Players can agree the enmity system is broken. We can't agree how to fix it, but we know it needs to be fixed. Might as well toss my method into the ring as well.

Currently enmity is measured by presumably 2 counters of Volatile Enmity (VE) and Cumulative Enmity (CE) and the total determines the target of a particular enemy. I would propose splitting the enmity values even further into Physical Enmity and Magical Enmity. Could be more but let's start with these two for simplicity's sake. CE and VE would be treated as usual within these two categories.

Physical Enmity would be gained by melee attacks, physical WS, BLU physical spells and the like. Magical Enmity would be gained through magical WS, spells, healing, etc. Enmity control abilities like Provoke and Accomplice could alter both.

A damage dealer like WAR would fill the Physical Enmity counter like normal and upon hitting the Physical Enmity cap would then begin to fill the Magical Enmity counter at a severely reduced rate. As an example, say the WAR's attacks generate 100 Physical Enmity. After hitting the Physical Enmity cap, the WAR's attacks only generate 30 enmity on the Magical Enmity Counter. In a similar vein, a WHM will generate Magical Enmity and upon reaching the Magical Enmity cap, the overflow would begin to fill the Physical Enmity Counter at a reduced rate.

Ideally, this overflow would also be removed at an increased rate as well, so an attack on the previous WAR would remove 20 enmity if they were uncapped on Physical Enmity, but the same attack would remove 50 enmity if it was in the overflowed Magical Enmity category. Normal generated enmity would have to take precedent over overflow enmity. For example, if the cap for both categories is 1000 and a RDM is at 1000/1000 Magical Enmity and 700/1000 of Physical Enmity caused by magic overflow and the RDM decides to start attacking with a sword, the Physical Enmity counter will increase at the normal rate until capped at 1000/1000 at which point the overflow enmity starts to convert into normal enmity. This does require that we have at least 8 enmity counters: Physical CE, Physical VE, Physical CE overflow, Physical VE overflow, and the 4 Magical Enmity counterparts.

In general, most jobs focus on a single type of enmity generation. Damage dealers will usually not use magical WS and mages will not melee for the most part so current tactics should not be affected too much.

The system does give a significant benefit to jobs capable of generating both Physical and Magical Enmity such as PLD and NIN. PLD and NIN would both be able to generate and lose Physical and Magical Enmity at the regular rate rather than the altered rate of overflowed enmity since they cause melee damage and cast cures/ninjutsu. This also extends to DRK if they use their magic liberally, melee RDM, magical casting BLU, bullet shooting Wildfire COR and DNC using Waltzes so some regulation should be advised.

Now for some reasoning behind the new system. We don't want to simply raise the enmity cap because we are left with the same problem as before, just 30 seconds later. A flat reduction in the amount of enmity generated could completely disrupt current lowman battles. A raised cap on tanking classes would just allow everyone else to go nuts with complete abandon. We want tanking to be a viable tactic, not a requirement for all situations.

The two enmity types allows current tactics to still be utilized while giving an advantage to our tanking classes. The overflow enmity, provided the altered enmity rates are reasonable, prevents specialty jobs from having perfect safety from being targeted.

Additionally, this would open up some new challenges from the AI. Instead of a single target, the AI would now have 3 to choose from and could tailor attacks based on the type of enmity. A player with high Magical Enmity could be more prone to being hit with silence and addle while a high Physical Enmity player could be targeted more with flash and amnesia. The player with the highest total enmity could still be the main target. Alternatively, the AI could rotate between the 3 bringing a new need for Libra or in the case of something like Hydra could attack all 3 targets at the same time.

I know a big problem might be from the fact the overflow enmity cap would need to be constantly adjusted based on the current level of the normal enmity i.e. (Physical CE overflow cap) = 1000 - (current Physical CE). I've heard there's a calculation lag for DRG when macroing Wyvern HP% gear, but not for Wyvern HP+XX gear, so there might be some issues with four constantly varying caps that can forcibly alter the current value.