Sunday, January 25, 2009

Healing overhaul Part 2

Last week in Part 1, I wrote some ideas about how to address two commonly cited problems with the healing game in WoW: Healers are too focused to their UI, and not enough on the game world. There are a few other issues that seem to come up a lot. Tobold recently summarized one well, making probably the shortest post he's ever made, saying only the following:

The fundamental problem of healing

* If I heal 1 second too early, my spell is wasted.
* If I heal 1 second too late, we wipe.

It's an interesting statement, rife with implications. It's a bit of an exaggeration of course. Tanks don't normally go from full HP to dead in 2 seconds. But healers are in a unique situation here, and the need for such careful timing and vigilance are part of what make healing less enjoyable for some. A possible solution is also immediately implied: some way to make a heal that lands early still count for something. Before getting into that, I'd like to point out another fundamental problem with healing in MMOs that's somewhat related.

When working on new content, a raid must carefully balance its healing, damage and tanking to achieve the best result, because their second-to-best effort will be a wipe. Just a bit short on healing means the tank goes down or too much DPS dies to beat the timer. Too little DPS means the same thing. Of course once the encounter is beaten and the same group picks up more gear the fight becomes much easier. The healers that struggled to keep everyone up initially now find it easier because they have better longevity and throughput. But the tank also has better gear, meaning they have to heal less. And the DPS also has better gear, shortening the fight and meaning they have to heal for a shorter period. So they have the gear to heal for greater and greater amounts and longer and longer periods, but need to heal for less and less. The result is that the top end guilds usually bring extra healers initially and drop them as they move into farm mode. This is frustrating if you're one of the extra healers. But it's also frustrating to get better and better gear as a healer and to not really get to make use of it until the next tier of content. Note that tanks had the same problem in TBC, before Blizzard's "tanking overhaul" fixed it. Their best gear gave too much mitigation and not enough threat to use on farm content. It's frustrating to get new gear and not get to use it 95% of the time. DPS gets to feel their progression with each new upgrade. They get to slap each other on the ass with each new record crit they land. But healers pass a threshold and then are kind of just doing risk management. Healers almost never talk about record crit heals, because by the time we get the gear for them, we are almost never in a situation where they get to land.

So here's an idea. Let's make overheals count for something. If we give each player some buffer health, an overheal can get saved up even if it takes them above 100%, sort of like Living Seed. We have to make this count for less than a regular heal tho, or else we're just moving the problem further down. So let's make the buffer health over 100% decay over time, bringing them back to 100%. The higher you save up HP, the faster it decays. We could say that characters have a hard cap of 200%, over which any heals truly are overheals. But at 190%, it's practically on overheal too, because the health drop from 200% to 190% would be almost instant. At 110%, health would drop towards 100% in a few seconds, enough time to be a useful buffer against the next hit but not so useful that you'd want to try and keep everyone way up above 100% constantly. Unless you overgear the content and have such good regen as a healer that you have lots of excess mana and can afford to keep everyone up above 100%. And here's where your over-geared healers can feel useful again. Besides providing everyone with extra buffer health, let's also make all that additional health count for extra damage. If you have enough mana regen to keep your star mage up above 150% health during Heroism, he will get a 50% boost to his damage. Now the freak 15k crit heals actually mean something. Your mage will notice some huge hit land at the same time, and now the healer gets to partake in the ass-slapping as well.

Saturday, January 17, 2009

Healing overhaul

Blizzard design representative Ghostcrawler has made several comments beginning from around the WotLK beta about overhauling the healing mechanics in some fundamental way similar to how tanking was changed with WotLK.

To help illustrate what I think needs to change with healing, let me describe the following hypothetical encounter:

Your party enters a room with a dragon lying injured in the middle of the room. It is a powerful dragon with near god-like abilities, but near death. When you talk to him, you begin the encounter. Ghouls are raised from corpses lying littered around the entire room. They appear in random places, lots and lots of them. Most of them have very little health and immediately begin consuming the corpses around them to regain health. If any of them reach 100% health, they cast corpse explosion on a single player, instantly killing them. This is the only damage the raid takes. DPS must quickly run around killing them off before any of them gain full health. Occasionally an elite ghoul will also appear among them that starts with more health, regains health faster and also has a larger total health bar, requiring heavy single-target DPS. Meanwhile the dragon in the middle of the room must be healed to regain his powers. He has several million HP though and must be healed to full before 5 minutes or the ghouls begin spawning too fast to deal with. Once the dragon reaches 100% health he casts a powerful spell that destroys all the ghouls, ending the encounter.

Does this encounter sound familiar? It should. Does it sound fun? In particular does the DPS or healing sound fun? (For simplicity's sake, I've left tanking out of it.) Now imagine a variation that involves making some mechanical changes.

All DPS abilities for this encounter become 40-yard omni-directional abilities. Even rogues can target practically any mob they want and 1-1-1-1-1-2 away without moving or turning. Also the Blizzard UI allows any enemy mob to register as a target bar on the UI (sort of like Proximo) something which is currently not possible with the API. This means that the DPS doesn't need to target the mobs from the game screen, but can target them from the UI. To compensate for the ease and speed with which DPS can dish out damage this way, we will also scale up the difficulty, making the ghouls spawn faster and gain health much faster. Now does the DPS sound more or less fun in this variation? Much less in my opinion. I no longer need to move very much or be aware of the game screen.

Now what's the point of this hypothetical encounter? This is actually just a basic tank and spank encounter, similar to countless encounters in many MMOs. All I've done is turned it on its head. Damage is healing and healing is damage. The healers need to heal a single boss mod to 100% rather than DPSing it down to zero. Conversely the DPS need to deal with small and large amounts of health gains on multiple targets, preventing any from reaching 100%. This is exactly what the healers do in a typical encounter, but in reverse. The second variation used mechanics changes to make the DPS role match more exactly what a healer currently does. But the first version actually sounds more fun, zipping around catching ghouls as they spawn, alternating between AoE and single-target damage depending on the situation. It's a bit like the adds in Vashj, commonly cited as one of the best encounters of all time.

Every player that rolls a healer learns pretty early that heals are omni-directional and long range, and that it is much quicker to target using the party or raid interface than to target in the game screen. They might think it's a good thing when they discover this, imagining how frustrating it would be to have to find and manually target people taking damage. But I think this would only be a frustrating mechanic because people take enough damage that healers need to react almost instantly to prevent casualties. If the typical damage received was much smaller it obviously wouldn't be as hard.

I think TBC actually took the healing a bit in the wrong direction. I think the same thinking that led to long-distance, omni-directional as a fun mechanic that eases frustration further led to so-called 'smart' heals that target multiple players and heal based on who needs it the most. The problem is that the UI makes healing so easy that in order to make it challenging the damage needs to be ramped up to ridiculous proportions. Instead of freeing the healer up from his UI, the damage levels require them to be glued to the UI, spamming their beloved smart heals. I think maybe Blizzard is starting to turn around on this issue, as evidenced by the new 6 second cooldown on CoH and WG.

I think it might be interesting to go even further with the changes. Imagine removing and/or heavily nerfing most of the 40-yard omni-directional heals and giving healers some new powerful heals that require specific positioning. Some might require melee range. Some might be PBAoE, like holy nova and tranquility, but with much smaller ranges. Some might be short-range frontal cone heals. Would it be much harder and more frustrating to heal? Sure. But now tone down the damage and on top of that give healers some more ways to zip around like nearly all the DPS classes have and I think you've got an interesting new healing paradigm. Healers would be watching the game more than the UI. Healers would need to move constantly. My favourite fights have been the ones that require me to heal and move a lot (Archimonde might be the best example) and I've seem many people agree. Healers would feel closer to their charges, and less attached to the symbols on a raid frame. It might be a small thing too, but for me less damage feels a bit more realistic. The tank's health bar constantly going up and down from near-full to near-death every second with me magically repairing damage by watching all these bars kinda breaks immersion a little bit for me.

Now there are a few reasons that this would not be a simple change. For example, it's impossible to quickly target a single player on the game screen when many players can overlap each other in space. But it's possible to imagine game design changes to overcome them. It will be interesting to see what new directions the WoW healing overhaul takes us.

Saturday, January 10, 2009

The guild as a social contract

One of the most interesting things you observe in an MMO as opposed to single-player gaming is how the rules of the game inform the culture of the players. If the best loot in an MMO was available from solo quests there would not be nearly so many guilds, and the ones that you did see would be far less structured and more social. But since the best loot is obtainable only from 25-man raids, guilds exist (ignoring PVP in this example). This part is obvious, but the game design decisions of how loot and dungeons work all contribute to tendencies in your player culture.

There are a couple interesting posts about loot and fairness. Matticus tells about one situation that makes him question loot systems:

A paladin decides to take a break from raiding until new raid content because he already has all his gear from the current raid content, ahead of everyone else.

And Gevlon the Greedy Goblin defends the paladin's decision and tells of one of his own:

A player decides to quit his guild after an extended break in raiding due to holidays.

How do both these players make you feel? Is it fair for them to leave in these cases? Both situations incurred the disdain of the two players' guildmates, and spawned lots of comments from posters voicing their disapproval.

There is something very basic going on here when people feel an instinctive negative reaction to these situations. We are a social species. The rules of the 'game' we have played for thousands of years as a species are similar to the rules of an MMO. We can't do as well individually as we can as a group. A guy living in a shack hunting his own food won't do as well as someone in an organized village. We need families and communities in order to supply us with the best security (it's easier to defend a city than a shack), food (it's easier to grow and store food for a large group than for one person) and education (it's more efficient to train groups than individuals) and so on. We all know instinctively that we depend on the group we are a part of for our own welfare, and that we can all contribute to the group effort as individuals. Individually, one guy can decide that since everyone else is contributing to the group effort, he might as well sit back and enjoy the benefits of everyone else's work and secretly do nothing. Sure, everyone wants to get the most for the least effort. But we immediately realize that the entire thing breaks down if everyone decides to cheat the system. The village burns down and we're all living in our individual shacks like idiots. So what do we do? We all enter into a social contract, one so basic and universal that it rarely needs expressed in its most fundamental form: Do your fair share to help the group and the group will help you. If anyone tries to do less than their fair share, we immediately react negatively. It's a fundamental instinct. Some psychologists believe that cheater-detection is hard-wired into our human brains. Try this simple test on yourself to see if you agree.

This all manifests itself in an MMO like WoW. A guild is like a village. We enter into a guild because we want the best loot available and we need a group to get it. By doing so, we enter into a social contract so basic it rarely gets actually written in a guild charter: the guild will help you provided you do your fair share. When someone brazenly breaks this contract, such as ninja looting and gquitting, we react with immediate disgust so powerful it manifests itself physically in some people. We feel the same way every time someone isn't "pulling their weight" by being prepared with consumables, farming for crafted gear they could use, or showing up for raids on time. When someone tries to do less than the group but expects the same share of the reward, we instinctively feel negatively.

The two situations above are a little less clear-cut than a ninja looter. There are several factors in play and we might ask for more information before passing judgment. In the case of the paladin, he was the only holy paladin in his guild, which is not uncommon. Blizzard's current loot itemization means that holy paladins are the only spec out of 30 that want the gear they do. They are the only spec with such a unique set of gear choices and, presumably to avoid frustration, Blizzard has designed dungeons to distribute their gear way more than 3% of the time. So naturally the holy paladin will tend to gear up faster than his raidmates. (This is more generally true of tanks and healers.) So I think it's fair to feel that the paladin didn't hold up his part of the implicit contract: that the guild would raid the current tier of content as a group for gear. He didn't get geared up faster due to exceptional effort on his part, but due instead to the design of the game. He should not expect others to put in more effort than him for the same reward, which is what he's doing by taking a break and effectively saying "Thanks for the gear and good luck."

With Gevlon gquitting, I think it's unfair to decide after having got a lot of gear from your guild raids that the raiding schedule of the guild is suddenly not for you. However if things change within the guild and their schedule is consistently unfavourable for you, how long do you wait and how much effort do you personally contribute to trying to improve the situation from within? I think everyone accepts that there has to be a limit. Everyone's probably got a different opinion on that. I do find it interesting how often someone is judged by how they act in their final days as a guild member, once they have neared that point where the decision is made to leave. Someone who waits for one last piece of gear on which to "blow their DKP" before leaving will be judged much more harshly than someone who deliberately avoids taking gear, knowing that he is likely to leave. It's a question of whether you've "paid up" when the contract period ends.

It's interesting to see how a guild will structure their DKP system and guild rules around avoiding these situations. I think guilds would create better rules and structure if they start out by thinking of the guild as a community with a simple social contract.

Wednesday, January 7, 2009

Healing meters

Are healing meters useless? Ulkesshern seems to think so. I think he's taken a pet peeve here and overdone it. I agree that meters can be misleading, especially in the hands of someone unfamiliar with healing who glances at the meters after the fight and uses them as a barometer of skill. But he seems to be saying that the meters are completely useless and should never be looked at, except perhaps in extreme situations.

Now I deal with numbers a lot professionally and often come up against these 2 extremes of thought. One side looks at numbers and sees "big" and thinks "good" then sees "bigger" and thinks "better", without taking any context into account. A little information is dangerous, as they say. But the other extreme is to acknowledge that numbers in isolation are difficult to interpret, but then to basically throw ones hands up and say "well those could mean anything" and dismiss them as useless. Numbers are what they are. If you're unsure of the context, then yes they're meaningless. But the proper response should not be to dismiss them but to try and understand the context and use that to divine their meaning. Both extremes are the easy ways out and the path to true understanding, as usual, involves effort.

What do you think? Are meters ever useful? How much and when? I'd like to address his 4 reasons why healing meters are useless and offer my thoughts on the other side of the issue.

More Healing does not equal Better Healers
This is kind of a scarecrow argument. It's a very good point that healing needs scale inversely and very quickly with your raid's gear level and this will probably be the subject of a future post. But I don't think anyone looks at healing done on a fight and compares it to healing done in previous weeks and says "Our healers are doing less healing and are therefore getting worse." In fact I'll bet hardly anyone remembers the numerical amount of healing from fight to fight. In pretty much all cases, people are looking at the comparitive healing amounts from person to person.

Situational Situations are Situational
This is very true. Different healers can be assigned to different targets and types of damage and this will probably be a bigger factor in the meters than their individual skill. But that doesn't mean the meters are useless. At the very least, we can compare healers of the same class in the same situation and make meaningful comparisons. But even in differing situations, we learn a lot from healing done. Say you have 2 healers each healing a different tank and Healer A is doing twice the healing of Healer B. Assuming neither tank was at risk, we can safely say that Healer B should probably be assisting somewhere else in addition to his tank, either with another tank or with some extra raid healing. Every single class has a way to spread heals to more than one target without losing efficiency (gaining efficiency actually in nearly all cases). At this point if I were really interested in improving the healing group, I might even go into the breakdown of heals by target and see if someone was being too restrictive in who they healed. It's true that certain fights require healers be separated by too much distance to really heal anyone else. And perhaps all the meters can tell you then is how to better allocate your zone coverage.

Or, say you have 2 healers both on raid healing and Healer A is doing twice the healing of Healer B. Assuming no one was at serious risk of dying, this usually tells me that Healer A has got a consistently faster reaction time than Healer B. Looking at the overhealing and the breakdown might tell you if Healer B is coming in 0.1 sec too late every time, or if they simply got frustrated that Healer A was beating them to the punch every time and gave up, or even if Healer B was simply choosing the wrong types of heals with which to 'beat' Healer A. Incidentally, this is probably also telling you that you have more healers than necessary. This is usually a necessary evil for most guilds at some point, but it's still useful information if say next week, one of your healers and can't make it and you need to decide whether to replace them with another healer or a DPS class. You can bet however, that later when you do a harder fight that does need lots of healing, the fact that one of your healers is a little slower than the others will be important. A good raid leader will put them in a less critical position for the fight, and perhaps even look for a replacement in the future.

Some Classes outclass Classes
This is a similar argument to the previous. Just as situations give healers an advantage, so do class abilities. This is very true. But it's no reason to throw up your hands and discard the numbers. Again, at the very least we can compare healers of the same class to our benefit. We can also try and understand why some classes are better than others on the meters and use that knowledge to our benefit. Classes with healers that land on several players for small amounts over time should always do more healing that healers that tend to hit single players with large heals, because pound for pound HoTs are less useful than direct heals. As a druid healer, I should be doing more healing than the paladins because my heals are much less likely to save someone from a sudden damage spike than theirs are. Sure, I could switch to a glyphed Healing Touch druid and try to compete at that style of healing, but I've decided my strength as a druid is to lay a blank of HoTs out that will free up the paladins from having to sweat the small stuff so they can land there big tank-saving heals. The most useful type of healing, even better than direct healing, is a kind that doesn't show up as healing at all: absorbed damage. Pound for pound, 10k damage absorption is way more useful than a 10k heal. This is why discipline priests are always low on the meters. But digging a little deeper into the report, you can probably find the PW:S casts and come to a better understanding of what the numbers are really saying.

I will say this: If every healer is firing on all cylinders and playing to their strengths then in a reasonably difficult fight the healing meters should not show a huge variation between the top and bottom healers.

I'm going to add another "objection" here to healing meters that I think is a much better argument against them that he surprisingly did not bring up.

Meters can be padded
A skilled player who decides he wants to "beat" the meters can do tricks to bring his total healing higher. Some of the common things you can do are: using fast blanket heals way more than is efficient or necessary; pre-healing or pre-hotting someone who might be about to take damage; learning the habits of your fellow healers and trying to beat them to the punch on whoever you predict that they're about to heal. There are other sneaky tricks, some more devious or downright dangerous. But during trash or easy fights, its fun for healers to play these little game of trying to outdo each other. And I'll tell you why. Each of these little skills that lets you pad the numbers artificially in an easy fight will potentially save you in a truly wipe-threatening situation. So you've learned to hot up everyone that might take damage to pad the meters? Well guess what. Now it's Sapphiron time, and everyone is taking damage and you had better know which kinds of hots and how many you can most efficiently spread around. You've learned to predict what your fellow healers are about to do? Great, because the DPS just pulled 2 extra trash groups and now the shit's about to hit the fan, and if you guess what everyone around you is about to do, you might just come out of this alive.

Trash and easy fights are when good healers hone their skills and reaction times, playing little games with themselves and each other. When push comes to shove, those skills will save the raid.

Meters are not infallible
More hand-waving. "I've seen variations in meters before so how can anyone say if they're correct? Might as well ignore them." No. If the meters don't look right to you, pull up the underlying numbers until you find out where they're wrong. WWS allows to actually inspect the entire log the report was built from, and Recount allows a good deal of drilling. If something is out of whack, there's an explanation. These numbers aren't just pulled out of the ether. I agree, things go wrong all the time. A log file could be corrupted and missing pieces, or the logger may have been absent for part of the fight. But better to learn how to get a better quality of logs than decide that since quality varies, all logging is useless.

Monday, January 5, 2009

Achievements and immersion

Matticus is discussing what he thinks the ideal dungeon design would be, beginning with a look back at what worked or did not work in previous WoW dungeons. I think his list is good. Regarding storyline, I think Naxxramas missed a big opportunity by failing to have a tie-in to the world around it. Perhaps they felt that since it had already been part of vanilla, it didn't need a new storyline, or perhaps they didn't want to create quests that people who don't raid had no hope of completing, as Tobold suggests. I think in the spirit of making the entry-level raid game more accessible, you'd especially want to have quests pointing to it. Neither of the other raids, Eye of Eternity and Obsidian Sanctum have quests so it seems like a deliberate decision rather than something that just got cut on the way to release. While questing in dragonblight, frequent reference is made to the coming of Naxxramas and it's impact on the region. But players deal with that impact on the ground in the mid-70s before they even have flying mounts, so naturally no one suggests that these players actually try to get in to Naxxramas. However, I think it would have been really cool for level 80 players to have a quest send them back to the old Dragonblight quest hubs. On their return, the old quest-givers would recognize and welcome the players as returning heroes. However, you have come now not to perform rescue operations, not to defend, and not to take orders. You are now leading the charge, in a direct counter-attack, with the intent of utterly destroying the threat hovering in the sky above.

It does seem as though there is intended to be a clear divide from the leveling/questing game, and the heroics/achievements game. Achievements must necessarily exist outside the game, unreferenced by NPCs or quest text, since they would break immersion. So quests never point to anything beyond merely defeating bosses or recovering objects or NPCs in the dungeons. Unlike TBC, no quest that I'm aware of ever requires heroic difficulty. And although there were a few achievement-style objectives that existed in TBC (rescuing Millhouse or the prisoners in Shattered Halls) this style of play has been moved squarely outside the realm of quests and into the non-game realm of achievements. This all makes sense from the perspective of immersion. What exactly is happening in the game when you switch the dungeon to 'heroic'? No attempt is made to explain this, and none come to mind that don't strain the imagination. Similarly, the hard-mode style achievements can't really be explained. Rescuing prisoners within time limits, was all fine and well. But WotLK expanded the whole idea of hard-mode achievements into weird things like deliberately doing encounters in more difficult ways, or with fewer players, something that you really can't, nor should, try to explain in-game. So perhaps the decision was made to make the entire end-game sort of a separation from the leveling/questing game which you basically finish and 'win'. Raiding and achievements are a new type of leveling that don't really exist in game, since they all imply that you are doing the same encounters and killing the same bosses over and over, something which no longer fits in the immersion of the story line.

Interestingly, Ulduar is slated to expand even further the idea of 'hard modes', introducing special loot for doing encounters in harder ways. The Zul-Aman timed runs and the Sartharion drake runs already did this, but these were mostly cosmetic mount awards. The Ulduar hard modes, will give real loot, not mere mounts, and not mere achievements. I like this idea, since it's in keeping with the idea of accessible content. Practically anyone can see the content, and defeat the bosses, but real tangible challenges and rewards are held out for players that want them.

Sunday, January 4, 2009

On healers and tanks

I've often thought of publishing a gaming blog, generally having much to say, and appreciating the structure that writing imposes on one's thoughts. Seeing as this is the first post (and therefore statistically speaking, the only post) of my fledgling blog, it will be somewhat general and unfocused. In the hopes that I continue to publish, that will mean that I can lay out some of the general groundwork upon which to build and focus. So here's a quick overview of what I've been up to in WoW.

For WotLK, I decided to reverse the roles of my two main characters. My Protection warrior, long time guild main tank, is now my alt. My resto druid became my main raiding character, and was therefore the first to level to 80. This decision was based primarily on guild needs. A large number of people had decided to switch to a tanking character, due to the new Death Knight class and the general changes made to the tanking role to make it a little more enjoyable. Conversely a number of healers decided to switch out of a healing role, which seems to be pretty common. Healers tend to burn out faster and want to try something else. So we were left with an over-abundance of tanks and a shortage of healers. I enjoy playing both characters and decided I wouldn't mind at all being a primary healer for a while. The promised upcoming dual spec feature meant I would also be able to try a completely different role, probably as balance, without a large investment in gear.

So far I've enjoyed the druid and have healed every raid encounter in the game in 10 and 25 mans. The druid is a very strong healer right now, with an impressive arsenal of heals added to their kit. There are very few fights where I do not end up with the largest amount of healing done, usually by a fair margin. Some fights where there is very little damage are hard to stay on top, since other classes have faster blanket heals (CoH primarily) but since these fights are trivially easy to heal, I'm not really too concerned with who has the most through-put. On harder fights, primarily Malygos, Sapphiron, and Kel'Thuzad, I pull way ahead due to sheer length of the fight and the amount of incoming damage. I have incredible regen capabalities and can churn out a lot of healing without worrying about mana. I give my innervate out to the mana-hogging priests on these fights, and rarely need even a pot for myself.

Healing is a challenging role. I tend to think the most challenging. Tanking I really believe is pretty easy. The hardest part of tanking is done before you ever log in, researching various statistics on gear and stats to determine optimal avoidance and EH and threat for each fight. WotLK has even simplified these elements, making it a pretty simple matter of picking your best gear, then managing your cooldowns. In this respect its pretty similar to the role of DPSer, with just a little added vigilance needed in reacting to damage spikes. The hardest part of tanking is actually nothing to do with tanking per se. The tank ultimately decides every pull, and controls the flow of the raid. It usually falls on the tank to be familiar with every pull, every mob, and every boss. Not only from a tanking perspective but a holistic one. It's possible to raid lead as a healer or DPSer, but one of two things will happen. You will slow the entire raid down a great deal communicating strategy to your tanks or you will find your tanks naturally picking up the mantle of leadership in all the moment-to-moment decisions, leaving the raid leader with only overall strategy decisions or leadership decisions with respect to their role of either healing or DPS, which is really how it should ideally be, with decision-making and leadership the shared responsibilty of about 3 people.