Absolutely agree. There's a reason Monaco isn't a research giant, even though it has a massive wealth-per-head-of-population.
The problems with sprawling Empires should be social and political. A sprawling empire should require management of splinter groups and suppression of revolt. It should require unpalatable political decisions like going to war with neighbours who you'd (rationally) be better off remaining at peace with, in order to create a common enemy. It would require a method of delegating some measure of autonomy to placate separatists (short of releasing them as tributaries/vassals, though that option should remain).
Given the obvious power advantage that size gives, diplomatic relations should also be affected. That neighbour that's growing, they're going to want to keep growing, and the only way they're going to be able to grow is to eat you. Neighbours of large nations should pre-empt expansionism and combine effectively. It's what Europe did to contain France in the early 19th Century.
A sprawling empire should require management of logistical elements as well. If all your food is at one end of your Galaxy-spanning empire, the planets at the far end should be paying more to have their food defecit shipped in than if it was coming from the neighbouring system. Consider how long it can take for a fast military ship to traverse an empire, then think how much longer a megatonne bulk grain carrier would take, and how many of them you need to move to provide for billions of people on a planet with no food production.
Commerce protection is also too easy. As the internal area covered by an Empire increases, the effort that is needed to combat piracy should also scale. Currently it doesn't, and when you get Gateways, it pretty much goes away entirely, because Trade uses the Gates, and, even if it didn't, your Trade protection projects through them anyway, so even firehose trade mains worth hundreds would be fully covered.
Admin Capacity vs Sprawl is a jejeune attempt to abstract dealing with these issues into a single balance with a single (core) means of managing it. And it is thoroughly unsatisfying. It's an understandable approach. The computational burden of determining the cost of every imported good (and all the other factors affecting and affected by Sprawl) would be immense. The AI would have to cope with the additional complexity, and it can't cope with the current level of economic detail.
Part of the problem with the way this approach has been implemented is that everything is linear. The hundredth system adds the same amount of Sprawl as the second. The 1000th Pop adds the same amount of Sprawl as the 10th. If non-linear progressions were used more (say the Sprawl increase per system added doubled every 10 systems added, and/or is multiplied by a factor, potentially also non-linear, based on the number of jumps from the Capital, or another administrative centre), the system would develop inflection points where the advantages of expanding would be outweighed by the advantages of keeping your numbers controlled and your lines of communication short.
Another area where the management of large empires could be made more problematic is in applying greater stability penalties for changes. At the moment, you can change a planet's Designation at will, for no cost, for example. Changing Sector structure is entirely without cost (or benefit, largely).
A third problem area comes with the management of diverse populations. This is largely irrelevant within the current paradigm and should be given greater emphasis.
Making empire management more nuanced would also allow more differentiation between government types. A centralised authoritarian Empire could grow to a larger natural size than a more laissez-faire one, perhaps, but the necessary stifling social straitjacket might cripple their research efforts. A liberal approach might improve research, but risk chunks of the empire falling away as independent states of varying friendliness. Monobloc empires like Hive and Machine Minds would have their own problems, possibly being extremely brittle/vulnerable to random problems of mutation (of control code, or genetics) that they struggle to cope with in a more sporadic and unpredictable fashion, while otherwise being more efficient. As ever, any choice should have consequences of future decision and/or direct effect.
Something that I personally think would add interest would be more random outcomes for standard things. Research as an example: when Sol III took on the research card "Fusion power", the date of completion was 30 years in the future. At various stages, it's failed random rolls, and it's still (50 years later) "at least a decade" (so, 30 years then) away. Very few people at the same time Fusion power research began would have predicted the meteoric speed of development of IT, which seems to have made every success roll at the first attempt apart perhaps from quantum computing. Social policy is not an exact science, nor economics. So any given socio-economic stimulus may have different results in subtly (i.e. below the game abstraction layer) different circumstances. However, gamers on the whole seem to despise random outcomes outside combat, even when they're the best abstraction available, perhaps because they can banjax carefully-planned min-maxing.
Edited: spelling. How come you always spot the typo as you hit "post"?
