Tuesday, April 21, 2009

Enfeebling nerf, more math, DNC gear

So, apparently during the last FFXI update, there was a significant nerf to enfeebling magic. Enfeebles are resisted a lot more, and do not last as long as they used to. This probably contributed to our Omega failure in Limbus last Saturday, though it was not the only cause.

It took a while before details of this change came to light... basics are simple... any type of enfeeble (RDM, BLM, BRD, BLU) has a much higher resist rate, and lasts only a fraction as long as it used to. This is even with a fully merited RDM with good gear, so Flayer didn't even have a chance with RDM74. This makes basically any type of low-man situation where enfeebles are used a large issue. This also explains why Rooks had so much problems Silencing scorpions in Dynamis-Bastok... it just didn't last as long as it used to.

There is some community backlash to this... Here's an editorial from one of the FFXI "Premier sites" which are supposed to have good two-way communication with SE. It's clear from this piece that there is no such thing, that SE is giving them as much of the cold shoulder as everyone else... you know, the whole lack of customer service thing I talked about in a previous rant. Just business as usual from a crappy company. :) This also made BLM soloing much more difficult, if not impossible.

Now the question is... is this an intentional nerf, or simply a bug? Why was the code changed in the first place?

1. It's a bug
If so... then at some point in the (near) future we should be seeing another emergency maintenance. There will be no need to adjust battle tactics and life will resume as per the past.
2. It's intentional
If this is the case, then let's consider why they would have done such a thing. Looking at the last update, they did add one little thing that could have an effect on resist rates: RDM Tier II en-spells.

So far, I haven't seen a single RDM use these spells yet. I've heard of RDMs going out to farm them, maybe some people buying them, but I've never seen a single RDM casting it and going out to melee. What if they changed resist rates so that it would force people to use RDMs on the front-line? Since RDMs are the #1 enfeebler out there, chances are if you want to enfeeble a mob, you'd have a RDM around, so having them use their en-spell IIs would make sense. The tier II en-spells lower the mob's resistance to the opposite element.

So... in theory, if you're fighting a NM and need to land Slow on it, have a RDM cast Enthunder II, run up and poke the mob and enfeeble it. The resistance lowering effect of Enthunder II should allow Slow and Elegy to reach their former lengths and strengths. The RDM would have to keep meleeing in order to keep the resistance lowered... and the RDM would actually have to HIT the mob, which is what Composure is for.

By the same token, this might also end up making Stutter Step more useful. While Stutter Step lowers magical resistance of a mob, in practice I've never actually needed to use it. Most mages are properly equipped with magical accuracy gear, so there really was no reason to use Stutter Step. By increasing mob resistance to a point where you'd actually NEED either a RDM or DNC in order to properly enfeeble it would be just what a sadistic developer who wants to influence player behavior would do.

Of course, this is simply a theory... it's more likely that it's #1, and they're not going to fix it since the devs probably do not play the game and aren't bothered at all by having their customers be angry at them.

Tanking Math
Last night, Fuuji was telling the BLMs that "nuking every 30 seconds keeps you safe" during Tinnin. He also said to give the PLDs a minute to get hate before starting to attack. With all the enmity math I did, I wanted to figure out how tightly you could nuke and still be "safe", and how long would it take for a tank to build enough hate before you could start nuking.

Let's make a few assumptions... we're fighting some sort of large NM, like a Tier IV ZNM or Proto-Omega. I'm using a level 85 mob for the purposes of the calculation. Let's also assume that we have a nice BLM that can cast a spell that does 1k damage. Merits and partial resists would make that go up or down of course, but 1k is a nice round number to play with.

For damage dealt enmity against a level 85 mob doing 1k damage, we have:
CE = 1.379 x 1000 = 1,379
VE = 4.14 x 1000 = 4,140
Very high amounts. That's a total of 5,519 enmity done in one spell. VE decays at a rate of 60/second, so to bleed off all VE would take 92 seconds. Taken another way, a BLM who wants to stay VE neutral can do up to 870 damage/minute. Your CE will cap at 10,000 after 8 minutes, but that is expected in any long fight. It's the tank's job to maintain CE + VE at higher than 15,000 total in order to hold hate against a prudent BLM.

So... looking at these numbers, nuking every 30 seconds isn't really that safe. You'd have to about 435 damage per nuke if you want to stay under the radar. Nuking for over this every 30 seconds would result in a net VE gain and you'd eventually be smacked... as proven by Fuuji himself last night :)

Summary so far:
- A BLM(or any DD) will generate about 5,500 total enmity for every 1k damage done
- Hence, before any serious attacks can take place, the tank has to generate at least 5,500 enmity
- DDs can do damage at a rate of 870 damage/minute and stay VE neutral, that is, they will not pull the hate of the mob no matter what if the tank is doing their job. If you do damage higher than this rate, you will eventually pull hate.
- For reference, any healer can cure at a rate of 826HP/minute and stay VE neutral according to the same 60VE/second bleed-off rate. Of course, this does not count the lower enmity effect of gear and Cure V

Alright, so now we have some numbers to work with... how cast can a typical endgame tank reach the initial 5,500 total enmity number, and can they keep it up?

PLD/NIN
This job has surprisingly the least number of hate gaining tools. The big ones are Flash (180 CE, 1280 VE), Sentinel (1800 VE) and Shield Bash (900 VE). The rest of the hate must be obtained through curing, either themselves, or other people, and doing damage.

So... spiking initial hate gain using Sentinel and all abilities, we get:
- Sentinel (1,800) + Shield Bash (2x 900) + Flash (2x 180CE + 2x 1280VE) = 6,520 total enmity
Now, the danger here is that most of this is VE and only 360 is VE. This means that if a PLD goes all out in the beginning, chances are they will be able to hold hate through a trigger-happy BLM. However, a PLD's recast timers are such that just through job abilities alone, they cannot maintain their developed VE. Even Flash has a 45 second recast... with Haste and Fast Cast gear, they can reduce the VE loss rate, but they'd have to make up the remaining VE through damage dealt.
- Side note: Having a PLD start off a fight with a 300% TP Spirits Within is definitely a good way to start the fight. This would throw up about 3,300 enmity from the start, giving the PLD a head start in establishing hate.

Of course, the biggest advantage of this combination is that with Utsusemi, you're reducing the rate you lose CE by simply not taking damage. In theory, this means that a good PLD/NIN tank that is conscious of their shadows while doing damage and curing OTHER people should be able to build up to the 20k total enmity cap over time. I'd wait at least a few minutes before seriously attacking the enemy with a PLD/NIN tank

PLD/WAR
This combination is almost never used for tanking anything big simply due to the enmity loss from taking damage. You need to reduce the CE loss as much as possible. As my last post pointed out, you can't hold hate even if you're curing all the damage you take. There is a net CE loss from taking damage then curing it again.

Still, PLD/WAR has one advantage: Provoke. This gives an extra 1,800 VE over PLD/NIN. Combine that with Sentinel, that's another 3,600, almost reaching the 10k from the get go. Since this is mostly VE, damage taken would not reduce it. The PLD/WAR would also be able to maintain their VE at a high level since the recast and enmity gain of Provoke is equal to the loss rate. A PLD/WAR could reach the VE cap of 10,000 and pretty much stay there as soon as the fight begins. This would allow all DDs to do damage immediately after the fight begins.

The disadvantage is that a PLD/WAR would never be able to build CE. Damage taken would wipe out any CE gains, even if the PLD cures herself. Hence, in any long fight, other members of the party or alliance would pull hate no matter how long they wait between actions simply because they build up CE to the point where they hit the 10,000 cap, matching the PLD/WAR's VE cap. Any action that generates VE would get the mob's attention.

So, this is a good tank for only short fights where you want to do as much damage up front as possible.

PLD/RDM
This is an interesting combination... you're still blood tanking, but you are using two things to obtain a net gain in CE.
- Reduction of damage via Phalanx
- Increase in CE generation tools

A RDM sub has a few interesting hate generation tools, specifically:
- Dispel: 320 CE, 320 VE, recast time 10s
- Sleep: 320 CE, 320 VE, recast time 30s
- Blind: 640 VE, recast time 10s
- Bind: 640 VE, recast time 40s

So, if you spam just these 4 spells, you can gain CE at a rate of 2,560/minute, and VE at a rate of 7,040/minute. A single minute of this brings you to about 6k total enmity, and 4 minutes would cap both CE & VE. Starting to attack at the 1 minute mark is about right.

This assumes that you are not suffering so much damage that it negates the CE gain. Your CE loss would need to be less than 2,560/minute, which means (assuming a PLD with 1200HP):
2,560 CE loss = 1800 x [Damage taken] / [Target Max HP]
Damage taken = 2560 * 1200 / 1800 = 1,706 HP/minute, or 85 HP/tick

This is about 2-3 Cure Vs in 1 minute... for any mob that does more damage than this in a minute, the PLD/RDM tank will not have any net CE gain (and probably run all mages out of MP)

Hence, PLD/RDM is good for long battles where the mobs do not do much damage. The time it takes to cast spells, spell interruption, etc. would make PLD/WAR a better choice for shorter battles.

NIN & /NIN
Overall, there are two ways to maintain a high hate line:
- Cap VE and have a method of maintaining it at the high level. WAR and /WAR have this capability simply through Provoke. PLDs need to do damage + cure in order to maintain and gain VE.
- Cap CE and do not let it go down. You need a way to avoid or reduce damage taken. NIN or /NIN is the best way to do this.

Of course, the danger with relying on Utsusemi to keep CE up is that if a hit does get through, you end up losing a ton of CE due to damage taken. The best way to guard against this is to have a 2nd layer of defense so that even if a hit gets past Utsusemi, you take minimal damage.

Now, instead of going through the math, I'll just talk briefly about a few other jobs
NIN/DRK:
Lots of high CE (Sleep, Aspir, Stun) and VE (Bind, Absorb-spells, Last Resort, Souleater) spells job abilities. Ninjitsu spells generate a decent amount of enmity as well, though you trade off VE for CE. Can hit CE cap easily and much faster compared to a PLD/NIN, though maintaining VE would be difficult. Would need to spam Absorb spells, Bind and Ninjitsu to maintain a high VE. You can probably start attacking about a minute into battle if the tank starts by spamming job abilities. Would also lose hate much easier compared to a PLD/NIN simply due to taking more damage.

NIN/RDM:
Now... this is interesting... I don't think I've seen anyone talk about this combination before. It would essentially be similar to both NIN/DRK and PLD/RDM. You'd lose the high enmity from Stun, but you'd gain Stoneskin, Phalanx and better be able to reduce CE loss if you take hits. You'd use the same casting cycle as PLD/RDM, Dispel, Sleep, Bind, Blind, etc. Might be a better choice compared to NIN/DRK simply due to Stoneskin.

RDM/NIN:
This is a very similar scenario to PLD/RDM, except with more damage mitigation tools to stop CE loss. Utsusemi, Stoneskin would cut down CE loss to almost 0, assuming you're not fighting something that hits too fast. It would require a good RDM who can both track shadows, and keep track of how much Stoneskin is left and when to recast it. Properly played, RDM/NIN should be able to hit both the VE & CE cap and keep it there, allowing all damage dealers to go all out.

NIN/WAR & WAR/NIN:
The WAR side of things basically gives you Provoke as the VE tool, assuming you provoke every 30 seconds. After that, hate gain is entirely through damage (and to an extent, ninjitsu in the case of NIN/WAR)
The disadvantage of these combinations is that if a hit gets through shadows, you'll lose CE very fast simply by taking a lot of damage.

DNC/NIN:
The main disadvantage of DNC/NIN is that the job is slow to build up enmity. We can maintain VE using our job abilities, but any net VE or CE gain needs to come from either damage dealt, or damage cured. A Dancer attempting to tank needs to strike a balance between keeping TP to cure, keeping Finishing Moves to use Animated Flourish, and using TP to do damage.
The advantage is that besides Utsusemi, we have Fan Dance that can help reduce damage taken and slow down the CE loss, giving us an advantage over NIN/WAR or WAR/NIN.

Overall Conclusion
Basically... RDM/NIN and NIN/DRK are your best bets for tanks that can get up to a high level of hate quickly and maintain it. PLD/NIN is probably the next best bet only because it takes a longer time to generate hate compared to the first two jobs. However, PLD/NIN has the advantage of being able to cut down on CE loss due to taking less damage. Throw in a Scholar or Blue Mage for some Stoneskin and you'd really help out in that department.

NIN/RDM is a combination that I think has a high possibility of being a more reliable tank compared to NIN/DRK. You'd lose access to Stun, but you'd gain more safety with Stoneskin and Phalanx and hence, lose less CE over the long-term.

WAR/NIN and NIN/WAR can maintain VE, but they do not have enough hate tools to really compete with the other three tank combinations, and would lose hate too easily from simply taking hits.

DNC/NIN is a combination that has yet to be tested out. From the theory, I can already tell that it would be harder to get going in a solo tank situation as it requires someone *other* than the Dancer to take damage in order to generate hate, and it takes a longer time compared to NIN/DRK or RDM/NIN to generate both CE & VE. This could lead to risky situations if the other party members are not jobs that can take the beating. In a co-tank situation, we'll have to see if a DNC can grab hate from the other tank and hold onto it.

Overall, it is safe to start attacking after about 2 minutes of the tanks engaging. After 4 minutes, all tanks should be at their respective hate caps. As long as you stay under 870 damage dealt per minute, or 826 HP healed/minute (individually, not alliance total), you'll be fine and not have to log out to clear hate.

DNC Gear
I picked up some +enmity gear last night to experiment with Dancer tanking... Grabbed a Harmonia's Torque and Eris Earring for now... I already have the Etoile's Bangles, and a Warwolf Belt from BLU. I also have a regular Breath Mantle from my brief period leveling PLD that I'll experiment with before investing any more into this theory. Connie pointed out that Dorado Sushi has +enmity on it, so I'll try some of that out.

I'm planning on customizing my ACP armor (assuming we actually finish this) with +5 enmity and +10 evasion... a very strange combination compared to what I'm sure other people are doing. However, I really do not have any use for this armor. On BLU, I'm set with AF+1 & Relic. SCH I'm set with relic + AF. On DNC, I'm set for a Haste body set, and I'll *eventually* pick up a better accuracy/attack body piece, either the Enkidu's harness or the Etoile body. Meanwhile a +enmity piece be interesting to experiment with, and +evasion would help shadow recasts.

Oh, and congrats to Zu for his pink harness... now he'd better get on more often and use it or else I might lose one of my daggers... in his back!


No comments: