#1: Where should you make the decisions about the remaining things we want control over
Both at empire and planetary level, I suggest. Leave the decision to the player. In the early game, when you have only 2 or 3 planets, you want to squeeze the optimum out of them and full control on the micromanagement level (which is already reduced by the fact that we are degenerating buildings into a infrastructiometer). Later on with growing empire size, you will not be able to control every single planet, so you issue your orders on the imperial level and the CPU assigns them to planetary queues. You may, however, have a few "favourite" planets that you feel worthy of keeping your continued personal attention (e.g. your home world).
I'll just rehash the argument against that option (of both options simultaneously) : in a competitive context (multiplayer, overwhelming cheating AI), enabling micromanagement effectively force the player to do micromanagement, else he won't as efficient as he can be (hence being at a disadvantage).
So, while having both option may provide the most fun factor (empowering the player while not forcing him to do the chore of micromanagement), it is always a losing strategy to neglect micromanagement when you can do it.
A friendly advice, don't be so vindicative. Readers will assume you are upset and will stop reading carefully. Then they will "read between the lines" lots of things you never meant and give you a bad reputation. From there it will spiral up to the point where they won't even read you post, just reply to the sub-text they interpreted the last time they read what you wrote.
Try donning an affable composure, you will certainly be more successful than trying to shake those you don't agree with using short exclamations. Mind I'm not implying your post could be summed up with this short exclamation.
While avoiding excessive micromanagement is a noble goal, we should make sure we do not reduce the game to a primitive level. The game must have its complexity, particularly since only complexity allows different valid choices, so our goal must not be to take away complexity, but to make it manageable.
It is all a question of priority, preference and democracy. Right now, the abstration flavour is all the rage.
I think we should adopt a good idea from MoO2: "rally" lanes. Have your ships produced at a certain planet, and that is where it pops up when it is complete. You preset MoO2-type rally lanes that immediately send the new ship to a planet of your choice (so that you need not micro every new ship). You will normally direct most of your output to one planet where you rally your new ships in order to form a fleet. You can, however, decide to have two or three rally points in order to defend key spots or create multiple attacking armies.
I think this falls into my idea of macrotool that I'm currently pushing (this use case would use ship order before creation and maybe multiple destinations).http://www.freeorion.org/forum/viewtopic.php?t=733
Everything else does not make much sense anyway. A ship is not an abstract thing; it must be produced somewhere, even if distant planets help by supplying pre-mounted parts or sending workers and equipment. And I do not see why a ship that has been produced at the most distant edge of your empire should pop up at the other edge of your empire once it has been completed, but has to travel normally afterwards if for some reason it desires to fly back to its origin.
It seems to me that you definition of "making sense" rely on a realistic element. While I value reality as a coherent and intuitive system, here reality alone is deemed "irrelevant" (as Aquitaine put it, "the proper phrasing would be 'it's realistic, so it's irrelevant.'") There is surprising strategical consequences to global production, but throwing global production away may not be the only solution. Wait and see.
Rally paths will normally lead across your own systems. If your systems are torn apart by an invader attacking a system in-between, then it is time for manual intervention. It is the task of the imperator (that is you) to react to invasions appropiately. If your routes are attacked, it is an obvious reaction to redefine rally spots for isolated systems. If you want to automate that as well, then you can as well automate the whole game, in other words, players become superfluous.
Repeating myself: I'm wary that with an abstracted model, detection of such situations (empire thorn apart) will be up to more or less coherent detection algorithm. Then, the details of the situation (that the imperator should strive to resolve) will be determined by other more of less coherent algorithm.
A good interface can include the ability to select multiple systems with the mouse (dragging up a selection box similar to those you use when you want to scan an image and select the important part to scan from the preview), and then assign a safety classification to all selected systems at once. That will help you to update classifications as needed with only a few clicks.
It seems to me that classification and topology are closely related to groups
What is KISS?
Keep it simple, stupid. It's a rule of thumb to avoid overly complex system (complexity beyond what is required be interesting). IMO, it's become a mighty powerful stick around here, but it does keep risky and time/effort consuming features from creeping into the project.
EDIT: well, you should know about KISS now. I'm eager to see how drek will address your concerns, as I share a lot with you.