upcoming rdm changes?

[Jollyy5]

Since there seems to be some confusion when it comes to data used for balancing:

90+ or 95+ percentile is used to see what the balance between classes is like, and to make adjustments, because at those high percentiles each class' kit is used to their fullest potential and gives us an idea what kind of DPS a class can conceivably put out, there's not much value in looking at low parses because there these kits are not fully utilized, and in some cases some skills are used wrong, or not used at all.

And I would argue that high percentile does matter during progression, because while you may cancel a cast or miss a GCD otherwise to focus more on mechanics, you'll still be trying to use your tools the best you can, especially after seeing a mechanic a few times.

[Gruntler]

Excluding the mistakes and missteps that make parses go lower to address balance means you've excluded the part of balance that includes a kit's vulnerability to mistakes and missteps.

Here, let's explain it better so you can understand.

The question 'Which data set is the best to look at for balance?' is the problem. You shouldn't be asking which data to best exclude, you should be looking at ALL the data, and then using the data as a WHOLE and asking 'what does this mean? What does this mean?' and so on.

The alternatives are ludicrous; they're suggestions by people who don't understand how data works and want to ask the most banal question and call it 'balance' instead of looking at -everything- and coming to a better, fuller, more well rounded conclusion.

It's like deciding what the best healer duo would be by looking only at the best parse; it doesn't tell you anything. On the other hand, if 60% of all the top 100 parses are the most played healer duo, you might actually find out the most played is not the best--if it's played 80% of the time, for example, it would be underperforming -tremendously- and it would indicate another healer duo, or even both, are outperforming it, even if the top 5 parses were all the most popular duo. This is data that you don't pull from percentiles, but rather by looking at the top 100 parses of everyone, and comparing to usage. No percentiles taken! See? ALL THE DATA, and a question you can't get an answer from by looking at percentiles.

Looking at median data can give you -actual- expected production, because the median is the production you can expect from a given unknown player. Looking at the range and standard deviations in the top end to the median can tell you what the return for further investment and mastery in the job is--more information staring at one data point, the 90th percentile, will not tell you.

Looking at low end data can tell you how capable the job is at the intangibles in clearing--the worse it performs at lower percentiles, the worse it can perform and still claim a kill. This is also valuable for balance.

See? There's information at -all- the data points that relate to balance. If there is information that relates to balance, there is no need to exclude it. Staring at one set of data permits you to make a single statement: 'This job seems to have the potential to do X damage in a capable hand.' This is a valuable statement, but it is not the only one that matters in balance, a much larger, broader question. So you need other information. Not 'can use.' NEED. Thinking otherwise is greyparsing in statistics.