This post was originally written during Cataclysm. Its content may be outdated due to changes that come with expansion packs. Please consider this while reading.
Questions? Feel free to contact me – contact@sometimesatree.com. |
This is a follow up to my previous post, On Nerfs and Skill, Part One: Nerfs. I recommend you read that first if you haven’t already.
I ask that before you read this post, you understand – I am not trying to assign people ranks of any kind based on skill. Though I talk a bit about both individual and raid skill and how content should be assigned based on this, your personal progression does not assign you to a skill level – some people do not progress as far as they potentially could, and some people progress further than they would otherwise be capable (i.e. getting carried).
When we talk about skill in WoW, we talk about the combination of a lot of variables. Raid awareness and not standing in fire is one. Knowing the right spell for the right time is another – as a resto druid, this includes making the correct choices as to which heals to cast on whom. Timing cooldowns and abilities correctly is another. Switching targets at the correct time. Adapting to what the players around you are doing. There are a lot of variables. All of them are finite, as in there is a limit – a point at which perfection is reached. No amount of effort or change beyond this point will affect the encounter for the better (as long as it behaves according to design). Therefore, applying the Central Limit Theorem, skill in WoW is normally distributed.
Raid skill is also distributed normally, for the same reason. However, individual skill does not equal raid skill. There’s more to it than that including class/spec composition and buffs, interpersonal relationships, you name it. A player may be better or worse than his or her raid group. Ultimately, group performance is what determines the success of a raid. For example, a pug comprised entirely of above-average skill players with full clear achievements may not clear a raid as quickly as a regular guild group whose skill level is average, who have not yet downed the last boss.
I’m going to back up a bit. The below section contains a little background on what this normal distribution actually means. Feel free to skip it if you aren’t interested or are already familiar with the basics of what a normal distribution is.
In summary, this distribution is basically how things are distributed for a large population with finite variables. It tends to occur a lot in nature which is why it is kind of a big deal, and we can be pretty safe in our assumption that WoW skill takes this form. Other examples of this distribution appearing in real life include IQ, height, etc. Basically, the data is centred around the mean, and deviates equally in either direction.
The numbers on the x-axis (… -1, 0, 1, 2 …) indicate standard deviations, and the area under the curve between these points is approximately equal for any normal distribution (well it’s exactly equal for a perfectly normal distribution, our data approaches perfection with increasing precision as the sample size increases, but with a fluctuating WoW raiding population this necessarily deviates slightly). 68% of a population falls within one standard deviation of the mean, while 95% falls within 2 standard deviations.
So with that explained, let’s talk about who the content should be tailored to at which point during a tier of content. Please bear in mind that the population we are looking at are current content raiders – not the entire WoW population (which would include people who only do quests, collect pets and/or obscure achievements, or of course those who PvP exclusively; people who don’t raid don’t matter as far as tuning raid content is concerned, and due to the normal distribution holding, it doesn’t actually matter).
Below is what I think the system should be – who the content is aimed at for each “phase” of a tier of content (a phase being the time between changes to difficulty, which should be done in batches of patched nerfs, not hotfixes).
Please note: I’m not saying this is what the system is or ever was, just what I think the targets should be.
At the beginning of a tier of content, heroic content should be accessible only to about the top 5% of players, with completing heroic content accessible to possibly the top 1% of players. Normal mode content should be accessible to anyone within one standard deviation of the mean, however completing the end boss on normal mode is only accessible to raids 1 standard deviation from the mean or more. Players below one standard deviation of the mean are probably not going to be seeing much normal mode action – these are the people at whom the LFR tool should be aimed at. Until 4.3, their only option would have been the previous tier of content.
The timing of a nerf is the point at which progression drops off – we know Blizzard can tell when this is happening because they did it with Firelands. When it does, you scale it back one half to one third of a standard deviation. As I’ve said, Blizzard scaled it back too much in Firelands, which accounts for the huge disparity between before and after – people who were already on heroics began killing heroics at the rate of normal progression.