If we assume that comm relays are involved, we can also argue that Romulans jam them to varying degrees of success near the RNZ. And if we assume local conditions are to blame, we can again argue that Romulans jamming is
a local condition, and the UFP just doesn't realize this; that the RNZ runs along old battle lines where local natural conditions happened to be particularly nasty and contributed to the outcomes of the fights; or that Romulans are not exactly jamming as such, but actually manipulating natural conditions, Diane Duane style, out of sheer nastiness.
But if we enter speculation territory at all, I'd give a lot of weight to the brilliant Okuda/Sternbach TNG TM idea of having subspace communications be instantaneous over a given range and suddenly snail-paced beyond that. The insta-range in the manual is given as 22 lightyears, but that might just be for a Galaxy
class transmitter (with quoted amazing output power, possibly greater than for any of the ship's weapons). If the range is shorter for TOS ships, then all the data can be trivially fitted into one and the same model.
That is, minor differences in the location of the ship related to the nearest buoy would give major effects at extreme instantaneous range - and in the frontier, ships might be forced to loiter at this extreme range quite regularly, after which the adventure of the week would immediately take them the crucial few lightdays beyond and, despite the best efforts of the navigator and the communications officer, create these observed delays of a few days. This isn't a result we could easily obtain if instantaneous communications had a range of hundreds or thousands of lightyears without relays.