On World of Warcraft PvE Group Content
TL;DR - There’s not enough feedback in WoW for how you’re doing. Using the premade group tool to find groups for Mythic+/raids is really frustrating when you’ve not previously completed the content. Blizzard can solve both of these problems by analysing log data, giving you tips on how to improve, presenting a score/grade for how well you’re playing that’s publicly visible, and providing a dungeon finder queue for M+ that matchmakes using previous skill scores. Possibly also for LFR/normal raids. It’s a lot of extra compute, but could be cost effective if it improves player retention at max level.
I’ve been playing World of Warcraft on and off for 12-13 years now, doing PvE content almost exclusively. This is the third consecutive expansion that I’m about to stop playing within the first two/three months. I’m yet again spending all of my time standing in the capital city staring at the Premade Groups UI. As a DPS player who’s slightly behind the gear curve and doesn’t currently have a schedule that allows organised guild play, I’m less and less willing to spend ~20 minutes during peak times applying for groups before I get to do the content I want to. I’ve cleared multiple M+6 keys, but currently struggle to get accepted into groups for +4/5/6/7 dungeons. In other games, I can get into a game within 5 minutes. That’s a much better use of my leisure time.
The main problem I see is that there is no readily available heuristic for player skill within the game. When players are left to form their own groups without data on player quality, they will naturally reach for poor proxy metrics like iLevel, achievements for having already done the content or external rating systems like RaiderIO. This lets the group creator maximise their chances of success within the run, probably by selecting players with a higher iLevel than themselves or players with a sufficiently high RaiderIO score. This creates a system where it is difficult to get onto the PvE content ladder in the first place, and it’s also difficult to climb it without a huge time investment.
I think this is causing three further problems within endgame PvE play:
-
It removes challenge from the game - If the group creator is optimising for their own chances to succeed during the run and the easiest way to do that is by choosing players who have previously completed the content, then the group leader is probably getting carried by players with better gear than they have. When your contribution to the group doesn’t make a noticeable difference (e.g. you can die at the start of a boss fight without it having a major impact on the group’s chances of success, or you’re consistently the bottom dps) the game is much less fun.
-
It creates an economy for boosting - To get into a raid group 2-3 weeks after launch, I need the achievement for having killed the last boss. To kill the boss, I need to get into a group. You can break this cycle in a few ways, such as creating your own raid groups or organising guild runs, but the lowest effort solution is to get boosted. For a few hundred thousand gold, you can get the achievement and use it as a ticket into groups for the rest of the patch. You’ll probably also get a few items to boost your average iLevel at the same time. This may also make players more likely to buy gold from third parties.
-
It turns off new players - Players reaching max level for the first time already encounter a huge difficulty gap between the levelling content and the endgame content. With no measure of their skill and very little feedback in the default UI, it’s difficult to even know that you’re not playing well. This leads to problems like being kicked from groups seemingly randomly because the group perceives you’re not playing well.
I think you can solve these problems by letting players queue for M+ key ranges using the dungeon finder, and matchmaking using a skill rating system. Skill could be calculated by automatically rating player performance during PvE content using log data. Performance metrics should be gathered for all players during standard group play and be highly visible after a dungeon run, possibly with tailored recommendations on how to improve. I think this would improve retention rate for players after new content is released, make it easier to learn new roles (also boosting retention) and improve the chances that a new player will engage with the endgame content.
Note - I’m writing primarily from a DPS perspective. Healing and tanking have similar issues, but DPS is both my main role and the bulk of the playerbase.
Item levels and skill levels
It makes sense for players to measure the quality of their own gear and their peer’s gear to assess how strong they are. It’s the most significant way that a character can get more powerful in the endgame, and is the whole point of doing endgame content. The endgame content loop looks something like:
-
Do a thing
-
Get loot
-
Do harder things
-
Get better loot
-
Repeat for as long as you want to
With a loop like this, having both your individual item levels and your average item level easily visible is pretty crucial. You can assess the strength of your character at a glance and also see where your weak points are. This lets you target the content you want to do to maximise your chances of making a character stronger if you want to, so that you can do harder things more quickly. Similarly, it lets other players know if your character is capable of clearing the content. You probably shouldn’t be taking freshly dinged characters straight into M+/Nathria, as you’ll be pretty much useless (there are some notable exceptions, like when a full blue group cleared Ulduar in WotLK).
To quote a friend of mine though:
“I logged back on after two weeks and there was just no point. I was too undergeared to do anything.”
The main issue with using iLevel as a proxy for skill is that it’s a measure of character potential, rather the player’s ability. There’s no better place to see this than LFR, a queue I refuse to use at this point. You throw in a huge mixture of player skills and iLevels, all above Blizzard’s predetermined minimum iLevel, and get a cocktail of sadness. Some of the strongest classes will be bottom of the dps meter, regardless of how well they’re geared. I’ve seen players literally AFK through the entire run and just expect to be carried for free items. You’re probably going to wipe multiple times, especially if the boss has oneshot mechanics that haven’t been watered down from normal. All the players who have already cleared it (probably the ones who were previously carrying) will get very frustrated in the chatbox and maybe leave after 2-3 wipes. The whole cycle of wiping and shouting tactics in raid chat repeats 4/5 times, especially on final raid bosses, until Determination stacks high enough that it’s extremely difficult to fail. And this is with a group of people who passed the generous minimum iLevel threshold. In contrast, the players I know in real life walked into M0 while barely making the minimum iLevel threshold for heroic dungeons and cleared them without too much difficulty.
It’s not surprising that people AFK or intentionally underperform when there’s no incentive not to. You can happily hide in the DPS pack, then disappear at the end of the run without any negative consequences. Your performance isn’t being rated, you still get loot either way and you’ll never interact with any of the players in the group again. I can’t blame the people who /follow their way through LFRs to ensure they’re useful in guild raids or M0, but I don’t want them in my groups.
I think the smart people at Blizzard HQ are well aware of these problems and have tried to tackle them in the past. The semi-mandatory GearScore addon was replaced by the in-game iLevel counter at some stage (Wrath? Early Cata? It’s all a blur to me at this point), after it became so widespread. Similarly, I think the Proving Grounds challenges were attempts to give some kind of measurement of player skill independently of gear level. I think an opt-in system like Proving Grounds has a couple of problems though.
-
You have to stop doing the content that you want to do (e.g. an M7+ key, a raid) to prove to other people that you’re good enough to do it.
-
You do it once and carry the achievement through subsequent patches, and ignore balance or rotation changes in new content.
I think the only realistic solution is a performance monitoring system is always on and always giving feedback. Opt-in solutions mean that a portion of the player base is never measured, and I’d be willing to bet that it’s the lower performing percentiles who need the help most. One-off solutions tend to not age well.
On measuring skill
The default WoW UI has evolved a lot over the years, with the additions of things like WeakAura-style spell highlighting or the auctioning revamp. I’ve found the number of addons that I actually need has reduced significantly over time, even if I do still find myself reaching for things like ZPerl that aren’t strictly necessarily as a quality of life improvement.
It doesn’t give you nearly enough information though. For the player with no addons installed, all that you can see is a couple of numbers which float above a target’s head on hit. While this works for enemies with small health bars (‘oh Pyroblast hits WAY harder than Fireball’), it’s not effective for raid bosses which slowly tick down over time. It’s also ineffective for classes which primarily use DoT spells which are harder to track.
This makes damage meters semi-mandatory for DPS players. They give you a lot more feedback to work with (‘why is the other mage doing 3x my dps?’), as well as showing you the effect of your cooldowns in real time. They also happily aggregate your damage across multiple enemies and display it in one easily digestible number. I’d argue that if Blizzard was going to make any one change, it should be including a DPS meter or similar feature in the default UI to give everyone these benefits.
DPS meters can cause some pain though. Players try to inflate their numbers by cleaving when they don’t need to (e.g. Necrotic Wake exploding necromancer mobs). They can also become overly focused on increasing their DPS number and ignore other mechanics. This has led to the rise of more sophisticated combat log parsing tools like WarcraftLogs, which can ignore damage that was dealt to the wrong target in certain fights.
Tools like WarcraftLogs and WowAnalyzer are really the best available solution for giving feedback on what you’ve been doing. I’ve learned so much from comparing my log parses to those of similarly geared players, then optimising my talents or gameplay appropriately. The parse percentage number is a wonderful snapshot of how well you performed relative to other players too. The standard of community tooling (have you ever browsed through the SimC repo?) is incredible, and I’m consistently impressed by the things that are just available for free.
This smells a little though. To improve as a WoW player, I have to go to external, community maintained resources. More importantly though, I have to already know that those resources exist before I can go there. For something like log parses, this introduces a significant sampling bias. A 40th percentile parse in the uploaded logs might well be a 70th percentile parse among the overall population. Offloading all skill measurement and play analysis onto the community means that those who would benefit from it most are probably not aware that it’s an option.
So what are you actually suggesting?
A few things:
-
Create M+ key range queues (e.g. Random Mythic Dungeon 2-4)
-
Give players a PvE MMR based off log parsing of recent dungeons. Give skill points for doing well relative to your iLevel, take skill points for standing in the bad. Make it public.
-
Matchmake dungeon finder groups with skill as an input.
-
(Maybe) Apply similar strategy to LFR.
-
(Maybe) Add normal difficulty LFR queues
Most 5-man keys, at least up to +10, are easily accomplished by premade groups without any kind of prior communication. The criteria that players are using to build groups (e.g. must have at least one of the Covenant required for this dungeon, need a Bloodlust class, need a ranged player, want 3 MM hunters) are entirely automatable by a group finding tool. These might not make completely optimised groups, but it would give more of the playerbase a chance to experience more difficult content in a way that’s appropriate to their ability.
All of the data used by tools like WarcraftLogs or WowAnalyzer is readily available to Blizzard, even if it’s not stored at present. The game servers accept all of the keystrokes/spells and are the source of truth for what’s actually happened within the game. It could be analysed. Putting this directly into the game client makes it easily visible to everyone.
Matchmaking lets people actually do things while searching for a PvE group. You can work on your Covenant storyline, world quests, professions or transmogs while you’re waiting on queue to pop. This was one of the major benefits of the Dungeon Finder; you can play the game while you’re waiting to play a different part of the game.
Conclusions
I’ve really enjoyed the bits of Shadowlands I’ve played. I think that the 5 man content is excellent, but haven’t managed to fit a Nathria run in yet. There are a bunch of interesting mechanics like the Guessing Game in MoTS (reminds me of Lost Woods in OoT in a good way), and the game balance feels as good as I can remember. I just wish there was less friction in doing it.
I’d wager Blizzard haven’t done anything in this space because matchmaking with multiple variables and tracking a performance number is harder than just not doing that. Instead, they’ve opted to not solve the problem by providing the Premade Group tool. This might well be a completely reasonable decision. I don’t work for Blizzard, so I don’t have access to the amount of data they do. I don’t know how much PvE data the servers process in a day, I don’t know how many players actually stick the landing for endgame content, and I don’t know what any of the infrastructure behind the game actually looks like in practice. And there are only so many developer hours in a day. I’m sure that a lot of people who really care about this game have been thinking about problems like this for years in a lot more detail than I have. I’m also aware that I’m asking Blizzard to continuously process a LOT of game data, at no small cost. Similarly, maybe the ‘new player’ or the ‘low skilled player’ either doesn’t exist or doesn’t care about getting better. It’s really hard to say.
Ultimately, I’m unsure if any of this makes sense for a 16 year old game. I’m sure WoW and its game engine are stretched to the limit at this point, and it’s amazing that it still works as well as it does while still being one of the most popular games in the world. But I’m unsubbing again for now. I’m just writing this in the hope that when I inevitably return, the game state will be a little bit better.
Other stuff
What about queue times?
-
This is a legitimate concern. I imagine there’s a greater and greater surplus of DPS players as you move up the keys/skill levels.
-
Personally I’d rather queue for 20 minutes and have my brain free to do other things in game rather than spending time hunting through groups in the premade tool.
Why should performance profiling be on by default?
-
Tools like Warcraft Logs have massive sampling bias - you have to care to log
-
New players need the most help and are the least likely to install third party addons to help them improve, or even be aware of them
-
This shouldn’t be pushed onto the community. I think it’s necessary for improvement and Blizzard are best placed to implement. Potentially with open source community contributors who are already doing the work.
If you tell somebody they’re bad, they’ll get sad!
-
I think you can give feedback nicely. League of Legends has a good rating system of D-S+ depending on how well you did during a game (although it doesn’t actually show you how to improve)
-
If you give feedback like the LoL client rather than the LoL players, it’d be fine
But what about the community?
-
Based on the premade group finder, this is already a common way of playing the game
-
Gating M0 increases the barrier to entry
-
If anything, you’ve got more in common with players at a similar skill level
Why don’t you just make your own groups?
-
If I only run my own keys, I can probably run 3 dungeons total per week before I start failing them because I’m undergeared
-
I’ll naturally pick players better geared than I am and get carried
-
Especially for raid groups, making your own groups is time consuming, You’ll probably end up just standing in Oribos doing it.
(Writing style feedback also appreciated, it’s been a long time since I wrote anything resembling an essay)