Tuesday, July 27, 2010

Some Revised Numbers

My previous post on cataclysm numbers did not include the spell power bonus we are going to see on weapons. Right now wowhead has a crafted blue with 1532 spell power. I'll assume that the epic version will have about 9% more spell power, and that spell power increases 8.5% per tier (maybe I'll post soon on why this is not 13%, it's a pretty funny bit of math). That means my revised numbers are:

Raid Tier 1Final Raid Tier
Spell Power7,98711,333
Ratings/Spirit total5,0998311


I was reading Thespius who referred to Dills regarding spells that should leave the game in Cataclysm. Dills posted a list of spells he believes should be removed and Thespius said that Dills' post was ill-considered.

I basically agree with Dills regarding his list of spells that should go, aside from the "Soothes" which can stay as non-combat class quirk spells that don't really get in anyone's way or affect raid content but make classes feel different, even if only to a few people. Amplify and Dampen are special raid buffs that should either get in a raid buff category and be shared between classes or hit the road. Hunters really don't need three different melee strikes (though I personally think they could certainly keep one melee strike).

But the spell I really want to talk about is Lightwell. Obviously this spell is not a favourite of the community, though it has its die hard supporters. The problem is that debates over Lightwell tend to degenerate into debates about how the game should be played where people lament the fact that in today's GearScore obsessed world everyone just cares about upping their personal dps and can't be bothered to try to keep themselves alive. This annoys me because I don't like nostalgia for imagined pasts, but it also annoys me because it misses that actual problems that Lightwell faces. The biggest problem is that Lightwell isn't good.

The most interesting part of this post and response was an analogy to the made up spell Shadowwell. Imagine you could put a well of darkness on the ground and your healers could click it to do damage to the boss. Would that be a good or a bad spell? It's sort of a silly question without knowing how much damage it does. If clicking it did 200 damage to the boss over 10 seconds and used a GCD then everyone would wonder why the spell existed. If clicking it did 100,000 damage to the boss over 3 seconds and used a GCD then everyone would complain that you needed to stack priests to beat everything because Shadowwell was a necessity. 4k damage over 10 seconds at the cost of a GCD: bad. 50k damage over 60 seconds at the cost of a GCD: good. Pretty much any amount of damage without the cost of a GCD: good. What if the DoT breaks if the target takes a hit for 10k or more? Then it's bad. What if the DoT breaks if the target takes a hit for 30k or more? We don't really care about that.

The numbers actually mean almost everything to whether a spell is good or not. If the numbers are high enough then the mechanic can be as awkward as it wants because people will use it anyway. If the Lightwell HoT didn't break on damage and healed for 10k a second on a 1 second tick for 10 seconds then people would use it for sure. As it is few people use it, and quite rightly.

We all know they could make Lightwell good and we all know they could make it too good. But this is what brings us to the fundamental problem with Lightwell, and the reason it should go. They've been working on this spell for years and they haven't got it yet. Lightwell has always been a contentious spell, and people have always argued that it really isn't that great. When raiding used to be set up in a way that would have made it better, the heal broke on any kind of damage so it was virtually useless on the fights that you most wanted to use it on. They've eased that restriction but it is still there threatening to negate the heal. Fights have become much more dynamic and mobile in the mean time, making useful Lightwell placement difficult on many bosses. On top of that, dps checks have surpassed tanking and healing as the primary factor in whether or not you can win a fight, meaning dps can't afford to switch off target to get a heal on themselves, they have to trust their healers to do their job.

Lightwell has also been very hampered this expansion by the fact that healing is so powerful and fast. A dps choosing when to put a HoT on themselves is pretty meaningless when they might not know whether a 20k heal is about to land on them. Dps should be in a position to judge their own health, but they are not in a good position to plan healing for the raid, and putting a HoT on someone that won't do anything for the next two seconds is a question of healing the raid, not healing that one person. From an individual perspective the tick from Lightwell pretty much always be too little, too late.

And with that powerful healing comes powerful damage. I would love to have the option of putting an extra HoT on myself when I run out to drop off a shadow zone on Halion, but I don't have the luxury of going by an immobile place to do so, and even if I did the ticks from the zone can actually do enough damage to break the heal.

The fast healing and brutal damage will be less of an issue in Cataclysm, so Lightwell might have room to work. But it still comes back to the numbers.

I think on Lightwell the developers have a tough decision to make. Either make it good by making the numbers compelling, or scrap it completely. The breaking on damage feature probably needs to go, the cost should be dropped to virtually nothing and the duration/cooldown extended so you are expected to cast it only once per encounter. Make it the Lightwell untargetable (like a door), so clicking it doesn't deselect your current target. Then make the heal big enough that it can be meaningfully incorporated into a plan for the encounter (e.g. if you get hit with ability X, use the Lightwell or when the boss does ability Y, everyone click Lightwell). But, as Dills says in his post, it's current implementation on position on the tree make it really bad. Making Holy priests takes a spell they don't want to cast to get a talent that is supposed to be a big new feature for them this expansion is not going to make the priest community happy.

Monday, July 12, 2010

Guild Perks

In an early preview of Cataclysm, one of the selling features was guild talents. The idea was that your guild would go up levels based on participation from the players, and then you'd get talent points to invest in a tree that would benefit your entire guild.

Since then the developers have decided that having an actual talent tree is probably a bad idea. Choices are fun, but in the case of guild talents they would be fun only for those who got to make the choice, and they would be a source of strife and drama for the multitudes who didn't get to be involved in that decision making process. For example, they said, suppose you are in a guild because you are friends with the other guild members. They all raid while you mostly PvP. They will choose raiding guild talents and you will be faced with a tough choice that doesn't add to anyone's fun: stick with your friends for the social aspect of the guild or leave for a guild that chose the PvP talents to support your playstyle. Instead, they have opted to get rid of the talent trees and give guilds "perks" as they level up. Basically it's all the same content, but with no choices involved and all guilds get all the perks.

A very good point, but it carried with it a strong implication that ran contrary to how I assumed guild talents would work. They were basically saying there would be guild talents that made you better at PvP and PvE. Talents that would have enough of an effect to leave you feeling handicapped by not having them. Is the problem really the choice and the guild drama, or is the problem that this sort of talent exists in the first place?

WoW.com had a view of the current state of guild perks a couple of weeks ago. If you haven't seen them yet then take a moment to look at them before you continue.

The talents that the developers would be concerned about here are the talents that increase heroism and honor point gains by 10%. While most of the talents favour PvE over PvP, those are the only talents that have to do with gearing up and beating your enemies that you might feel really let down if you didn't have. The ability to summon your guild helps raiding and doesn't really do anything for PvP, but it's a minor convenience issue, not one that affects your progression.

The real problem is that these perks that increase the rate at which you can acquire last-tier's gear aren't a good idea in either system. They will feel pretty exciting when you first unlock them and start getting extra points, but several months into cataclysm all of the big guilds will have those perks. That will mean that everyone will expect to have the perks and rather than seeming like a bonus, not having them is going to seem like a penalty. No one will feel like they are getting 10% more honor points, but there will be people out there who feel like they are getting 9% less honor points. There will be people who want to start new guilds and can't because no one wants to join a guild without the perks. They will be a source of aggravation without being a source of satisfaction or fun.

Those aren't the only talents that fall into this category either. Getting more reagents when you use gathering professions is the same. Everyone is going to feel like "15% more" is the standard, and someone who chooses to be guildless or who plays in a very small guild of very casual players will just get 13% less. In fact this may be even more true for gathering professions than for honor/heroism points. The value of gathered materials is set by the market, so other people getting more directly devalues what you get. What is the reasoning behind imposing this on casual players?

There are lots of great examples of perks on the list. Instant mail between guild members, summoning a remote guild bank and having 10% of cash gains added to the guild bank are all spectacular. Perks that give personal benefits to the members of the guild are not good ideas as guild benefits because they do not benefit the guild. They benefit the members of the guild, which is a very different thing. Perks that give individual benefits mean that there is a selfish incentive for people to attach themselves to large established guilds over working with their own guild. This is an incentive the game does not need.

This is something that is particularly risky because once these benefits have been handed out it will create a lot of backlash if they are taken back. I hope that individualized benefits as guild perks can be stamped out entirely during the testing phase, before they go live. I don't have a huge interest in perks vs. talents (I would be the one choosing the talents anyway) but I do have an interest in the benefits given by them being guild benefits rather than individual benefits.

Wednesday, July 7, 2010

Three's a Crowd

Blue posts have talked a fair bit about the new healing environment they want to create. The basic idea is to give you different spells that you will choose between depending on how the fight is progressing, and to make mana a consideration. I agree with the central direction of this, but there is something about it that didn't make sense to me from day one: a normal heal, a big heal and a fast heal.

Heals vary on two easily quantifiable axes and also on a third, harder to pin down level. The first is healing per second, the second is healing per mana, and the third is the way in which the heal delivers healing. That third one encompasses the difference between direct heals with different cast times, HoTs, Shields, and bizarre things like Prayer of Mending. They are very different, and some are clearly better than others, and that betterness goes beyond and must be measured against HpS and HpM.

But when it comes down to two different heals that function in precisely the same way, there is really only one way they can be meaningfully different. One heal can have higher HpS and lower HpM than the other. Any other variance doesn't work. If Heal A has both higher HpS and higher HpM than Heal B then you don't really have two different heals, or at least things would be in all ways the same for you if you actually did have only Heal A.

And so the normal heal, big heal dynamic works. The big heal is presumably more expensive per point healed. They have the same cast time. You cast the normal heal most of the time and when you think it isn't going to be enough (either because you've been falling behind for a while, you know a big hit will land in a couple of seconds, or whatever) you cast the big heal instead.

But the way that heals deliver healing is a very important consideration. Imagine instead of casting a heal for a specified amount of time you could just channel a heal that continuously drained your mana and increased their health and you'd stop when you felt you'd done enough. This would be the ultimate in giving the ability to quickly react to damage as well as not overheal. I would basically be the limit of making heals faster. On the other end you have a heal that takes 10 seconds, is extremely cheap, and heals the target to full. This is a terrible spell because 10 seconds is too long to go without a heal1.

So a fast heal with the same mana efficiency and healing per second is better than a slow heal. Thus, a fast heal can afford to be a little weaker in terms of efficiency and throughput and still be useful. But how much cheaper can it be? And is the difference between 1.5 seconds and 2.5 seconds enough of a difference to justify a big disparity in effectiveness? Developers were talking about making it so you didn't have to worry about the tank going from full to dead in two swings. If that's really the case, then when am I going to pay more mana to get a heal that lands faster when the slower heal is fast enough?

I was having trouble getting my head around how the normal/fast/big heal system was going to work. They would make reference to the way healing used to work in BC, but we didn't really work on that model. We either worked on the shadow-priests-make-it-so-mana-is-a-joke-and-I-spam-CoH model or the six-ranks-of-heal-and-g.heal-on-my-bar model. We chose between multiple points on the efficiency vs. throughput spectrum much moreso than we chose between fast and slow. These days your tank does die in two hits so you don't dare cast spells that take 1.5 seconds, let alone 2.5 seconds.

Without the breakneck pace for healing, I was having trouble seeing how a fast heal would fit in. It seemed like it would make more sense to give us small/medium/large than fast/medium/large.

Now we've seen the costs and healing amounts of Flash Heal and Greater Heal and it doesn't seem like the developers had any better idea than I did about how to find space for these different heals. Bars show efficiency on the right axis and lines show throughput on the left axis:

There is never really much differentiation between the two heals in terms of efficiency or throughput. Since you will presumably spend a lot of time casting Heal, and switch to these more expensive heals only when you have to, it is hard to believe you will decide that doing 3% more healing is worth the trade-off of being slower. Doing more healing because you aren't keeping up doesn't seem like it will often coincide with wanted to go more slowly.

But I think finding space for both of these heals is much more than a numbers game. Imagine that instead Flash Heal did only about half the throughput of Greater Heal (which is around what I expect Heal to be) and was also less mana efficient. Then we wouldn't cast Flash Heal at all because there would basically never be a situation that warranted throwing your mana away on such a weak heal. You might say, then, that logic suggests there is a medium point where both spells are castable.

That is not necessarily the case. When you are actually healing, you don't get to sit around and look at graphs or run simulations to figure out which heal is the better one to use in the situation. You have to choose very quickly, and you have a limited amount of brain cycles to devote to making the decision. The closer two things are to even, the harder it is to choose between them. Having two different heals that basically do the same thing that are close enough together in effect that it is a subject of much debate which is better means you are probably best off only putting on of those heals on your bar. If you are really smart, you might switch which one depending on the nature of the fight. Choosing on the fly which one to use, however, is putting too much thought into a decision between two very similar spells. Casting either spell now is almost certainly better than casting the "right" one a couple of tenths of a second later.

Bar space is also an issue by itself. I currently have 37 keybinds that I use in combat while I'm healing (though some of them are very situational, I still use them). In the next expansion I'm getting Leap of Faith and Power Word: Barrier (and there is some chance I would have Inner Will on my bars). I don't know how to fit an extra heal-of-a-different-colour into the mix unless it is very noticeably useful. They still have a few months, but I really don't know how they are going to pull off the normal/big/fast trio.

1. In reality this could lead to full-heal rotation setups which would be bad for a completely different reason. But extend the cast time to 20 seconds and you couldn't even do that.

Tuesday, July 6, 2010

50%? Really?

Pre-shielding when you know big damage is on the way for select targets is fine. Trying to keep shields up on the entire raid is a strategy that we think A) is too good, and B) isn't fun. I think a lot of the players who think it is fun feel that way because it's overpowered, not because it's a really engaging activity. When your decision making consists of cycling through targets and casting PW:S on any target without Weakened Soul, then you aren't really responding to the encounter -- you're acting like a glorified macro. Now maybe some of you will argue that you take a lot more into consideration before executing a pre-heal, but the fact is that Disc priests can be unbelievably effective without doing so.

...An Icecrown parse for a Disc priest might have PW: Shield (including its glyph) and Divine Aegis at 80% of healing done. Imagine that is more like 50 or 60% of healing done. That's still a lot of power for shields. The remainder would be filled in by spells like Penance, PoM, Heal and Greater Heal.

Now that I have some numbers I can complain about broadsides like this. Suppose you are trying to keep a tank alive and conserve mana, but that just spamming Heal is not enough.

If I am trying to be mana efficient with heals, then I am definitely going to cast Power Word: Shield about every 15 seconds. If I get a Rapture from each shield tick then shield is about 26 healing per mana, while Heal, my next best spell, is 16.2 (more on the efficiency of Heal later) in Tier 1 gear. At final tier shield is returning mana each cast. Renew is, next to Heal and ignoring Rapture, by far my most mana efficient heal and my best healing per cast time, so I would presumably keep that up on the tank if spamming Heal is not enough. I could fill in any additional healing with Penance. Lets assume that's enough and I'm not resorting to Flash Heal to further up my healing per second.

In this scenario tank healing would make my shields about 18% of my total healing (counting both Power Word: Shield, it's glyph heal, and Divine Aegis). Obviously that is making not a lot of us of shields, just one on the tank every 15 seconds. But with a spell as expensive as shield (the first one is free, but each additional one costs the mana equivalent of about 3 Heals), how often can I realistically be casting it? To get my shielding up to 50% I would need to go up to 5 shields per 15 seconds, filling the rest in with a single renew, a single penance and 2 heals.

That would mean spending about 1760 mana per second instead of 1024 mana per second. It would also increase my Raptures, probably from about 1 every 17 seconds to 1 every 14 seconds. So the after rapture expenditures would be 1581 vs. 877. Replenishment is 1k mana per five seconds, so that reduces it to 1381 vs. 677. If we can manage a good 500 mp5 in combat then that would be 1281 vs 577. We get a shadow fiend that gives us half our mana back, so we have about 150k mana to work with.

In other words, if we get about 18% of our "heals" from shield then we can heal for nearly 4 minutes without a break. If we get 50% then we can heal for 2 minutes without a break. The former sounds pretty sustainable, the latter not so much.

But what is really troubling about that particular blue comment is that it is very disconnected with reality. I went and checked some logs. My shields are about 30% of my healing (though I cast renew when I am discipline, which most priests seem to think is crazy). A lot of discipline priests run around 50%. In heroic Lich King, which is pretty much the shield spammiest fight ever conceived I found numbers as high as 70%. Regardless of what people would have you believe on the forums, Discipline priests (at least ones who get anything done) don't just stand there and spam shield and right now it is our both our most mana efficient and our highest hps single target heal even if you ignore Rapture. It would stand to reason that when it is significantly less efficient than penance and much, much less healing per second than renew we will cast it less, not more.

The reason this is very important is that we have a mastery bonus, and it is balanced around how much of our healing comes from shields. If we get 50% of our healing from shields then our 20% more absorbs mastery bonus is nearly 10% more healing, and any bonus we get to that from mastery rating will be about .5% more healing per 1% mastery bonus.

If we get 25% of our healing from shields (probably a more realistic number) then our mastery bonus is half as good as that. It is very important that the developers have a good idea of how much healing is actually going to come from shields.

I think we all agree that is it less than now. It would be nice if we could agree on the reality of what it is now so that new numbers could be based on that.

First, Some Numbers

I want to complain about some of the numbers we've seen now that the NDA has been listed, but before I can complain about them I have to say what they are. I know that these numbers are not at all final. That makes whining1 about them all the more critical. It is much better to complain before things are live because they can be changed easier.

Before I really start complaining, though, lets look at some numbers.

The mana cost of penance: 3294
The cost of penance as a percent of base mana: 14%
Base priest mana at level 85: 23528

Stats on crafted pants: 341 int, 228 to two ratings
Total stats from a full set of gear excluding trinkets: 3403 int, 4550 in ratings/spirit
Trinkets: 197 int, an on use ability for 500 int, and we'll guess our other trinket can give the equivalent of 197 to two ratings
Guesstimated base int for level 85 priest (really a guess): 220
Stats from known enchants: 100 mastery, 65 haste/crit, 20 all stats, 110 int, 40 spell power, 30 spirit
Raid buffs that will affect int: Kings, Flask: 300 int, Food: (estimate) 60 int
Final int (presumed for now to equal spell power): 4740
Arcane int: 5401 mana
Total mana: 100,029

Guesses on scaling:
Tiers of gear: 4 (plus one phantom tier for heroic modes)
Levels per tier: 13
Level increase from tier 1 to tier 4 heroic: 52
Stat increase from 52 levels: 63%2

Lastly, we can include talents and abilities. Inner Fire is 1080 spell power at level 85, and presumably we will talent for improved inner fire since it doesn't look like we'll have anything else to put the points into, so that 1566 more spell power. We get 6% more maximum mana from enlightenment.

So, finally, here are the stats I'll use to look at how the cataclysm numbers will shake out for priests:

Raid Tier 1Final Raid Tier
Spell Power6,3068,800
Ratings/Spirit total5,0998311

1. Whining at WoW developers should always be in the form of constructive arguments if you want your whining to have any useful effect. E.g., "I think that your current criticisms of the twitch-reflex world of limitless mana healing are excessive in their downplaying of the skill involved and make it seem like people don't make choices because it doesn't matter much what they do. While I agree with going in the direction making mana matter, and I find healing low level dungeons where mana does matter to be more fun than the current high-level healing environment, you have to realize that people make the choices they do for very good reasons: it is because certain spells are much more powerful than others. If a spell is instant cast, is efficient and has very high healing per cast time, then people are going to cast that spell exclusively whether mana is a consideration or not. Rejuvenation and Power Word: Shield are both examples of this... WAAAHHH!"

2. In reality gear inflation is more than this because the first tier has zero or one gem socket and the last tier has two or three, which increases the numbers beyond iLevel increase. Probably a couple extra percent, and a very stupid couple of percent at that.