The designer behind Bubumiru had just as much intention to have his environment be useful as the designer behind Dunes. The intention to provide parallel leveling places is already present, has been since the time of MUDs. We know exactly how effective it is.
If you make zones similar enough to be functionally equal, you've made them copies in all but name. If you make them different, they become victims to the Path of Least Resistance. Just saying "they need to do it better!" isn't enough. This philosophy has been trying to have it's cake and eat it too for so long it's old enough to drink, drive and die for it's country.
A game shouldn't be trying so hard to make people go against the grain of human nature. If we can observe people following a largely herd-like path through zones in spite of their options, the answer isn't to fight against it tooth and nail. Whatever the solution to the problem may be, it need to embrace what players naturally already do.
The design of the game can have a huge effect here.
For example, the easier travel is, the less it matters which specific zone you go to, and thus the more spread out people will be; conversely, the more time-consuming travel is, the more location matters.
The more group-centric a game is, the better it is (to an extent) to go where everyone else is; the more solo-friendly it is, the less it matters where everyone else is (and at some level, the better it is to go where no one else is.
So in a solo-friendly fast-travel game (like WoW), people will tend to level in zones they simply LIKE better, mostly regardless of what everyone else is doing, while in a group-centric slow-travel game (like FFXI), the bulk of the population in a given level range will be in one specific zone, because that's where everyone goes.
And that's before including the effects of other FFXI design decisions that reinforce the community's herd mentality, such as the need to research camp spots in less popular or totally unused zones (because camps need to be relatively safe to allow resting, but with enough mobs close by to sustain the party) and the wildly unbalanced mob difficulty (for example, imps and colibri being significantly easier than other mobs of the same level, or eruca and spiders being far more capable of wiping a group than crabs and crawlers).
For example, with regards to Valkurm vs. Buburimu, the design deck was stacked against Buburimu pretty much from the get-go. Valkurm is fed by two nations, compared to one for Buburimu, so there's already going to be more people there. Before the level requirement for outpost warps were lowered, the fastest way to get between the two zones (barring higher-level help) takes 20-40 minutes, placing great value on picking one over the other. The placement of guilds and vendors makes it harder to gear a tank in Windurst compared to the other nations, and since parties need tanks (and the party's success is affected more by the tank's gear than anyone else's), this is another strike against Buburimu. So by time the player base was in a position where partying Buburimu was less of a hassle, they were already in the habit of going to Valkurm by default or were using better options.
If San d'Oria and Bastok led to separate Valkurm-level zones, if Windurst had a blacksmithing guild, if you could outpost warp to those zones at level 10 from the get go, if the game had stressed unlocking outpost warps to new players, if it were easier to obtain outpost warps (instead of needing to perform a supply run during a week you nation controls the zone), if the boat ride was shorter and the boats came more often, (heck, if the chocobo license was obtained at 15 in your home nation rather than at 20 in Jeuno) the 10-18 levelling experience would have turned out very different.
Newbie zone -> Valkurm -> Qufim -> Khazam -> Garlaige -> CN -> Kuftal -> Boyahda -> Terrigan -> Onzozo -> Boyahda -> Ro'Maeve (or later Newbie zone -> Valkurm -> Qufim -> Khazam -> Garlaige -> CN -> Kuftal -> Bhaflau/Wajaom/Caedarva) was not a foregone conclusion.