Seven-day parse aggregates:

Mage
So Selection Bias likely is a factor in these numbers. Good mages are moving to other specs to find DPS. Fire's actual DPS is probably a bit higher than the numbers suggest, but the same could be said for other middle of the road specs. I wouldn't consider it a significant enough to disregard the data which does show a significant drop in Fire DPS.

That said, looking at the data I wouldn't be surprised if we have one more small drop in Fire tomorrow. The higher % parses finally fell but I expected a larger drop there. I'm not sure exactly how Raidbots' timing works and when the nerf was live on all servers. I think we might still have a little pre-nerf data in there.


Rofl. Thank you Aftershock. I didn't realize I would have to explain what self-selection bias is after posting a wiki link to a very simple explanation of it, but apparently I was wrong.

Let me make it clear what selection-bias means.
Pre-5.0, frost sucked for PvE raiding. How many accurate and consistent parses do you think there were of the BEST mages in the game doing raids as frost to give their numbers? None. Because they all -SELF-SELECTED- to be fire.

I was simply saying that as word came in that "fire got nerfed," it's possible the numbers were skewed by the BEST and BRIGHTEST mages moving on to try another spec in an attempt to min/max their numbers, thus leaving only the mediocre mage parses as data.


You know, it's pretty funny that you would bring up the topic of self selection bias, and then in that last sentence you blatantly use it in your own argument.
Delusional:

Greg Street ‏@Ghostcrawler
@Velayna Not seeing Fire on bottom. Happy where Fire is so far. Many mages trying Arcane and Frost, skewing results of sites like raidbots.
Delusional:

Greg Street ‏@Ghostcrawler
@Velayna Not seeing Fire on bottom. Happy where Fire is so far. Many mages trying Arcane and Frost, skewing results of sites like raidbots.


It is very unlikely that anything the mage community does will change his mind. Players respecing to arcane and frost is something they are proud of, b/c it diversifies the class more. Too many mages played Fire before the nerfs.

@OP
Thanks for linking the data. It's great to see someone else passionate about the fire spec. But in all honesty it probably is time to move on b/c the goal of spreading mages to other specs has been achieved and you can not realistically expect Blizzard Entertainment to change their minds.

As for the argument about if it was nerfed as bad as everyone thinks/says it was.
It was nerfed enough to achieve the goal of scattering mages to other specs. Not much else to be said except it was planned and working as intended.
Curious, if GC's point is that players moving is skewing results how can he be happy where Fire is so far? Assuming they don't have some magical additional players that are playing fire where is he getting this data saying they're fine?

I suspect the scaling of Fire is still strong and its contributing to Fire looking "ok" at higher ilvl. But at low ilvl it seems week atm.

Anyway, I've been Arcane since the nerf which is going ok. Unfortunately I think the play style of it (Constant 6 stack) means its in for a nerf sooner or later.

Edit: He called out Arcane's spike as pseudo proof that the Raidbots fire damage is skewed. This argument is basically meritless, as they did buff Arcane in 5.1 and gave no one any incentive to try it by buffing Fire as well. The fact that some number of players are now trying Arcane post-nerf doesn't mean that there aren't still good players in Fire. There isn't any current data showing that the Fire results don't also contain good players.

Greg Street‏@Ghostcrawler

@JamaicanJasta Look at Arcane's curve leap skyward. Now considered they were not buffed.
Hahah both those tweets are absurd.

Arcane got buffed by 2 percent per stack. that's a 12 percent damage buff with the keep-stacks strat.
12/07/2012 04:02 PMPosted by Kaikou
You know, it's pretty funny that you would bring up the topic of self selection bias, and then in that last sentence you blatantly use it in your own argument.

Y... yes? That was the point. That self-selection might account for a large part of the change.

That said, is there any way to check how many parses for each spec we had? If the number of Fire parses dropped significantly, and especially if other specs suddenly became more popular (whether in absolute terms, or just relative to fire), that would seem to be good evidence for self-selection. Chasing the current best spec isn't limited to the best and brightest players, after all.
12/07/2012 06:07 PMPosted by Ravicana
You know, it's pretty funny that you would bring up the topic of self selection bias, and then in that last sentence you blatantly use it in your own argument.

Y... yes? That was the point. That self-selection might account for a large part of the change.

That said, is there any way to check how many parses for each spec we had? If the number of Fire parses dropped significantly, and especially if other specs suddenly became more popular (whether in absolute terms, or just relative to fire), that would seem to be good evidence for self-selection. Chasing the current best spec isn't limited to the best and brightest players, after all.


Fire still has more parses from the past week than arcane and frost had combined.

So yeah.
Well apparently he was referring to the hotfix, not the 5.1 buff.
Yea, its still not a particularly good point.
So Selection Bias likely is a factor in these numbers. Good mages are moving to other specs to find DPS. Fire's actual DPS is probably a bit higher than the numbers suggest, but the same could be said for other middle of the road specs. I wouldn't consider it a significant enough to disregard the data which does show a significant drop in Fire DPS.

That said, looking at the data I wouldn't be surprised if we have one more small drop in Fire tomorrow. The higher % parses finally fell but I expected a larger drop there. I'm not sure exactly how Raidbots' timing works and when the nerf was live on all servers. I think we might still have a little pre-nerf data in there.


Rofl. Thank you Aftershock. I didn't realize I would have to explain what self-selection bias is after posting a wiki link to a very simple explanation of it, but apparently I was wrong.

Let me make it clear what selection-bias means.
Pre-5.0, frost sucked for PvE raiding. How many accurate and consistent parses do you think there were of the BEST mages in the game doing raids as frost to give their numbers? None. Because they all -SELF-SELECTED- to be fire.

I was simply saying that as word came in that "fire got nerfed," it's possible the numbers were skewed by the BEST and BRIGHTEST mages moving on to try another spec in an attempt to min/max their numbers, thus leaving only the mediocre mage parses as data.
That doesn't invalidate the data.

If there are 1000 mages, and 400 went Arcane, and 400 went frost, that doesn't make his data invalid because he only has 200 Fire mgges to count.
It means that the problem has caused 800 mages to jump ship.
Some new tweats from GC today, makes more sense to me:

Greg Street‏@Ghostcrawler
@JamesFroggatt2 We're happy with it at the high end. Looking to make sure less geared Fire mages didn't drop too much.


That looks more in line with what we're seeing on the logs, with 10H and 25H showing Fire mid pack. I can't say I'm exactly happy that mid pack is what Blizz is happy with, but its a more reasonable position. My problem with it is if thats the case they have a number of ourliers to nerf then. Currently Afflic logs and Combat rogues are far ahead of others on the H bosses (Using the overall DPS numbers, using Raidbots spec score that changes, but Fire also drops to bottom 4 on both 10H and 25H).

And then given how long they let Arcane suffer I don't expect the low gear buff to happen anytime soon.

There's probably some more good data to parse out of Raidbots, but seems Blizz has made up their mind so I won't be spending any more of my time on it.

Join the Conversation

Return to Forum