This is fair commentary, I need to do the math on Assassin’s Annihilation. I accept there will always need to be some stuff for PvP but there’s is still room for improvement.
Did a quick run of the numbers, and while I made a couple of assumptions that may not turn out to be accurate (such as AA scaling with power/prec/ferocity at a similar proportion to other damage sources) it does seem the AA is definitely worthwhile if you have a reasonable expectation to be able to get flanks. Swift Termination possibly pulls ahead if you have the potential to AoE multiple opponents, and if you’re not likely to get flank (such as when you’re on your own in PvE and not using orbital hammers).
Assume that a 12-second period against a single target that’s below 50% and which can be easily flanked (so one of those Nuhochs, say) involves 1 usage of UA and 4 repeats of the autoattack chain. This is a rough estimation, but until we get good information on the aftercasts, ~2.5s for an autoattack chain seems fairly typical.
UA does 7×202 = 1414 damage (it grants might, too, but I’m assuming that scales everything equally)
Preparation Thrust does 293 (and grants vulnerability, but that will increase all damage)
Brutal Blade does 183, but strikes twice, for 366 total
Rift Slash against a single target does a total of 440. (Against multiple targets closely packed, this can rise to 660 through overlapping explosions)
Total damage from autoattack = (293+366+440)x4 = 1099×4 = 4396
Total damage over 12 seconds specified = 5810
In practise, the actual damage will be higher, but assuming Assassin’s Annihilation scales proportionally with power, precision, ferocity, might, and vulnerability, the proportional numbers will be the same (if the actual number is, say, 50% greater due to power, vulnerability, and crits, it is likely that the same will be true of Assassin’s Annihilation procs).
Assassin’s Annihilation ideally strikes once per second. In practise, some of that is almost certainly going to be lost by not striking at precisely the same time. Let’s say it triggers 8 times over the period. So that’s 243×8 = 1944 extra damage.
Swift Termination gives a flat +20%. That grants an extra 1162 damage.
Now, this ignores some of the complexities of interaction with Assassin’s Annihilation. There are few things that can increase the damage of AA further apart from decreasing the average time between finishing ICD and triggering again. AA doesn’t scale further with Impossible Odds, nor does it scale further when striking multiple targets (while I think every revenant set is good at striking multiple targets) and it is vulnerable to being shut down entirely if your target is facing you (unless you have Vengeful Hammers running). On the other hand, it still triggers if your target is above 50% health, and the extra healing is not to be sneezed at.
It’s also pretty rough with a lot of guesstimates regarding timing, but I suspect the ballpark is accurate, at least to the level of doing a comparable amount of additional damage.
Of course, this is assuming that Assassin’s Annihilation scales with berserker stats with a similar ratio proportional to base damage as the skills. If it can’t crit and doesn’t scale with power (or not as much) then Swift Termination will pull ahead with more berserker-oriented builds, but AA may still have its place for builds that are not quite so focused on DPS.
AA is a lifesteal and as such it cannot crit. This means if you’re going for a pure damage spec with zerker and such your best bet will probably be ST. However it’s still a lot of extra damage and healing, and can be a really good option for certain weapons, as being a lifesteal it ignores armor, it’s great for fighting certain classes like guardians and eles. It can also boost Mace and hammer damage, as mace’s physical damage is on the low side and AA works kind of like a condition, ignoring armor; and for hammer, the CD lines up with the cast time on the auto attack, meaning all your autos will proc it asuming you hit the foe from the side.