Two thoughts on this point. The first is that although it makes a sort of sense we do see that Stardates are consistent despite location with Enterprise D resetting its clock with the correct stardate more than once. ( in fairness the use of this indicates that stardates DO tend to drift )
The other is that we also see inconsistent stardates for the same locations on occasion. For example the DS9 episode Dax has a latter stardate than most dated stories later in season one, and the station maintains its position.
Again -- differences in interpretation. This is all fiction, a bunch of stories made up by different people; there is no consistent underlying reality that we can reconstruct from the evidence. The only thing that, say, TNG's treatment of stardates is evidence for is how the script coordinator for that show (since that was the person in charge of assigning stardates) chose to go about assigning the essentially meaningless numbers. The reason there's no uniformity to stardates is because they're just not important enough a detail to worry about.
After all, 99% of viewers couldn't care less about how stardates work. So the writers treat the stardates as if they're a consistent, universal calendar scheme because that's what's convenient for the story; anything else would be a distraction. The stuff about stardates differing from place to place is just a sop for that tiny percentage of us who bother to notice the discrepancies in the scheme. It's not supposed to be an integral part of the universe, not supposed to be something included in the stories -- it's just an excuse to get us obsessively nitpicky types off their backs.
True enough and from a production POV absolutely correct. Taking that viewpoint I guess the answer to the threads question could stardates make sense is Of course not, nor were they supposed to ( though Ironically the JJ Stardates can and do make sense).
It also makes it very obvious that Roddenberry's statement about stardates was nothing more than fudging ( which to be fair is hardly a revelation).
That said it IS fun to speculate on what we see, and if we do then we have to weigh up all the evidence not discount the bits that are inconvenient in favour of throwaway statements made by the production team. - admittedly we will likely tie ourselves in knots and not find a consistent answer.
Generally there is a trend for Stardates to go up - though they are hardly consistent ( indeed even the progression of stardates in individual episodes relating to passage of time are at varying rates in TNG for example sometimes 1unit is a day, sometimes it seems to clearly be the same day) And in TNG they are all over the place ( and show far more continuity problems if the episode order is rearranged into Stardate order).
We do know that starfleet synchronises them, Which would seem to be onscreen evidence that discounts Roddenberry's fudging.