Quote Originally Posted by Ultimatecalibur View Post
Those rotations, in my understanding, were more to cover flaws caused by lack of a 2nd viable formless gcd to fill in holes in the core rotation created by higher sks and GL. The rotation was fairly solid at 2.06 gcd (29 gcds per minute), but buffs and dots started clipping or falling off (depending on rotation) bellow that necessitating use of Cross-class weaponskills (of which Fracture being the best available at the time).
The entire way Ninja played optimally for three versions of the game was likewise in large part called 'unintended'. That didn't stop it from being fun, though, or having far more rotational choice and complexity than it has now.

I might even go so far as to say that having "flaws", where a job is balanced both despite them and if/when having "fixed" them, is often better than an originally "perfect" rotation as it at least allows for choice and flexibility.
We saw this with early Chaos Thrust, Demolish in either timing, the excess duration on SAM buffs, and the duration of Shadowfang (itself, in SB, and alongside Mutilate prior to that): in ARR, a higher speed variant allowed for an additional Full Thrust per Chaos and in HW it allowed for the 'Geirskogul-abusing' 2.33s rotation that was so fun in cleave, even if inferior to the 2.40 for pure single-target; until ShB one could use uptime-loss counts from longer Ninjutsu uses to trim or delay Shadowfang, eventually allowing for another Aeolian per minute and decreased SF clip once per minute and decreased SF delay another time per minute via increased Fuma usage; the increasingly excessive SAM buffs likewise allowed for trims and potency shifts around Meikyo Shisui -- all things that increased both the flexibility or number of rotational strings available and the skill ceiling for each affected job.

Note also that none of these things were permanent changes except at very specific GCDs -- they didn't usually involve just swapping one rotation for another. Instead, they increased the number of rotational strings available. One string might go over the DoT timer, trying to reapply it just before the next global tick after it had faded, while the next might try to rush it out, preferring to waste some DoT duration than risk missing a global tick. Having different parts to align, by which to encourage adaptation and variance, is a powerful tool for gameplay. And pursuing above all a single "way it's supposed to work", rather than a sliding set of choices over varying SkSs (with still multiple choices over time for any given GCD) kills that design tool.