Wednesday, February 16, 2011

Are loot distributions fair?

I don't know about you girls and guys, but for some time now I've had a strong sense that loot distribution is somewhat skewed. It just doesn't seem fair. I just feel like some loot drop alot more than it is needed. It all started back at the beginning of Wrath. I started noticing that some classes seemed to gear up alot faster. Not because they raided more than others, or there were less of them, no. But because their loot dropped alot more often than that of everyone else. In my case I'm looking especially at caster shamans and holydins. Also the tier token of druid/dk/mage/rogue drops alot more often than that of everyone else. Even when taking into account that there are ~30% more people that use that token compared to everyone else that token outdrops all the other tokens by far.

And I'm that kind of person that when I find something doesn't seem right, I start collecting evidence. I have to make sure that this isn't because of some idea I've gotten into my head, but that there actually might be a difference. So I started recording all the loot that dropped in our 25mans into different categories. How did they turn out?

I want you to note that these categories are far from perfect. I did clump some types of gear together just to make it easier to categorize, for instance I didn't put cloth with spirit as healer cloth, because there are actually some casters that use them (unfortunately healers still can't use hit, which means they take our stuff and we can't take theirs >:( ). Also when it comes to weapons I've categorized all weapons that lacked an avoidance stat as a dps weapon, although dks can use these for tanking (and other tanks as well in some cases). So this list is far from perfect, but it might still tell us something. In some cases it's easy to separate between casters and healers. Like with trinkets. They've often got "when doing damage" or "when healing" meaning they can't be used by the wrong type of role. Same goes with tanking and dpsing. I had to make a fairly easy system for myself because we often start clearing trash while disitributing loot, and people expect me to keep them alive, not collecting data (silly people).

It's difficult to say in which manner Blizzard take the prevalence of a class into calculation to design the prevalence of a loot type. Does the fact that paladins and druids are more popular affect the chance at which their loot types drop? I don't think so, but I do think however that hybrids have a slightly higher loot percentage than pures. All rogue specs can use the same type of loot, but druids use three different loot types, so maybe there are a little more loot for each hybrid spec than just a 1 on 1 ratio. But in the end I don't know, so all I can show here are if there happen to be great differences. There are after all some classes that use their gear type exclusively, like holydins. They don't share their loot type with anyone, not even a little, so one could imagine their loot type wouldn't be very prevalent. But it is! For some reason.

Let's take a look at the class distributions at a 1 on 1 level.
There are 10 classes, 4 different roles (counting melee and ranged dps separately) and 4 different armor profficiencies.
4 Healers (1 cloth, 1 leather, 1 mail, 1 plate)
4 Tanks (1 leather, 3 plate)
5 Casters (3 cloth, 1 leather, 1 mail)
7 Melee (counting hunters as melee since they use those stats) (2 leather, 2 mail, 3 plate)

4 cloth users
5 leather users
4 mail users
7 plate users

Here is my list of loot drops based on 156 drops all in all over approximately 27 bosses killed in 25 man (there are some 10 man in there as well, since the loots are the same). The numbers are rounded towards their closest one decimal.

Cloth
Caster 17 (~11% of total) (100% of the gear type)

Leather
Caster 4 (~2,5%) (29%)
Phys 10 (~6,5%) (71%)

Mail
Caster 12 (~8%) (67%)
Phys 6 (~4%) (33%)

Plate
Tank 7 (~4,5%) (24%)
Dps 11 (~7%) (38%)
Heal 11 (~7%) (38%)

Trinket
Tank 2 (~1%) (9%)
Casterdps 11 (~7%) (50%)
Heal 1 (~0,5) (5%)
Phys 8 (~6%) (36%)

Neck
Caster 3 (~2%) (60%)
Phys 2 (~1%) (40%)
Tank

Back
Caster 4 (~2,5%) (80%)
Phys 1 ~0,5% (20%)
Tank

Weapon
Caster 6 (~4%) (20,5%)
Phys 16 (~10%) (55%)
Tank 2 (~1%) (7%)
Bow 4 (~2,5%) (14%)
Wand 1 (~0,5%) (3,5%)

Shield
Healer 3 (~2%) (100%)
Tank

Offhands 3 (~2%)

Ring
Caster 6 (~4%) (55%%)
Phys 3 (~2%) (27%)
Tank 2 (~1%) (18%)

Tokens (Only drop from end instance bosses and aren't counted towards the total)
Dk/Mage/Druid/Rogue 8 (80% of total token count)
Priest/Paladin/warlock 1 (10%)
Warrior/Hunter/Shaman 1 (10%)

Is there anything we can say from this?
First of all we have to make two things perfectly clear - this is a rather small sample, and as many other things in WoW loot drops are under the effect of luck. It is possible all these numbers have nothing to do with the "real" distribution of loot, and that I'm just unlucky. Well me and the people I happen to raid with. Remember I said I thought, already since back in the beginning of Wrath, that some loot dropped more than others? Have I been unlucky for 2 years? And has that bad luck rubbed off on the vast number of people I've come to raid with during these two years? Eventhough there is an element of luck and my sample isn't great, I think it says something. It points towards something. And in a game, if some people have bad luck all the time, there really should be something done about it imo.

In most cases the numbers might be interesting but the variance isn't big enough to warrant a comment. These can be ascribed to bad luck. Let's look at the things I thought were out of place before I started gathering these data - have I found something to warrant my suspicion?

There are way too many holydin drops;
I really dislike how Blizzard have designed holydins. It is one class with one spec, and they have their very own loot. In my opinion holydins should use caster mail, but that is another discussion. In any case we can probably agree that holydin loot shouldn't be very common, considering the above factors. So how common is it? Well in my guild, 38% of all plate that drops (so far) is holydin loot. That means that out of all the plate that can drop, 38% is usable by one class, one spec only. This is even more annoying to us since we don't even have a main spec holydin at the moment (feel free to apply! *cough*). Do holydins make up nearly 40% of the plate users in a 25 man raid? Not often. Are they as many as the dps dks, paladins and warriors? Nope, and yet their loot type drops as often. Fair? Not so sure.

There are way too many caster mail drops;
I feel like gearing my shaman is way too easy. Mail drops everywhere, both melee and caster, but mainly caster. Do my numbers point this way as well? Well yeah, but not by a big margin. 67% or 2/3 of all the mail drops to us so far have been caster mail. Do caster shamans make out 2/3 of the total mail users? No, they make out half - ele/resto shamans vs enhancement shamans and hunters. Still, the difference between the drop rate in these numbers aren't big enough to warrant any whining from my side. Maybe just a little.

The only damn token that ever drops is the dk/druid/mage/rogue one;
Throughout the nearly half year of raiding I did in 25 man ICC, I saw 2 or so priest tokens drop. Remember we killed all the end bosses (except LK) each week, so we had plenty of drops. The hunter/shaman/warrior token dropped about as seldom. We might be extremely unlucky, but in that case we have been so for more than half a year. And what do the numbers from Cata show us? That the dk/druid/mage/rogue token drops 8 times more than the other ones. 80% of our token drops so far, have been for dks/druids/rogues/mages. There is one class more that needs this token than with the others ones, but not 8 times more. 80% of our raid group does not consist of either a dk, druid, rogue or mage. In a typical raid these classes make out some 45% of the total. Not 80%. Priest/paladin/lock make out about 25% of our raid, not 10%. I'd like to think our raid compositions are pretty standard (although as mentioned we lack a holydin. Apply!) and I don't think dk/druid/mage/rogue make up 80% of anyones raid (maybe in 10 man occasionally, damn druids are everywhere!).

There are other stuff as well, things I hadn't thought about much before. There seems to drop way more dps trinkets than healer/tank trinkets. But there are generally alot more dpsers than healers/tanks in a group so that's not so odd (although 9 to 1 does seem like a little too much). Overall does tank loot seem to drop very rarely, but then again, tanks only make out 1/10 of a 25 man and 1/5 of a 10 man, so naturally the tank drops will be low. 3% of the total drops seems very low though.

Where am I going with this? Do I have a point? Maybe not. I wanted to see if my hunch was right, and unfortunately it seems to be to some extent. Is there something I can do about it? Besides whine? Not much. I could make Blizzard aware of this, but I am pretty sure they know of their distribution rates already, and maybe there is some guild out there that is taking all our priest tokens as I write this. Are you? *evil eyes*

10 comments:

  1. A nice breakdown of local stats. I'm wondering ... could the data from WoWHead be leveraged to supplement your research? Might need some mathemecations to normalize the data, but it almost looks possible (says he that has not yet actually tried all this hifalutin' theorys). It would at least give some feel as to whether your initial perceptions are on track, or not.

    Still, it's nice to see a methodical approach to the problem instead of just 'gut feeling'.

    ReplyDelete
  2. @Grimmtooth
    Love asked me to check wowhead too and I was like "do you know how much time that would take? o.O" Yeah that would be my dream, but it would take forever probably ^^
    And it's just hilarious - I have a personal WoW saying that goes "As soon as you whine about something, RNG will make you happy again". This is really how WoW is set up because if the game never pleased us we wouldn't play it. If it pleased us too fast we'd be bored and stop playing too. What it does is that it teases us into near nerdrage frustration (just think about all those poor people doing archaeology) and just when you caps "I HATE THIS GAME!" in guild chat, you get what you want. I nerdrage about not getting about priest token on my blog and when we killed Nefarian tonight he dropped 3 priest tokens out of 3... Ironic.

    ReplyDelete
  3. While your idea to have holydins wear mail makes a lot of sense in terms of loot distribution it also streamlines things a lot and removes a bit more of the individuality to each class that makes things interesting. So, personally I would be against it, even though I think it would be more fair. And no, I don't have any holy paladin char. :)

    Regarding your data, I think the sample size is way too small to actually give us any sort of conclusion here, but you could use the WoWhead data instead...
    Check this: http://www.wowhead.com/items?filter=minle=359;maxle=359;cr=148:148:148;crs=5334:5094:5638;crv=0:0:0;ma=1#0+12+1

    Now, to tell horrorstories about loot! :)
    During the year we raided Icecrown Citadel, the bow from Deathwhisper dropped once on the first day when I wasn't there and then didn't show up again for 2 months! And then when we got into heroic mode a few months later the bow from Deathwhisper actually NEVER dropped between summer and til' the end of the expansion. I was like this: AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA!!!

    So, there you have it. :)

    Cheers,
    Gav

    ReplyDelete
  4. @Zinn - Well, I do think that would be a great use of it, but I do agree, lots of work. If I didn't already have like eleventy-zillion irons in the research fire right now, I'd be on that like Bieber onna Peep.

    (I'm a bit of a spreadsheet geek - there must be something wrong with me.)

    I'm glad your tokens dropped on Nef. Paging Alanis ... :)

    ReplyDelete
  5. Nooo..., don't condemned my little Holydin to that hooky spooky stuff. She loves her plate, she's polishing and buffing it everyday to be the shiniest around and will be very unhappy when she have to wear sorceress leftovers :(

    What's fair and what's not. I've more problems with people who need on things not suited for their role they are in that moment with the excuse that they can use it for that other role they do sometimes.

    ReplyDelete
  6. Not only is your sample size statistically meaningless, but your loot distribution isn't going to be equal, unless you killed every boss the same number of times.

    You should go to wowhead, figure out the odds of each item dropping, and then compare the odds of each type of item dropping in a "full clear." Anything less is going to lead to horridly skewed results that have no validity or use.

    ReplyDelete
  7. @Anon
    I am sorry if you missed that this post was more aimed at being fun and/or interesting than statistically correct.
    I did say "Is there anything we can say from this?
    First of all we have to make two things perfectly clear - this is a rather small sample, and as many other things in WoW loot drops are under the effect of luck. It is possible all these numbers have nothing to do with the "real" distribution of loot, and that I'm just unlucky." which I hoped would clarify this!

    Anything can have validity and use, it completely depends what you want to use it for.

    ReplyDelete
  8. I had a whole big reply typed out for you, but it was unneeded antagonistically argumentative, so I'll be succinct and try to avoid being an ass.

    Drawing conclusions from bad data isn't partial confirmation of your belief - it's a description of what your last few months of kills have given you, with no credible extrapolation to loot in general.

    ReplyDelete
  9. @Anon
    Well like I concluded - "Where am I going with this? Do I have a point? Maybe not. I wanted to see if my hunch was right, and unfortunately it seems to be to some extent. ". It's still just a hunch.
    I've already stated in the post that it is based on nothing more than a personal "feeling", that the sample is small, and that a point probably can't be drawn from it. So when you say "it's a description of what your last few months of kills have given you, with no credible extrapolation to loot in general." I say - yes thank you, that is exactly what my post is about and what I've already said myself. I don't mind you telling me the obvious, I just don't understand why.

    ReplyDelete
  10. I main a Warrior tank, and so I need on plate. Boy I can tell you that I see plate stuff with Intellect way too often!

    ReplyDelete