You... did, though, unless you were just going to stand around waiting for the patrol to pass, you may well start on one side or the other. Additionally, a WHM could solo the central winch (Holy was a 5s stun at the time) while the rest of the party went left, right, or through (if pat didn't block it or one had a PLD to tank it) to grab the oils and winches from East to West or West to East by just nuking down the oil-bearer and then Sprinting onward, outranging the mobs where possible and using the boss wall to hold them in place where not. We had that shit down to a science, but it still technically had some variation.
Yes, it could have been better balanced for variety and some 80% of runs, so long as they were done at full speed, all took the same route. In the remaining 20%, though, yeah, variations happened to recover or to better make use of BLM's AoE Sleep/Blizzard II (or Holy/Miasma II) and potentially Sprint w/ Paeon.
To my mind, it's just a matter of the added initial extent of, and the added longevity of, the dungeon's novelty vs. those obviously diminishing returns. Honestly, though, in terms of 'dungeon content novelty/longevity per hour of development time' would already seem to incentivize spending a bit more time on each dungeon (and systems encapsulating them).The rest of my post is just about trying to make a linear dungeon seem non-linear whilst still being linear. Whether that solution will please people who wanted a non linear dungeon, who knows (probably not), however, it is still something that can make each run of the same dungeon different, to a point. Only so many routes you can make after all.
In terms of varying up pathing slightly (again, less important to me than shaking up the pacing, flow, and encounters, but still worthwhile)... plopping mobs on the map doesn't take up much more time. Nor is pathable ground so much more time-intensive than making an equally interesting path or slope leading (inaccessibly) off the main trail for purely visual purposes. So, if the dungeon seems like it should look that large (even if just for visual purposes) already, then... yeah, some more paths would be cool and probably a slight net increase to value for the dungeon's distinction and longevity.
Ultimately, though, a lot of the benefit of making more interesting dungeons depends on the amount of interest that can be generated from the differences and agency within the party members' respective kits, which it turn requires that the dungeons not just be a steamroll. So... also really hoping that minimum item level difficulty would actually be worth speaking of and that the ilvl caps would fall far closer to merely the ilvl the dungeon rewards OR a real Expert Roulette (dungeons upscaled to level cap and to the party's average ilvl).
(No, I see absolutely no reason why Savage gear would "need" to "reward" players with a gutted dungeon-running experience from steamrolling the whole thing. Frankly, I'd be fine with most content having a maximum item level barely over what they reward, and just reducing the weekly grind requirements in exchange, or having the amount by which ilvl has been synced down increase rewards near-proportionately.)
EDIT: Now, if we REALLY want to incentivize more varied layouts for dungeons, then there'd simply have to be further use for those variations, even if that might not be for "Dungeons" themselves -- and instead some new content type (I used the term "Delves" before) that borrows from those prior "Dungeon" assets (to perhaps less of a visual rehaul than Hard Modes, but still likely significant).