen under the SS). additionally having the three branches seperated too much is nonsense because in the end the same companies worked for all. MAN developed engines for tanks and uboots and planes but they did not necessarily build them all themselves or even developed all models. assigning one team is simply nonsense.
I think that your attempts at "defending" the game in its current form whenever you can are well meant but slightly misguided. The current research system is functional as support for a pure combat simulator, but it fails miserably in several ways, all of which are addressed by the proposed system.
When it comes to realism, there it fails both in terms of process and outcomes. In terms of outcomes, the current system ensures that at any given time all countries that actively participate in the conflict have exactly the same models of every item that they use. The obvious question is then what is the point in having a research system at all? To make sure that minors were left behind and that items that countries don't need and do not build would have been worse than corresponding items from other countries that do build them? And a process requiring a demi-god player with 20/20 hindsight to directly intervene to decide exactly when certain items will become available is hardly an ideal reflection of history.
Another problem, which I mentioned in a previous post, is that the AI has no way of correlating the techs in the tree with the goals it is trying to achieve, and thus has to rely on very explicit scripting to produce the in itself rather undesirable outcome above. A system that relates strategic goals to research goals is not only much closer to the historic process, but will also be a natural way for the AI to interact with the research system.
If you had read the details of what I wrote, you would see that the player would never have to take decisions at the level of developing a specific engine (as a side note, major German aircraft engines were generally designed by Daimler-Benz and BMW, not MAN). It also makes no sense to simulate bidding, prototyping, and the chain of subcontractors. However, having several light (Bf, Fw) or medium (Do, He, Ju) aircraft manufacturers clearly created more opportunities for a more diverse fleet of aircraft. In the general category of light aircraft, you will be able to get slightly different results by giving the primary responsibility to one or more of the teams available. Furthermore, in the light aircraft category you will develop technologies for fighters, close-support aircraft, carrier aircraft, etc, depending on the set strategic priorities, as well as the preferences of the design teams and their superiors.
In terms of the details of the German process, you miss a number of details (Udet was, for instance, not the contemporary of Milch), but this does not matter. The issue at hand is that the proposed system provides a very good general schematic for the decision flow in all countries, whether they had a market economy like the US, or were communist like the USSR. The challenge is to create the right level of abstraction, and the proposed system does just that.
That said, there is a number of aspects where one could develop various implementations. For instance, instead of splitting it up into convoys, strategic air power, etc, one one could have a single global strategic warfare goal, which could be assigned offensive and defensive priorities, and it could have naval and aerial subgoals. Thus, Germany could have a primarily defensive aerial strategic warfare goal, and an offensive naval strategic warfare goal, while Britain would have the opposite, with both having significant weights assigned to each, while the USSR probably could have very low weights assigned to strategic warfare in all forms. This would mean that Germany would be more likely to research subs than escorts, and interceptors rather than heavy bombers or long range fighters. The USSR, on the other hand, would probably focus on land combat, which the aircraft researchers would interpret as primarily tactical support aircraft and short-range fighter-bombers. The latter could, of course, be used as interceptors as well, but would be driven by different needs, and perhaps have slightly different technical emphasis and doctrines (In reality, for instance, the Soviet fighters had very poor high-altitude performance since their focus was battlefield support. Incidentally, this was also true for the Fw190, which ended up as their German equivalent.)
Regardless of how the details are implemented, however, the proposed system will create a more interesting and historically correct process, with a natural historic bias in terms of outcomes, and a way for the AI to correlate research with its strategic priorities. This would be a major advance compared with what HOI3 can offer.
