Oct 28th 2011 3:44PM You're right, it's a complex system and we can't really tell WHY cataclysm was such a failure. We tend to pick out the things that bug us personally. Anyway, the numbers don't lie-- whatever the reasons may be, it was indeed a failure.
Oct 28th 2011 3:24PM Two months before Cataclysm, WoW had "over 12 million" subscribers. In March 2011, they said they lost 5%, which is a very substantial drop of 600k. Most recently in the June conference call, they reported dropping to 11.1 million, so we're looking at a drop of at least 900k subs. That's more than any other entire GAME. Rift doesn't have 900k subs total, and WoW lost that many subs in the past year.
Cataclysm was intended to bring in new players. That's why 3/4 of its development effort went to revamping the 1-60 zones. From that stance, it wasn't just a failure-- it was a disaster. Not only did they not expand their userbase, they lost ~8-10% of their total subscriptions.
As I see it, the moral to the story here is that users don't want rehashed content.
And guess what? Three of those new Pandaland dungeons are, you guessed it, rehashed content. Rehashed Scholomance and Scarlet Monastery.
They just don't learn.
Oct 6th 2011 5:43PM See, that's a great example of why the all parses dataset is superior. I would reasonably expect EVERY caster in the top100 to have a legendary.
Looking at balance druids (#1 in top100) it represents the top 100 parses out of 32,887 total. That's the top 0.3% of parses in the full dataset.
Oct 6th 2011 5:29PM I would still love to see this analysis with the full "all parses" dataset at both median (man on the street) and 80th percentile (elite player) ranges. Using the full dataset would support your conclusion in a way that the comparatively extraordinarily miniscule top100 does not.
Oct 4th 2011 6:02PM The problem with the top 100 is limited to its very small sample size. It represents the top 100 parses for each spec, which means that every caster has a legendary, rolling tricks of the trade buffs, the stars aligning for crits, great latency, perfect positioning, and so on.
Lets look at some numbers to see what that really means.
The arcane top100 [b]median[/b] is 41.5k DPS. If you switch that to all parses, the median is a mere 29.5k. The all parses [b]95th percentile[/b] for arcane on Baleroc is 37.9k DPS. The 95th percentile represents seriously elite players, and it's still way, way below 41.5k. You need to go all the way up to the 99th percentile to match the top100-- and would you really consider those numbers meaningful to the vast majority of your readers?
The spec score can only really be used to rank specs, it doesn't have any direct relationship to DPS. It's useful for ranking the specs, and nothing else. You need to go into individual fights to see the true DPS variance.
Your graphs are definitely not wrong, but the conclusions you're leading your readers to take likely will be.
Oct 4th 2011 5:30PM Excellent post, I just posted a similar comment before I read yours.
Oct 4th 2011 5:27PM Frostheim,
You made two mistakes here. The first was in using the "top 100" parses. The problem with top100 is the tiny sample size, only 100 per spec. If you want to look at the very best players, use the "percentile" dropdown with individual fights.
The median is simply the 50th percentile, meaning that 50% do worse and 50% do better. If you want to look at elite players, set all parses then choose the 80th or even the 90th percentile. Don't use the top 100.
Your second mistake was using the "overall DPS" charts. These charts do not account for gimmick fights like alysrazor or rhyolith. I could go into more detail, but for a better explanation just click the "what is spec score" link at the top of the page. The raidbots.com owner describes it better than I would.
Hopefully you take these criticisms to heart and rewrite the article, because, well, no offense, but it's just plain wrong.
Oct 2nd 2011 12:09AM Lots of specs get nerfed every patch, but that isn't because their resource mechanic is inherently broken, it's because they scaled too well, or people figured out neat but unexpected ways to use their abilities. I still don't see any direct link between the fury nerfs and rage.
Oct 1st 2011 9:32PM I don't understand why you feel rage normalization didn't work. Are you always rage capped now?
I don't raid on my fury warrior, but he's sitting at 361 ilvl, and he certainly is not constantly rage-capped. He does hit 100 rage occasionally, and that's when I heroic strike. But I'm not using it on cooldown.
You talk about inner rage not being useful for DPS warriors. Doesn't that mean rage normalization is working? Please explain your reasoning.
Fury is probably fine in T12 content, but once the spec gets 10% increased attackpower and better weapons, it will overtake everybody else due to superior scaling. I don't see that as a rage problem at all-- the solution isn't to change rage, it's to drop the attackpower coefficient and weapon damage multiplier on a couple abilities until fury's damage is where the devs want it.
Sep 29th 2011 2:03PM So was I. And not just tanks-- if I queue for a "random cataclysm heroic" I want a quick one like throne of tides, not deadmines, shadowfang, or the Zuls. They all provide the same reward, 150VP, so why would I take an hour when I could take 30m?