Exactly. You can put down stats of a particular thing that released but if you omit all the variables then you aren’t really predicting anything with it.
Predictive analytics needs a lot more information than just focusing on major patch release cadance. Otherwise you are gimping your own study.
About every 6 months
You’re conflating different variables.
This analysis is about the timing of major patches, not patch quality or the number of intermediate updates.
Whether the Fyrakk Assault or Flame’s Radiance had bugs doesn’t tell us anything about the interval between major content releases. That’s a different metric altogether.
And while the variation means you can’t predict an exact date, it doesn’t mean you can’t estimate a likely release window based on historical trends.
That’s a fair point about the change in Blizzard’s development model, and it’s why I wouldn’t expect older expansions to have the same predictive value as Dragonflight or TWW.
That said, saying the previous data “means nothing” is a much stronger claim than the evidence supports. A change in cadence doesn’t automatically invalidate historical data, it changes how much weight you should give it.
Also, the mean and median weren’t presented as proof of anything. They’re descriptive statistics used to summarize historical release intervals. Explaining why those intervals changed is a separate question from describing what they were.
And that’s an estimate, not an exact number. When we’re talking about statistics (in the mathematical sense), we’re dealing with probabilities, not certainties. That means estimates, ranges, and intervals with upper and lower bounds, not exact values.
I disagree with your methodology specifically because of the scope of the middle patches in this and the previous 2 expansions compared to the scope of middle patches in previous expansions, they’re regularly scheduled and have a reliable scope, compared to expansions before that where the patches would sometimes not get scoped in or released. You aren’t doing any sort of hedonic adjustment to account for the full scope and quality of the patches or doing any sort of analysis on the value of your subscription.
You’re just trying to reduce it down to one number, which is effectively the interval between new raid tiers.
As such, I’m going to make two claims and I want you to actually use your real brain to evaluate them.
1:) The WoW developers have never had a more productive dev team.
2:) your subscription dollars have never been more valuable.
I think we’re discussing two different questions.
My analysis was never intended to measure developer productivity, patch scope, or subscription value. It was intended to analyze the historical intervals between major content patches.
If the goal is to evaluate developer productivity or value for money, then I’d agree that you’d need a different dataset and a different methodology. But that’s a different analysis, not evidence that this one is invalid.
So you’re just collecting a single data point devoid of any context?
No, it’s called defining the scope of an analysis. Otherwise every study would have to explain the entire universe before presenting a graph. 
But in order to bake an apple pie you must first invent the universe.
You forgot to calculate the standard deviation.
“Forgot” implies it was required. It wasn’t.
Nice quote from Sagan. The difference is that statistics requires defining a scope, not explaining the entire universe before analyzing a single variable.
Now calculate the mean recycled slop frequency. I bet it’s through the roof.
That’s a completely different study, but I appreciate your enthusiasm for expanding the scope.
Exactly what is the point? Why is one measure of central tendency more or less important than another?
Ok, but you’re getting more patch per patch now than you were getting in TBC, Wrath, MOP, Warlords, Legion, BFA, and shadowlands, and on a more regular schedule.
Also your Dragonflight and Shadowlands calculations completely ignore the entire fated season they had, so your methodology even fails at the test you’ve set up for it. If you just want somebody to explain the current patch cycle to you using plain english, its “They have an 8 week patch cycle in order to deliver content approximately every 2 months with new raids every 6 months, and their internal goal is that if they do need to delay for something like a holiday that it not be delayed more than 2 weeks and right now we don’t know if that delay would result in a shorter interval for the interval patches or if they’d simply adjust their delivery targets, e.g. if they delay by 2 weeks and then deliver an interval patch in 6 weeks instead of 8 or if they’d just continue to deliver the interval patches in 8 weeks”
Then you can get into WHY they made that decision.
Because they answer different questions.
The mean describes the average interval, the median identifies the central observation while being less sensitive to outliers, and the mode shows the most frequently occurring interval when one exists.
None is universally “better”; they’re complementary descriptive statistics.
That’s actually consistent with my methodology.
I’m measuring the cadence of major patches from Dragonflight onward, because that’s when Blizzard adopted its current expansion model.
Fated Seasons and minor patches are different types of releases with different objectives. Including them would answer a different question than the one I set out to analyze.
Then its poorly defined and not fit for any sort of academic or practical use since those fated seasons were major patches.
I excluded Fated Seasons because they’re extensions of existing patches rather than new numbered major patches. They recycle and extend the current season instead of representing a new major development milestone.
Since my dataset is based on intervals between major patch releases, including them would be inconsistent with the criteria used throughout the analysis.
I think it is a mistake with BlizZard to have more middle patches while having major patches come later after expansion launch. I know in Legion the first major patch was very quick but at the same time the first major patch really doesn’t set the tone for many expansions. BFA first major patch didn’t really set the tone either but the second patch for BFA was in my view integral.
It is the second and third patch that tends to make or break how people perceive an expansion. Given since DF the expansion cycle is now mainly about 2 major patches (third patch is just a fated patch ). This is because of the shortened expansion cycle 1.6 months we can conclude that the second patch is taking the place of what used to be the third patches in older expansions.
Shortened expansion cycle sounds good on paper but it seems to be too lofty of a goal.
I actually agree that the second major patch carries more weight in the current expansion model. That doesn’t change the purpose of my analysis, though, which is simply to examine the historical timing of major patches rather than evaluate Blizzard’s content strategy.
Right but you’re discarding patches on whimsy, the fated seasons counted and were a major patch. Even if you’re only counting raid releases then sporefall would throw off your calculations.
What are you even comparing it to?