WoW Hunters update: Deading the dead zone

Well this is how quickly it goes. Just my last blog entry was a class review of Hunters. In the meantime the mood in the hunter community has gone to a rather violent mood swing.

First there was an onslaught of upset posts, including very prominent hunters about the sentiment that the 2.3 changes did nothing to address hunters’ real problems.

Then Tom Chilton who is the chief original designer of classes posted under his “blue” acronym Kalgan in the hunter forum.

The mood swung from massive upset to ecstatic cheers. The news was that the dead zone will be reduced or removed.

Now I actually like the dead zone. It makes space matter for hunters. Still I understand the cheers. These really are the cheers of “help is finally on the way” for a class that had it rough since early Naxxramas.

To explain the dead zone: Melee range in WoW is 5 yards. But virtually all ranged weapon attacks (minus one talented shot for hunters) starts at the 8 yard range. This means that if a the distance between a hunter and his opponent is 0-8 yards hunters can use their very weak melee abilities between 0 and5 and virtually nothing between 5-8. Because one can’t do anything this 5-8 yard range is called dead zone.

The dead zone has various impacts. Is general need to be at distance. If you have a hunter at point blank range, they need to get to 8 yard distance and keep that distance to stay effective. So gaining distance is a crucial aspect of traditional hunter play. A core problem is/was the balance of classes to immobilize/snare vs. hunters abilities to remove or escape or outlast these.

 In the case of immobilizing abilities the effect was that indeed the hunter could be almost completely defenseless. The approach for mages against hunters was simple: Run to the hunter, frost nova, and nuke from the dead zone. Two possible counters: Scatter shot and Intimidation. Only other possible disruptive action: feign death. If any of these was on cool, the hunter was completely defenseless against a mage. Being temporarily defenseless isn’t new though. Rogue stun-locks or various fears are similar in game-play feel. The main difference here is that non of these were class-specific from the recipients point of view.

In the case of snares, this had an impact on hunter play against most classes. The strategy is simple. Run to the hunter. You can match normal running speed with a hunter. If you have a slow, you can usually match slowed speeds. Hamstring, frost shock, crippling poison, mind flay, chill effects all are easy favorites against hunters as are short-term stuns (maze stun for example).  All that is needed is keeping hunters from having enough time and excess speed to get outside 8 yards and stay there. 

Hunters have a few abilities that are designed to give range: wing clip, concussion shot, frost trap, freezing trap, scatter shot, intimidation. Of these only wing clip is spammable, but it requires to be in melee range and facing the opponent. It is single target. In 3-5 man arenas the distinct problem was that getting away when assist trained was very difficult.

Even if a hunter gets out of 8 yard range the range/melee efficiency difference means that a hunter wants to keep range. I.e. if opponents try to close in a hunter has to match this by running for distance. However, a running hunter has also a reduced damage potential. The main shot is arcane shot, which is on a cooldown and has high mana use. I.e. even keeping a hunter moving is a good way to take the main danger out of hunters.

Hunters are dangerous when static. The same is true also for mages. Though the relative efficiency when on the move is different and mages have a larger instants arsenal to make them dangerous while moving.

Removing the dead zone will mean two things: One is that casters cannot use a no-defense approach to hunters. When hunters will try to gain distance they get an earlier change to use  instants (arcane, concussion shot). This will add one more usable escape snare in some cases and more damage potential.

I don’t know yet but I might just miss the dead zone. Maybe I would have prefered a solution that gives hunters a long cooldown escape mechanism (ala blink in terms of group impact). But I’ll be very interesting to see this on the test realm.

WoW Class Design: Hunters

I decided to review hunters. I can’t say that I feel it will be a wholely balanced review, simply because I care about hunters too much. My main and first level 60 and 70 character was a hunter, it was my main raiding character for WoW 1.0 and TBC and I have tried pretty much everything with the hunter until late spring 07 when I stopped WoW for a while.

What is probably good is that I know other classes rather well too. I have 6 of 9 classes at 70 and I have been raid leading extensively hence contending with all classes at high level.

I don’t play my hunter much anymore, even since I returned. To understand why let me try to give a sort of post-mortem or rather my personal current review of the Hunter class design.

History

Historically hunters have seen two major revamps: Patch 1.7 (Hunter class review patch) and Patch 2.0.1 (TBC prep patch, which included all class changes). The reason why 2.0.1 was a more drastic change for hunters is evident by the fact that a main stat mechanism (the AGI to attack power scaling) got changed. That this was no small feet showed by the fact that Blizzard touched many items to adjust for this change.

Going mentally back to 1.7, hunters had a reputation for being useless in groups. In part that was unjustified. Rather hunters had it easier to accidentally cause havoc. One is hunters inherent need for space. The other is the fact that hunter pets are on defensive (not passive) by default and that pet handling is not so simple and added another add potential to the mix. Furthermore, traps were not as trivial as sheep and not as known as sap. Hence hunters had the reputation of bringing no crowd control and causing wipes. Part of this was based on what happened in groups the other part was just lack of using existing abilities (like consistently putting traps down). Traps were hard to use mid-combat, and required feign death+retrap acrobatics. Early on pets didn’t show as frames in the party making them for practical purposes unhealable, and additionally early pet survivability was rather low.

By 1.7 the hunter situation got markedly better. Pets showed on party frames, they got more survivability. Hunters got a generic buff with 1.7 which they needed and time was very good for that patch period. In fact overall hunters were fine until late AQ40 and Naxx raiding, when their DPS couldn’t keep up with other classes (even hybrid option classes like DPS warriors). Warlocks suffered similar issues. Both these classes began to being benched to make space for other classes. A distinct limiting factor for hunters was the lack of mana regens, as well as scaling of efficiency with itemization.

Beside the mana issue, itemization also led to another criticism of the hunter community: Lack of weapon selection and the lack of weapons meaning much in terms of upgrade relative to what melee weapons meant for physical melee dps, or subjectively even wand upgrades (raid level wants often read higher dps values than top end ranged weapons). But mostly it turned out that slow weapons were best, and that itemization didn’t deliver slow weapons often. By that effect a crossbow from Blackwing Lair (second raid instance) remained the best weapon through the full next instance and only got challangers late in Naxxramas. While new weapons dropped, due to the nature of shot rotations, they turned out to not be actual upgrades, certainly also contributing to the relative drop of hunters in the raiding game.

For this reason some theorycrafting hunters called for a new shot rotation in which diverse weapon speeds could be meaninful upgrades, and also that weapon DPS more closely represents what damage one will deal with the weapon.

Back then a hunter rotation was dominated by three spells, aimed shot, multi-shot and possibly arcane shot. The bracket of the rotation would be the cooldown of aimed shot, and the other shots would be multishot when available. Hunters get automatic attack shots (auto-shots). These are on a steady rhythm but are delayed when another shot is cast. In the old mechanism an auto-shot that was coming up while aimed was casting would chase the aimed shot immediately after it released. The reason why slow weapons were best (~ 3 second cast time) was because it matched the aimed-shot cast time best, hence minimal time delay for an auto-shot. With haste effect (that speed up weapons) the slow BWL crossbow came very close to hardly losing any auto-shot time, while a very fast bow at 1.5 second would lose a full auto-shot damage each cycle.

However the good part of this rotation was that the timing was spell-cast-time and global cooldown time driven. Because the auto-shot chaser was automatic to aimed shot there was no real way to squeeze oneself out of the auto-shot damage.

I personally felt that the shot rotation itself was good. The real problem was that itemization didn’t match the mechanism. A solution would have been to tune most weapons around 3.0 weapon speed (plus or minus) for raiding, and have a few fast weapons for interrupting.

Patch 2.0.1 and TBC

By the time guilds got to 4 horsemen it was rather blatant that hunters needed changes to help them in PVE endgame. Many boss kill shots showed 1-2 hunters and 2-3 warlocks at most, when balanced would mean 5 each. I certainly was excited that on paper the announced changes for 2.0.1 addressed hunter concerns somehow.

First off there was a new aspect in the leveling progression: Aspect of the Viper, finally an inherent mechanism to manage mana return. Second was the announced steady shot, advertised as the new boss of shot rotations. In turn aimed shot was declared to never having been intended in shot rotati0ns. Now when Aimed shot fired, it would no longer chase an auto-shot, but rather it would restart the auto-shot timer (hence add weapon-speed delay).

Additionally pets got another survivability boost, and they would finally show up in the default raid interface.

On paper this looked possibly good. In practice it turned out that Aspect of the Viper had too low mana return to alleviate the mana problem. The second problem was that another aspect, Hawk, combined with it’s talented improved version meant a drastic difference in DPS. When hunters started raiding it became blatantly obvious quickly that in order to keep up with other damage dealing classes, we could not leave Aspect of the Hawk, in fact we needed to pot heavily and chain chug mana pots to keep up with mana using classes who still had their mana return functions functional. Raid DPS of hunters before 2.1 was woeful. On the up side, the new 5-man heroic and 10-man instance design encouraged CC and make traps attractive. By allowing retrapping in combat, hunters desirability in 5-man became appropriate.

Still during leveling arcane shot, which was made to scale with gear, was nerfed down. Some argue that this may have been exactly the damage that hunters then missed to compete, but I’m not convinced.

At the same time first arena results came in. The results were stunning. Hunters were the least played class in arenas and suffered lower average rankings than any other class. Hunters weren’t just in trouble in PVE, they were also in trouble in PVP (even though they had a reputation to the contrary).

2.1

Blizzard was busy to complete lots of things, fix the consumables situation, complete black temple, retune the raiding game. Hunters did get a DPS buff. Before 2.1 a deep marksman build was the best DPS build to be had, with BM close. Survival tree was lackluster even for the low hunter standards. Not surprisingly there was a buff to both BM and Survival in 2.1. The survival tree got extra mana return options, and more damage potential, the BM tree got extra pet damage and pet survivability. Pets before 2.1 were very vulnerable to any AoE and finally got a talent to help mitigate that damage. Mend pet got significantly buffed allowing hunters to be more proactive in keeping their pets alive while not losing DPS. It used to require channeling hence taking hunter out of damage dealing completely. These were all good changes.

At the end, the tree that didn’t get buffed remained at it’s gimped state. Mana return was still an issue though, as only survival hunters got some extra mana return.

The real issue in many ways was however something that is somewhat unexpected: The shot rotation. The change of aimed to steady shot as the core spammable shot had some interesting implications. Steady shot has a 1.5 cast time (affected by haste). And since the cast time is so short and one can chain steady shots a new phenomena became apparent: How auto shot was implemented internally. Auto shot, while firing automatically, has an internal cast time of 0.5 seconds. This is to prevent hunters from auto-shooting while running. How the effect of this is that if a hunter spams steady-shot without pausing, they actually prevent auto-shots from firing completely. In fact any action, also a multi-shot or arcane shot can delay an auto-shot and it turns out that the timing of all these shots has very significant impact on the DPS output.

What’s more is that the relative timing of these shots matter, rather than global cooldown or long channeling times. I.e. if steady shot is on 1.3 and auto-shot is on 2.45 the relative difference from 2*1.3 to 2.45 in this example, which is just 0.15 seconds is what the player has to time for.

The result was that people with significant lag had very poor dps. Also people who just didn’t know how to precisely weave their shots for optimal DPS did suffer a substantial drop-off. Hunters inadvertently became the only class where sub-second timing because the crux of their damage dealing potential.

Soon types of work-arounds were found, mostly in form of shot rotation macros, which would try to prevent the hunter from accidentally “clipping” an auto-shot. These macros were sub-optimal as they relied on a server response to go to the next stage. For high lag hunters they still lost DPS using these macros. To perform optimally hunters needed to consistently weave their shots at these tight timing levels. Playing a hunter in raid at high level meant playing guitar hero on steriods for 4+ hours straight. And that still mostly to just compete with other classes who’d have 1.5 seconds minimum to consider their next button, often longer.

Basically what happened is that in an attempt to solve the Aimed shot rotation, a new rotation was found that had many pathologies that didn’t occur at all before. The fact that auto-shot had a cast time was inconsequential in an aimed shot rotation. So was latency and timing, as the auto-shot chaser was guaranteed and aimed shot was not chainable.

This is the mechanisms hunters have since 2.0.1 and it’s unchanged till today. I personally considered that mechanism broken the moment I realized how it functioned. It cannot be intended that in a RPG based setup, tight timing matters. Clearly this is unintended, just as the dominance of slow weapons in the old rotation was. But at least the old rotation wasn’t broken in the sense that it disadvantaged lagged players significantly. Also while in the old system people could perform better and worse depending on certain decisions, even a not very well informed hunter could perform sensibly. In the new system uninformed hunters (which are most players) will hopelessly underperform by default.

An attempt to illustrate the complexity of the shot rotation can be found on the EJ board. However, you have to add that haste changes with procs throughout the game, making these timing additionally unpredictable.

2.2

The main thing to mention about 2.2 is the change to Aspect of the Viper. It’s mana return utility has been improved. Otherwise 2.2 contains mostly nerfs to CC options in PVP, like duration of freezing trap and scare beast in PVP.

However, this is now 6+ months since the first arena results are in. Here are the current ones:

From Vhairi’s Sketchpad. Hunters are consistently the least played class in top ranked teams in all categores (2v2, 3v3, 5v5). One has to additionally consider that hunters are among the most popular classes to play still.

Looking at 2.3

Hunters are announced to get a ranged dispel on arcane, and mild scaling buffs to traps and stings. Unfortunately the scaling buffs are inconsequential. The range dispel has the potential to be both overpowered and useless.

The real issue is something different. Hunters have more positional constraints than other classes. The dead zone is a close proximity area in which the hunter cannot use ranged attacks (and not necessarily melee ones). In melee hunters are comparatively weak. So the main concern for opposing classes is simply to get close. The main concern for hunters is to keep or gain distance.

The dead zone is a brilliant concept and in my view one of the best concept in the hunter class design. It makes space matter. The real problem for hunters is not that space matters more to them than other classes, the problem is that we don’t have sufficient escapes. In order to keep distance a hunter usually has to move, this does reduce our damage potential significantly, which is very good, but the problem there is that in fact we do have insufficient methods to guarantee that we can even move.

What is needed to help hunters is more control over space. Now since 2.0.1 most CC options of hunters in PVP have been nerfed. This even though hunters are as bad off in PVP as they were when things first surfaced.

A second related issue is LoS, another great concept. All ranged casters have to handle LoS in arenas due to bridges and pillars. However, hunters have to combine the LoS issue with their space constraints. That is if a melee runs around the pillar to survive, we cannot just close in on the pillar to reduce the running radius to get the opponent into LoS, we additionally will be not able to deal damage the moment the opponent does get into LoS. Pillar running is additionally bothering for hunters. The problem isn’t LoS, it’s that gaining LoS doesn’t gain attack potential or ranged CC for hunters generically. Again, hunters need an extra control in that situation or pillar running against a hunter will remain effective.

None of the PVP buffs for 2.3 address any of these issues, hence it doesn’t appear that hunters will be any better of in arena PVP the next major content patch cycle.

Summary

Hunters are currently:

• Playing with a broken shot rotation mechanism, that is latency and micro-timing sensitive.
• Need additional ways to control space to be competitive in arena PVP but there is no help in sight.

Currently the hunter design has major issues and need fixing of what’s actually broken. Unfortunately the first makes endgame raiding not fun as hunter, the latter makes endgame PVP not fun. This may well explain why my hunter (and former main) of almost 2 years is parked and occasionally gets to farm materials.

It’s sad really because one core concept of WoW hunter design: – space sensitivity (death zone, trap positioning, flare spotting), reduced damage while moving, dangerous if left unbothered static – is brilliant.

Academia plays MMOs #2: How social is WoW?

Second installment on highlighting academic MMO research. This time an article that came out last year and focused on social dynamics of MMOs on the example of World of Warcraft. The article is Ducheneaut, N., Yee, N., Nickell, E., Moore, R. J., ‘Alone together?’ exploring the social dynamics of massively multiplayer online games. ACM Conference on Human Factors in Computing Systems (CHI 2006) April 22-27; Montreal, Canada, 2006, 407-416.

They looked at a number of factors and the main theme was overall time spend either grouped or together. For example they look at amount of time spend in groups split up per class. The result is possibly surprising for one fact: While warlocks and hunters are in groups the least, warriors are the third least grouped class. Of course they indicate that warriors, hunters and rogues are the most popular classes. Unfortunately the numbers are missing, so it’s not clear if warriors group less simply because classes are unevenly distributed and there is more “supply” than “demand”?

Grouping increases with level. As early as lvl 4, already 16% of time is spent in groups, which I thought was surprisingly high, given that there was no explicit group content. That rises to 20% before level 10 and to 30% around level 20 (when the first instanced content becomes available). Above level 55 the curve increases drastically and folks close to 60 spend close to 60% in groups. The authors make the claim that WoW becomes more social during end-game.

Of course at the same time, with level the involvment in guilds rises as well. 90% are guilded above level 43, compared to an overall average of 66%.

On guild play together they find that average guildies spend under 23 minutes playing together, while member of a tight core of a guild come in at 154 minutes per month.

The paper is really about the leveling game and about this the authors make the point that at least from the data it appears that WoW isn’t mainly successful because of its social aspects. The argument goes, that WoW is an addictive game that can and supports Solo play in its own right and social is possible but not the main determinant of the success.

Unfortunately the core arguments here are hypothetical. They argue that social factors do make a difference but truly social play isn’t one of those main factors, but mostly playing along among others, having factors like achievement (being uber or leet or both, showing off gear), hanging in chat channels while playing, or having shared comedic value of the game. Unfortunately these aren’t really supported by a systematic study.

It’s true though. WoW as a game is mostly a solo game while leveling. This is much stronger now simply because when leveling your alt, you may not have enough people in your level range to sensibly organize instance runs. That the solo-ability is an important factor also showed in Lord of the Rings Online, where people complained about solo content having dried up in the 35+ range before the first content patch that came out size, in fact some still complain.

But play style and sociability aside, are there maybe other factors that contribute to folks not playing together until level cap? One thing that always struck me was the difficulty of game support to form groups. In WoW we are talking about the evolution of meeting stones, LFG channels, LFG interfaces, the reappropriation of the meeting stone for something completely different, then back to some form of channel. An important point about instances is that not only do they take a committed block of time, they also require planning and organising ahead. In the WoW leveling days iching towards level cap, it wasn’t unusual to spam for an hour to form a group for certain destinations. Is the difficulty to organise groups part of the issue here?

Warhammer online has advertised (see WAR Production Video Podcast #7) public quests, which are quests that you can join at any time and do together. How that will work out precisely. If it is safe againstexploiting behavior and above all if it’s a truly “playing together” experience remains to be seen. But it’s an idea that cuts out the searching for a group aspects, because one can join when ready.

But for social there is a lot more to be said then during leveling. At another time, I want to write some about the social and raiding (some more). But not now.

Academia plays MMOs #1: WoW guild survival and churn

I should not be so surprised. Academics are also people and people play games. In fact academia has often been the first to embrace technological developments and often prefigure or even found the very concepts that later made a market. MUDs or other early online games often ran on university servers, and students and faculty threw their time into creating and running them. To study “games” was certainly frowned upon for long, and academic careers were not made by studying gaming. That’s for off hours mostly (and a lot of those).

This has changed though and the academic eye on gaming has become more acceptable and by gaming being a sizable and maturing industry and academia always having sought affinity with industry, the old taboos are falling.

But a lot of my negligibly small readership would never care about all of this so I’ll stop here and instead go on to the first in a series of posts in the spirit of “Academia plays MMOs #1” where I want to highlight some stuff academics look for in MMOs that may actually deserve being pulled out of heavy tables and statistical significance tests for more general consumption.

 Today: Nicolas Ducheneaut, Nicholas Yee, Eric Nickell, Robert J. Moore “The Life and Death of Online Gaming Communities: A Look at Guilds in World of Warcraft” in Proceedings of CHI, 2007.

For the uninitiated, CHI is the major conference on all things human computer interactions (CHI stands for computer human interactions, because people come first!).

These good folks looked at guild history over a year and some subaspects over shorter periods. Time is roughly June 2005-June 2006, so squarely during WoW Vanilla times. This was run across 5 different servers of all 3 main types (PVP, PVE, RP).

Here a few interesting highlights of what they found:

• Dominant guild size is 35 (90%), average guild size is just over 16. Anybody surprised why Blizz downsized to 25 man raids and emphasizes 5 and 10 man content more?
• Typical churn (i.e. people leaving and joining) for a guild, interestingly mostly independent of size is at an amazing 25% per month.  That is typical guilds are not all that stable! Not sure how my former guild managed to go with a much lower churn for so long.
• However churn does drop as a server matures, which makes sense.

Also they report analysis of dimensions that contribute to stability (i.e. doesn’t implode), and advancement speed of a guild. They find that a guild is more stable if it (in order of significance):

• Has a balanced in class population, is large, has a wide spread of levels, has small subcliques, spends a lot of time instancing, are well connected with each other.

 Most rapidly advancing guilds:

• Are small, have large subcliques, have high schedule compatibility, has a large number of subcliques larger than 3 people, is class balanced, has a wide level spread.

Of course all of this is WoW 1.0. Interesting that WoW TBC seems to merge both profiles, i.e. the raid size got reduced to match the more typical group sizes. Yet WoW TBC raiding goes through varied group sizes (5, 10, 25) hence demands large yet stable subgroups to be able to split safely. By having highly paced attunements, also high schedule compatibility is encouraged, even though this is a characteristic of the most rapidly advancing groups not one of the most stable.

 In some sense it seems to me that WoW TBC chose accessibility by size, but chose achievement by raid content format. But of course that’s just me guessing. Maybe in a year or two we’ll see some academics writing a nifty paper showing the actual impact of the TBC changes in hard, statistically significant numbers. Whee.