Greedy Goblin

Monday, February 28, 2011

The fall of the core M&S myth

I made a comparison table for the guilds on the server (Feb 24 data), including all guilds that killed something. I'm fully aware that it's not a complete list and high levels can be earned by completely social guilds with no raiding at all, but I assume that if a guild is active, it must have 10 players who killed something, at least in different /trade pugs.

Current level
Bosses killed
Level 85
Pescorus (A)
3 heroic
2 heroic
1 heroic
Ultimate paradox
1 heroic
Sons of Liches,
The Noble Europeans,
Do nothing and win
Suomislackers (A)
High Definition (A)
eXplicit (A)
Red Army (A)
The PuG (A)
Kitty Spins
Vanilla Core (A)
The Unforgiven
The Army of Darkness
You Got Carried
midlife crisis
Midnight Madness
Saunan takana on tilaa
Nordic Crüe (A)
Why no love
Ancient Shadows
Anxiety (A)
Warchiefs Secret Service
Saga Inside
Night Stalkers
Flying Circus (A)
Wrath of the Alliance (A)
CenturionS, (A)

At first this data is directly opposing my old belief that only banker guilds will not max out guild XP every day. Only 11/41 guilds are level 17 despite when I wrote this post, our guild is lvl 17.8, so it's not like "they did not cap on Christmas eve", the less-than-17 guilds are at least a week behind us.

Secondly I calculated correlations between guild XP and other stats. The XP-level conversion is needed because level is not linear with effort, if you have half the XP of a lvl 17 guild, you are level 11 and not 8. (Technical info: since data of very different magnitude are to be compared, I divided everything with its average. So "1" is 350M on the XP scale, 208.5 on the character scale and 7.1 on the bosses scale. It does not affect the R2 values and linearly scales the fitted coefficients)

The correlations with the guild achievements are weak positives and theoretically not interesting. Let's see the correlations with guild XP:
Guild XP practically measures "how much the guild play" as you get it from quests, dungeons, raids, rated BGs. While it's possible, it's unlikely and unnatural to play without earning guild XP. "how much the guild play" is obviously strongly correlated to guild size: more people play more. The number of bosses killed has zero correlation with the guild XP. A more progressed guild doesn't mean a more active guild.

If you are good at maths you already see the point. But let's make it perfectly clear. The axises of the next chart are the guild XP/character (simply divided XP with the character count, no normalization this time) and the bosskills:

The XP one person gains is completely uncorrelated with bosskills. The myth of the no-lifer, the claim that better players play (much) more is therefore defeated.

The evidences for the myth are the anecdotes about the unquestionably unhealthy amount of playing some topguilds do. But this is a one-sided result, as the equally unhealthy play without bosskills remains hidden. Check out this guy (source) for example: no raid kills, not even HC Stonecore or Grim batol, no 2000 rating achievement either. "Casual guy playing a little for fun" you'd say. Well, check out his statistics: 1.2 million honorable kills. I have 35K after living in Wintergrasp for half a year. Some casual! Or think of the 9K+ achievement guys or the ones with the "what a long and strange journey". But their no-life make no headlines, no site makes toplists of their "achievements", so you can act like no-life is only among HC raiders. However their playing contributes to guild XP, so finally someone can came up with hard data to prove it.

Disprove this with better data or accept: there is no connection between play time and progression. After that, I'd love to hear what excuse you have left for the fire-dancing 3K DPS. Gear maybe?

Update: to commenter suggestion I collected how many lvl 85 are in the guilds, as he assumed (not without ground) as lvl 85 count is a better data for player count than character count, due to alts. The correlation between GuildXP/lvl85 and boss kills is still R2 = 0.012, so no matter how we approximate player number, the result is the same: boss kills are uncorrelated from amount of playing.

Update2: Squishalot noticed that the average guildXP/person differs between the top, mid and bottom of the list. I tried to formulate this find and got this:
Now both R2 values are low, so it can be just noise. But if it's true, it says that "casual raiders" with 5-8 bosses play the smallest amount, and both HC raiders and "fun ppl" play more than them. The 1/12 people had two times more XP than the 5-8 people! Again, the correlation is weak to claim it but if it's true, that's really a slap in the face to the M&S who has "no time to raid". (note: I removed the guild "yourtearingmeapartlisa" because it only has 12 lvl 85's and no recent activity, so probably a dead guild. It had insane amount of XP/person, as the XP was collected by people who are no longer in the guild)

PS to "this is nonsense, if I'd play more I'd kill more bosses, the connection between play time and progress is obvious" trolls: of course. But if everyone would play more, than your 2HM would be 6HM, my 9/12 would be 2 HM, randomguy's 2/12 would be 4/12 and Mr "i haz no time 2 raid i haz life lol" would be revered instead of honored with bloodsail buccaneers. You would only inflate the bosskills, their relative values (compared to the average) would not change. In mathematically correct form my statement is "The variation of progress in the playerbase has no connection to the variation of play time."


Anonymous said...

you are actually ahead with achievements, looks like your prize money pays off, I wouldn't have thought people would hunt achievement for so little gain.

on your first point, a guild doesn't get a kill achievement if 10 members killed something in different groups, it has to be a guild group to count.

Anonymous said...

Guild XP only measures amount of time spent playing doing specific activities (those which give guild XP, obviously). How do you know that accurately reflects all play time?

Also 'number of characters' surely includes alts for some guilds, but not for others (the PuG). What you really need here is 'number of players'.

Trelocke said...

I find your conclusion extremely weak due to the massive amount of incomplete data and, in my opinion, illogical connections.

First of all, guild XP is capped. You have no idea if a guild is hitting their cap on day 2 or day 7. You may feel this doesn't matter since you can point to guilds who aren't capping as being more progressed but it is still a massive amount of missing data.

Secondly, you have completely ignored individual contribution to the guild's XP. Data which, last I checked on the armory, is actually partially (explained in a moment) available. So you're not telling us if 75% of a small guild is contributing to the guild on a weekly basis where maybe only 20% of a large guild is contributing.

And lastly, even an individual's contribution to guild XP is capped (at 1,575,002 per week). I know of no way to track a person's guild activity beyond this number. So you have no way of knowing if the 10 people in a small guild are playing enough to contribute 10 million XP to the guild but are being capped at 1.5 and thus unable to reach their guild's weekly cap or whether a large but less progressed guild is capping because their members play a lot less but have a lot more people to pick up the XP "slack".

Due to the individual XP contribution cap, you can only tell if a player has played X amount. You have no way of knowing if or how much a player played beyond X.

Gevlon said...

@Trelocke: you just reformulate "raiders are in a small no-lifer guilds whiles casuals play individually little in large guilds"

However it was explicitly disproved by the second chart. Guild size do not tell us anything about progression. If you would be right, progression would strongly negatively correlate with size (there are lot of small progressed and large unprogressed guilds)

Anonymous said...

Trelocke, you're wrong about guild XP (your point 1). *Individual* guild XP is capped on a weekly basis. People who got lucky with the overcap bug the first week (like me) were hitting exalted even last reset, maybe even earlier.

Guild XP is capped on a daily basis, not a weekly basis. Your guild can earn 6246k per day (at least till level 23, apparently). If you don't earn it on day 2 of a reset, you can't make it up on day 7. Guilds that are at the current max of roughly level 18 (except for a few guilds that benefitted from bugs) have maxed their rep every single day, not every week.

Grim said...

I do not really believe that no-life is a requirement for bosskills, but your logic is just false:

"Now let's combine the two results: Bosskills are completely uncorrelated with guild XP (how much the guild members play) and also completely uncorrelated with "how many people are in the guild". As a result, bosskills are completely uncorrelated with "how much a person plays"."

Guild XP strongly correlates to player count, so you are basically comparing bosskills twice to the same thing and finding that it correlates the same way both times. Then making a non sequitur.

Where is the bosskills vs xp-per-player chart? Oh wait! No data for it!

Gevlon said...

@Grim, the data is right front of your face.

Oh wait, I can do much more elegant proof, making the chart now.

Trelocke said...

Either you missed my point or I did a poor job of formulating it.

The data you are using to track game play is capped, both on the guild level and the individual level. I'm saying that regardless of how progressed a guild is, regardless of how large or small it is, you have no way of knowing how much the players responsible for the progression are playing because the data only tracks up to X. All data beyond X is lost.

I'm not saying HC raiders are no-lifers in small guilds nor am I saying they're not. I'm saying that the data being used to make your point is being poorly interpreted. A deeply progressed HC guild *could* be spending 80 hours a week raiding and you'd never know because the data only tracks the first 10 hours (example only) before it stops tracking altogether.

Also, I have not seen any sort of conversion chart that says how much XP is earned for the guild for different activities. I know I can reach my cap before the end of the week by doing nothing more than the cooking and TB dailies every day. Will a person who spends 60 hours a week raiding pull down their XP cap and if so do they do it within the first 15 hours or does it take all 60?

Your point may very well be valid. But the data you are using to try and prove it is terribly incomplete and doesn't even remotely begin to track what you are saying it does.

Gevlon said...

@Trelocke: I don't question that data above the cap is lost. However reaching the cap is not trivial, most guilds don't do it. So if there would be a group of "XP cappers" (HC-s or others), they would slide the chart in a way that "all raiding guilds are lvl 17 while casual guilds are everywhere between 10 and 17". If you check the data table, you see lvl 17 guilds with 12/12 and 2/12.

Trelocke said...

@anon: You are correct about a guild not having a weekly cap. However, due to the individual weekly cap, you still get essentially a weekly cap for the guild as well since once players have reached their cap they no longer contribute to the guild's XP gain.

However, whether a guild has a weekly cap or not is a moot point because what Gevlon is trying to prove is that progressed players aren't no-lifers who play ridiculous amounts of hours. Since a player's XP is capped on a weekly basis AND there is no tracking beyond that cap AND there is no X time played = Y guild XP formula, the data being used doesn't prove (or disprove) a thing.

I think Gevlon has a valid point to make, I just think using this data as proof is false and illogical.

Sean said...

Yep, I've always known this to be true. But of course, I've never gone about proving it with hard data.

A lot of people are arguing that time does help progression - well, it definitely does but I believe the progression distribution at any particular "time spent playing" is the same. In other words, it is skill that correlates with progression and not time.

The progression = no-life myth is brought about because people want to make themselves feel better about not progressing.

However, it is important to note that the world's top progression guilds are EXTREMELY hardcore during the early stages of content and would raid for many many consecutive hours (e.g. 12+). Have a look at Paragorn's efforts during ICC in WOTLK.

Grim said...


The data is not in front of my face for the reasons that Trelocke explained - everything is capped.

But I just realized that's not the worst of it. Remember that its actually players who progress, not guilds.
So basically you are trying to correlate the progress of 2%-20% of characters to the activity and count of the 100%.

Even if there were no caps around, we would still have absolutely no way to compare the activity of the progressed players.

Anonymous said...

I don't know if the data is available, but a more interesting piece of information would be a measurement of guild age against its guild XP. Instead of just a guild level, you would have a ratio of potential / actual guild xp, which would probably tell you more about the relationship between play time and progression.

For example, a recently made guild of hardcore players (with very high play times and a low guild level) might be skewing your data and giving the false impression that there is no relationship.

Something else that would be interesting to look at is the number of level 85 characters in a guild instead of total characters. that would control for alt-heavy guilds (or the opposite, such as your guild.) I am willing to bet that that would be some serious nose-to-the-grindstone research though.

Gevlon said...

@Grim: characters are not put randomly into guilds. They choose each other for similar interest. Despite there are exceptions, the normal case is that in a progressed raiding guild there are progressed raiders and in a funguild people are funplayers.

Anonymous said...

"The myth of the no-lifer, the claim that better players play (much) more is therefore defeated."

I am not sure the analysis in the post is enough to defeat this claim. Two reasons:

1. (Minor) A fair number of raiding guilds have a no-alts policy similar to that of The PUG. This policy makes it hard to account for the entire playtime of the guild members, since many raiders have alts, and the policy forces alts to be in other guilds. I don't know what is the ratio of raiding guilds who have this policy, so this might or might not be significant.

2. (Major) The claim "better players play (much) more [than worse ones]" is about players, not guilds. Guilds come and go all the time and people constantly move between them. So what if a particular guild which kills bosses in Cata is only level 9? For all we know, its members might be playing since vanilla, and for most of them this might be their ninth or tenth guild. They might not play as much today (or they might still play a lot, and their playtime just fails to be captured by analysis, see point 1 and a point on XP caps made by other people), but if they already played for years, their playtime is still considerably larger than those of others.


Grim said...

Even if we rule out all the non-raiders who hang around in raidguilds as friends/family, all the national guilds and casual funguilds that have a 10man or two who are actually worth a damn, we still don't know how many killed what and what is their activity.

Of the 70-120 characters in the top 4 guilds not all are going to be in the group that kills heroic bosses.

And it gets worse after that. How many of the 200-500 characters in the half-decently progressed guilds are actually progressing and how many are contributing to XP?

The PuG is a good enough example. Of the 238 characters, how many have killed 9/12? 15?
How many have killed say... 4/12? How many have not killed anything but just done a lot of dailies?

Also - what is the guildxp/hour of those who progress and those who do dailies?
Wipes don't give xp.

Anonymous said...

Guild XP isn't enough because it doesn't track enough. Wiping on a boss doesn't give you any XP. So someone could just postulate "those guys just raid more than I do, so that's why they're so far in progress!" and you can't disprove it. I don't believe that's true but current available game data does not allow any logical conclusion one way or another.

Trelocke said...

Let's look at this from a different angle. I have an alt which I log onto every day. I do the cooking daily, the JC daily, as many of the TB dailies as possible and arena for conquest cap. I spend an average of about 1.5 hours per day for all of this. I am always capped or very near cap by Saturday. I'll round up and estimate it takes me roughly 8 hours to reach my personal weekly guild contribution cap. I have another 3 hours of play time that isn't even recorded.

I would say it's a safe assumption that you play/raid at least that amount of time each week. Yet you aren't even 2/3 of the way to reaching your weekly cap.

My point is: Guild XP has no logical correlation to time played. Yes, most things you do in WoW contribute to Guild XP. However, there is no reasonable translation into Guild XP/hr. With no reasonable translation how can you say one does or does not play X amount more or less than someone else?

Anonymous said...

"There are three kinds of lies: lies, damned lies, and statistics"

The conclusions you are drawing aren't supported by your data set as the dataset is not fit for purpose and you don't test your assumptions. It may be the best you can manage, but it's not good enough to pretend you have meaningful results.

In addition to Trelocke's points; Using a dataset of 43 guilds and pretending its a representative sample of wow guilds is a bit naive. Not to mention the top end of your dataset is only 3 HM .. that's not even at the standard that people start accussing people of being no-lifers. Further more you're not including those guilds that have yet to kill magmaw, and don't even have a way of differentiating between those that can't and those that don't want to.

Second, your preposition that Guild XP is a measure of how much you play is flawed. 25 people spending 12 hours wiping on HM Sinestra doesn't count as much as 1 person spending 12 hours levelling an alt fishing in the barrens or archealogy or something.

The fact that not all of a "no-lifers" activities are recorded by the measure of guild xp whereas all of an M&S'ers will be skew the results in favour of your conclusion that no-lifers don't play more to such an extent that any conclusions become meaningless.

You don't even seem to acknowledge the importance of that pvp'er with 1.5million kills. How long has he spent playing in a field that doesn't earn guild xp? Hardcore pvp'ers don't tend to raid hard modes, so does that mean in your analysis they would be classed in amongst the M&S because they don't have raid experience?

If you're trying to prove that "bad" players play as much as "pro" players, then you do not have the data to prove that one way or the other with any degree of confidence.

"So if there would be a group of "XP cappers" (HC-s or others), they would slide the chart in a way that "all raiding guilds are lvl 17 while casual guilds are everywhere between 10 and 17".". Except that one of the better ways of levelling the guild is running heroic 5mans, something that's very common in non-raid guilds and gets less common in the more raid-orientated guilds and would probably be non-existant in a "no-lifer" guild.

Add to that the fact that its reasonable to assume if the majority of the HC mode guild spends the majority of their time doing something outside of your data collection technique (eg wiping on aforementioned heroic modes) it all adds up dodgy data not fit for purpose.

Anonymous said...

Actually if you only raid you get far less guild experience (and reputation) than by doing other stuff, especially questing and 5man heroics.

With 5man heroics you need around 5-6 guild runs to cap a day. This means a guild with 30-40 active players will cap just with daily guild runs between main and guilded alts (which makes much sense to guild to make them count in the guild progress).

With raids you need around 8 raid bosses down *each day* to cap. This is obviously not feasible, so if you focus your efforts in raiding only and don't spend much time in other activities you won't cap.

Recently they did boost the rewards for raid kills, so maybe things are a bit more balanced now.

Gevlon said...

@"Wipe don't earn XP" people: I'm not sure you have noticed that you are actually claiming that raiders just "raid and log off, don't do random stuff". This is practically the opposite of "no life": being online all the time.

Raiders always claimed that while they raid a lot, they don't play other "useless" aspects of the game like alts, achievements, dailies

Alrenous said...

Gevlon, did you look at the correlation between bosskills and achievements, or did you stop at kills vs. XP?

Gevlon said...

@Alrenous: Slightly positive (y = 0.35 x) and low correlation (R2 = 0.1). On the top of that not really interesting.

Taemojitsu said...

Minor critique: the first four guilds at top of ranking (fifth is at rank 17) had a late start in Cataclysm. Pescorus Jan 6, Control Jan 11, IMP had noticeable turnover on Dec 12 and 31, Ultimate Paradox was formed on Dec 30.

The influence of the rest of the raid group has always been known to be much more important to progression than any individual's skill. "No life" might better be interpreted as more commonly referring to scheduling priorities, not total played time; whether or not this distinction is argued effectively by those who use the phrase.

The 1.2m HK's player is, honestly, interesting in that many of the most vocal proponents of the current raid difficulty are players who left prior to or during WotLK, and have not had extensive personal experience with the mid-expansion 'planned obsolescence' that became standard during that time. If the date of completion of content is seen as a meaningless social achievement and completing 'current' content (a wording based on the implication that content released earlier is less interesting) requires restricting class choices or imposes other obligations, then it's very reasonable that a subtlety rogue (with PvE DPS that's amazingly significantly lower than even PvE frost mages) would choose not to do PvE content in Cata when it's 'current', or even not at all.

Grim said...

'you are actually claiming that raiders just "raid and log off, don't do random stuff"'

No, we are not. We are just claiming that raiders have lower guildxp/hour.

If someone spends 3 hours wiping, then 2 on dailies, he has the same guildxp than someone who was online just 2 hours doing dailies.

Also, if someone wipes for 12 hours, then finally kills the boss and has thus progressed, he has less guildxp than someone who has quested for an hour.

One bosskill generally gives ~7 max level quests worth of xp (councils give more) so go figure...

Will said...

"@Grim: characters are not put randomly into guilds. They choose each other for similar interest. Despite there are exceptions, the normal case is that in a progressed raiding guild there are progressed raiders and in a funguild people are funplayers."

But if you refer to a point I believe you made in an earlier post Gevlon, most social guilds have skilled but social players attempting to carry unskilled players. Sometimes there are enough of the more skilled players to kill bosses; sometimes not so much.

Anonymous said...

@Gevlon: I don't actually think that's true. "no-life" from the mouth of a m&s/social means just that, she doesn't care how that one spends its life in the game just that they do.

Of course, in reality these people just log on for raids and log off right after (spending probably about 3-6 hours straight). But that's not how these people are viewed. They are being though of spending whole days, hour on hour on progressing.

There is not a single evidence in your statistics proving otherwise (just that these people do not spend their time doing stuff that rewards guild XP). The absence of evidence is not an evidence of absence.

Anonymous said...

I am sorry to point out that progress defined as "number of bosses killed" cannot correlate with guild XP if there is any other activity in the guild beside raiding.
Guild XP is earned through raiding, non-raiding PvE (5-man dungeons and quesing) and PvP. Raiding progress (bosskills) can be achieved only through raiding.
If all members of the guild would log in only to raid than bosskills would be only source of guild XP (so correlation would be at 1).
If, on the other hand, the guild is PvP-centered and wins a lot of arenas and rated BGs, it will have a lots of guild XP but few bosskills (if any).

Furthemore, amount of XP earned is a poor measure of play time. 3 dailies on Tol Barad give approximately the same XP as 1 raid boss kill. 3 dailies take about 10 minutes to complete, whereas 1 bosskill takes up to 10 minutes for encounter and also some time for trash - and this is for bosses on farm. For Nefarian firstkill, the same amount of XP will be earned with many hours of tries.

Anonymous said...

"Disprove this with better data or accept: there is no connection between play time and progression."

That hypothesis entirely depends on how one defines playtime: With or without reading about classes, bosses, and strategies?

So it probably boils down to: M&S don't read up but have the same (ingame-)playtime.

Anonymous said...

"you are actually claiming that raiders just "raid and log off, don't do random stuff""

Er No. Personally I'm claiming that there's a whole chunk of "playtime" you aren't accounting for when trying to calculate whether progression orientated raiders play more or less than "scrubs" and because of this (and other factors) your "analysis" is mostly meaningless.

If I spend 10 hours wiping & 1 hour farming mats and MrScrub spends 2 hours fishing and 2 hours doing archealogy, then the way you are measuring gametime and interpreting that data would show that the scrub plays 4 times more than me because there are 5 hours of the pro's playtime that you aren't recording.

It doesn't matter whether or not the raider logs off after or goes on to grinding or mat gathering, what matters is the 10 hours you missed and the fact that proportionally more of the activities of a casual/M&S/scrub are recorded in guild xp.

Therefore your methodolgy will always underestimate the amount of time a "no-lifer" plays because you cannot track his amount of playtime as accurately as you can for a "casual".

If you can account for 99% of a casuals playtime, if you're only counting 80% of the "no-lifers" then you will be underestimating the number of hours a "no-lifer" plays it will always skew the results towards the outcome you are trying to prove (that there is no correlation).

You're normally a smart cookie, I can't see why you don't seem to grasp or acknowledge the shortcomings here.

You aren't proving or disproving the link between playtime and progression. At best you are proving/disproving that the amount of manhours spent on activities that level the guild aren't directly linked to progression.

One of your biggest assumptions:

"Guild XP practically measures "how much the guild play""

is a flawed fallacy. Can you at least accept that this assumption is flawed as it tracks everything a casual/M&S would do and misses out an activity which could be deemed a defining difference between the 2.

The data you need to prove or disprove your point is not available to you.

Gevlon said...

@Grim: here you assume that raiders spend lot of time wiping and otherwise just as effective as anyone else. It's not true, their random HC is 30 mins, even with alts, they do the necessary dailies in group and don't do other inefficient nonsense.

However guild XP/hour is something neither of us can approximate.

@Last anonymous: you wrongfully assume that only raiders have no-XP time (wiping) while "you can account for 99% of a casuals playtime". Let me give you a list of "casual" playtime that give 0 guild XP:
* standing in Stormwind chatting, anal spamming, begging gold, displaying "cool" mount
* archeology
* fishing
* any kind of farming except dailies
* leveling lowbie alts (OK it does give a little bit of guild XP, like 20 for a lvl 10 quest)
* random BG
* random 5-man without at least 2 other guildies
* /trade pug Baradin Hold
* Tol barad
* ganking lowbies
* exploration

Anonymous said...

More progressed guilds are long done with the daily heroics unless they have many people with alts or many socials.

It's not unreasonable to find today a guild with most of the people playing only for the raids, and even if they are successfull and don't wipe, they still won't cap a week's worth of guild xp if they clear every raid available in a lockout.

True hardcore guilds also play in spikes: when the race is relevant they play day and night. Once the race is over (first kills claimed) they go back to a low-playtime routine.

If you do a mean probably they play far less than "casuals" during a whole year, but during progression time they are able to put together a lot of "high quality" play time.

Kelindria said...

Most Hardcore guilds are working people who simply don't have time to spend 12 hours a day playing for the most part.

The only no lifers I have encountered are achievement hunters and people trying for world/server firsts. The main thing of note is that hardcore players spend extra time preparing to raid or they will be kicked for not being prepared. I imagine if you could graph time in raid/prep time vs casual social guild and hardcore you would notice the progression would be significantly higher for the people who focus their time on raiding related goals.

The main difference going from casual to hardcore is the time spent getting ready to raid. Social you just show up and kill stuff. hardcore you plan.

Anonymous said...

My whole arguement isn't invalidated by that one (incorrect) assumption though.

That's like saying your post has zero merit because you incorrectly claim archealogy doesn't give guild xp. You have one fact wrong, but your overall point stands (that there was an implication that I'd made an incorrect assumption)

Fishing and archealogy level your char, therefore grant guild xp.

If you'd said "fishing at max level" you'd have been right, but you didn't so you're wrong and I can ignore the rest of your post!!!

^ that kind of defensive reasoning is one of the things you seem to hate about defensive ape-routines. Shouldn't the Goblin thing to do would be to at least discuss the points raised and not to jump on any minor inaccuracy and use that as a reason to dismiss the rest of your arguement.

You don't have the necessary data to draw the conclusions you are trying to prove.

Your "estimations" at the amount of playtime played by each camp is wildly inaccurate.

Even if the amount of time Joe Bloggs spends in anal /trade spam exacttly matched 1/25th of the number of hours Method spent wiping on "HM Boss A" and you've accurately modelled the playtime of your subjects, your analysis is still ropey/flawed/unproven due to your sample.

It's too small to be representative of the WoW populace and it doesn't include any "no-lifers" to begin with ("no-lifers" are further progressed than 3 HMs.. first 3 are "easy"(ymmv))

Healer24 said...

Of course, this sort of breaks down at extremely low play times. The theoretical person who plays for one hour per week is unlikely to have any raid boss kills simply because with that low of a play time they have only just gotten to 85.

In general I agree with this chart and I'm glad you made it. It's very similar to the aim of the Undergeared project which I loved following.

Also, the person who has over a million honorable kills might just really like pvp and not do much pve. No raid boss kills does not mean failure at the game. I got a 404 error trying to access the armory, but does this person have a high arena/rated battleground rating? Basically, not enough information in the post to say anything bad about that guy as far as skill goes.

N said...

Personal experience tells me this is true. I've lost a lot of my interest in WoW, so for this expansion I've only played one character, and only really done the things I wanted to - raids and Arenas, basically.

I found that my gear was perhaps one or two items behind the people who grinded all day every day, because they were taking on tasks with significantly lower ROI in order to get "the best." Archaeology, daily heroics, etc are very low ROI activities, but they did offer an advantage early on.

I suspect that top guilds usually mirror that behavior... they know they'll get gear, and they know that one extra epic isn't going to make or break their progression. Certainly they'll go out of their way for best in slot, but none of the first kills in the weeks following release were by people with full epics farmed painfully over long days.

Anonymous said...

The more focus you have, the less time you need to learn what you want.

Is that not obvious to anyone?

If anywhere, the myth 'effective raiders are per definition nolifers' exsists only in the minds of the M&S (they might phrase it otherwise) - for me, it's nothing new.

Good leadership, strong focus and average time contribution, that is hardcore (okay, over world top 300) raiding.

The Gnome of Zurich said...

Also, it's not at all clear that time spent banging away at a progression boss is achieving guild XP. Is it? Does it do so only to the extent that you have to reclear trash? Normally you only get XP for positive results, like actually downing a boss. I would not expect to get any guild XP from a wipe unless it involved killing trash or linked mobs.

Also, it's very likely that, just like normal XP, some activities are very efficient at generating guildxp/time and others are quite inefficient. Sorting through your mailbox and selling items on the auction house or in trade probably doesn't generate any guild XP, even if crafting does.

Guilds that care a lot about maxing their guild xp, will be making sure they do enough of the highly efficient GXP activity to cap every day/week, while those who don't won't bother, but may be spending just as much or more time playing.

Anonymous said...

M&S can survive. They skew the truth to make themselves feel better.

M&S plays X hours and is at Y boss kills. He concludes that this is the 'norm' and anyone who is at Y+1 boss kills must play X+1 hours.

Replace boss kills with DPS or Gearscore as you wish.

It is their own ineptitude combined with massive ego that allows them to think this is correct.

Alrenous said...

I have to disagree, it's very interesting.

The hypothesis is that high bosskill guilds are full of highly skilled players.

The objection is that actually, they're full of highly experienced players, but the guild XP isn't showing this due to atypical behaviour on the part of the hardcore guilds. They're spending all their time wiping (learning hard fights) or something similar.

If indeed high bosskill guilds were fooling your metric in this kind of way, they would also be suppressing their guild achievements. Correcting for population, they would all have between 90 and 130 - the dungeon and raid achievements.

That the correlation is close to flat but positive definitely refutes these kinds of objections.

Anonymous said...

I really dont think you can draw all of the conclusions you have drawn from this data. Many have hilighted issues that are missing in the data.

Disprove this with better data or accept: there is no connection between play time and progression.

Of course there is some connection. If you dont spend time energy and effort at the start you will not be able to get further along. Playing 5 minutes a day gets you no where. There is a point where time will not impact progress. I dont think anything you have analyzed has shown this. All it shows is that guild xp....doesnt capture all playtime.

One other point...if we have alts etc in all levels of guilds....are we not distorting the data. The progression raiding guild that has no alts vs the progression raiding guild with lots of socials...give a different view of playtime and guild xp vs bosskills.

Deprava said...

The difficulty I have with this argument is that you make a lot of logical leaps based on your pre-conceived conclusion, which is that in-game play time does not equate with skill. I agree with that statement, but I do not believe this data is the best evidence to support it.

Looking objectively at the data, it is impossible to reach any conclusions about the average play time of any one person involved in the listed guilds. You state that guild XP is a measurement of how much the guild plays. That is true, but it does not measure how much the individual players play. If a guild, such as yours, has a strict “no alts” policy, then any play time on an alt is not added to the guild’s play time, and hence does not contribute to guild XP.

Consider the scenario where a hardcore raider has a raiding character in a raiding guild, but on nights without a scheduled raid plays on an alt in a different guild. His contribution to the guild XP of his raiding guild does not accurately describe his time spent in-game. If he spends his time in his alt guild running instances or otherwise practicing execution skills, he is gaining skill and racking up play time, but not contributing that play time to his primary raiding guild's XP.

Your second assertion, that guild size strongly correlates to “how much the guild plays” is likewise logically flawed, because you are assuming that every member of a guild is an independent player. As other comments have pointed out, a guild may contain twenty active players, with each player having five characters in the guild. This would show a guild size of 100, but the guild would still contain only twenty active players. The counter-argument that “If the twenty people have five alts and play them, then that simulates the XP gains of 100 people” is incorrect because the overall XP gain is greater when two individuals play for four hours each as opposed to when one individual plays for four hours total split between two alts.

The only conclusion that this data presents is that raid progression does not correlate to guild XP earned, and that a guild’s level is not indicative of its raid progress. That does not, by association, prove that a player’s skill is not connected to hours played, since guild level/XP is not analogous to individual player activity.

Sthenno said...

What about the fact that if every member of a guild spent 120 hours in raid dungeons every week since cata came out and were now up to 25/25, their guild would not even be level 10?

Guild xp does not correlate to time played, it really correlates mostly to heroic dungeons completed. Until the very recent buff, killing a raid boss generated only about 250k guild xp. Killing the second boss of SFK without attacking his adds in a 5 player guild group generates 1.9M.

I think your conclusion is accurate (with the exception of world first guilds who really do put in fantastic numbers of hours early on in new content), but I don't see how this data proves it, because you can't tell how much time people actually spent raiding.

Squishalot said...

Damn, I came too late to get an early say in the discussion.

I've got an interesting set of stats for you:

Boss progression, XP per Char, Xp per 85:

1-6 bosses: 1752k/char, 4913k/L85char
7-12 bosses: 1510k/char, 3539k/L85char
HM(any): 3219k/char, 5774k/L85char

If you want a bigger disparity:

1-12 bosses: 1610k/char, 4051k/L85
HM(any): 3219k/char, 5774k/L85

What this says is that guilds that complete hard modes are lower level because they have less members (obviously). However, they have significantly higher guild XP per character than normal mode bosses.

There is a good reason why this isn't seen in your second graph. You've fallen prey to the flawed idea that everything is linear.

The big problem that you failed to identify was that social guilds need to be treated separately from raiding guilds. The graph is skewed towards the all the social guilds with high activity and poor raiding (i.e. 1-6 bosses down).

With this limited dataset, it tells an interesting story - your social guilds generate high amounts of guildXP per character. But this is actually skewed by the fact there are so many low level characters in such guilds. Comparing heroic guilds to 1-boss guilds, they have the same total guild XP, similar L85 characters, but the 1-boss guilds have twice as many characters generally.

What actually seems to happen is that social guilds waste a lot of time generating guild XP by doing dailies and so forth. Semi-serious raiders (as in, the 7-12 boss kills) appear to be less active than socials, because they spend less time farming, and more time wiping and in progression.

Hard mode guilds, however, are serious about anything they do. Their actions generate twice as much guild XP per character as characters in other 'raiding' guilds. So it's fairly clear to say that, on average, on your server, HM guilds are more full of 'no lifers' than other raiding guilds (if any raiding guild can actually claim to have a life).

For your own exercise - take an average guild XP earned per character for each level of progression (i.e. 1/12, 2/12, ... 2HM, 3HM), and compare them. You'll find that the three highest numbers come from 3HM, 1HM and 2HM respectively.

Kurt said...

#1) I see a tremendous number of critical people in this thread, saying that the data is bad, the methodology is bad, the conclusions are bad. What I'm not seeing is any of them making any positively useful suggestions, or linking any of their own statistical analysis. Part of science/the scientific worldview is attempting to form and test hypotheses based upon limited initial data. I can only imagine how you all would have reacted to Einstein's theory of relativity, or any other correct but initially unsupported hypotheses from the past.

#2) None of you acknowledge that the other side of the argument is supported by absolutely zero statistical analysis. When you attack the person who attempts to analyze the limited data at hand with such emphasis, while ignoring the other side of the argument which doesn't even bother attempting to be rational about it, you are contributing to the current air of ignorance, where naive obfuscation is rewarded with general acquiescence, and an attempt to derive truth from data is scolded as an untoward effort to rise above one's place.

The general response to this post saddens me.

thehampster said...

Gevlon, there is another major problem with your conclusion: Overall game time is NOT equivalent to the time one has available to spend raiding.

Raiding typically requires at least 3 hours of uninterrupted time. This is virtually impossible for most people with a spouse and especially young children. Whereas, guild heroics and achievements can be done 30 minutes at a time, and if you get interrupted it's no big deal.

I actually play WoW more now than I did when I was single, but when you have a spouse and a young child, serious raiding is just not practical. Sure, some people can make it happen, but they're the ones blogging about it. The vast majority of people just give up.

Gevlon said...

@Squishalot: I need time to gather and evaluate data. Your idea is interesting and not ignored.

Squishalot said...

@ Gevlon: Thanks for letting me know.

For reference, what I did was copy and paste your table, assign guild XP values in a similar manner as you did, then use SUMIF and COUNTIF into a table containing progression levels. Calculating averages like this essentially gives you a weighted average across all guilds of that progression level (e.g. 6/12).

This is still subject to all the flaws of your original analysis - primarily that we lose data beyond the daily guild XP cap, or even the weekly player contribution XP cap. However, segregation into discrete progression groups is more useful than attempting to form a linear correlation. If there is actually linear correlation, you should see it in the discrete groups as well.

Gevlon said...

@Squishalot: check update2 in the post.

Squishalot said...

@ Gevlon: I'm not sure why you're always so insistent on using R2 as a measure of correlation. The p-value associated with the slope is the measure of how strong the correlation is. The R2 value only assesses how far away from that slope (i.e. how variable) the data lies. You can still have a strong association between variables with a low R2.

Having said that, your conclusion only applies to those in 1-4/12 raiding guilds. It says nothing about the non-raiders, who may indeed play less than the raiders. All you can conclude is that 'if you have time to raid in the first place, then you can be good at raiding'. If you don't have time to raid (for example, I don't have even 1.5 hrs in a guaranteed block to join a PuG style raiding group), then indeed, more time may result in more progression.

I know it's not available, but I think that we'd also need to consider the numbers on post-85 guild XP gain per member. Given that getting from L80 to L85 is approximately 24M XP, or 6,000k guild XP, the 'no life' statistics are heavily biased towards levelling guilds rather than raiding guilds. That may also explain the lower guild XP/member number for mid-tier raiding guilds.

Gevlon said...

@Squishalot: the R2 says how good the data fits to the function. You can ALWAYS fit a linear or quadratic or ANY equation to a data set (assuming it has less parameters than data points). This linear will always have a slope and an intersection. If R2 is low, than the to data dimensions have no connection (that the chosen function could describe).

You are right that the p value (Y = p*X + Q) tells HOW they correlate but it is only meaningful if R2 is non-zero.

A bosskill gives 100K guild XP, even BEFORE it's buffed. I killed 10 Argoloth, 5 conclave, 5 magmaw, 6 tron, 6 maloriak, 4 atra, 2 Chim, 9 halfus, 7 valiona, 2 council, all together 5.4M guild XP.

You are also right that there are time limits for raiding. The most blatant example: Blizzard told that 80% of the people who have a trial account will not subscribe. These players played 100 hours max alltogether with 0 bosses. The subscribers played more (even 1 hour/day over 2 years are 600 hours) and killed 3 bosses on average. Than you could conclude that one boss needs 200 hours playing.

I do not question that there are "very casuals" who have no time to raid. What I claim is "among people who raid at all the predictor of success is not time".

Sthenno said...

@Squishalot: Obviously if someone only plays for 10 minutes a week then they will never down a raid boss. I think that focusing only on those guilds that actually have downed raid bosses makes sense. Ideally, I'd say focus only on guilds that consistently down at least one boss every week.

It reminds me of the old "Money can't buy happiness" which is true assuming you have enough wealth to have a place to live and put food on the table. If you don't meet the minimum then money buys a lot of happiness (or buys you out of misery, if you'd rather look at it that way). If you do meet the minimum then the correlation pretty much disappears. I think the question is whether the correlation between success and play time disappears when you hit a certain threshold of play time, since it is obvious that it will be very strong below that.

@Gevlon: I still don't think that guild xp is a useful measure of time played. The variation between different activities is huge in their impacts on guild xp. 15 people dedicated to maxing guild xp can hit the daily cap in around 20 minutes. The same 15 people could easily play for 8 hours and never get to the cap.

Hardcore raiders spend more time wiping than casuals and casuals spend more time doing other non-xp generating activities than raiders. But we can't usefully guess how much more non-xp time each team puts in - the data just doesn't have the information we want.

Squishalot said...

@ Gevlon: "What I claim is "among people who raid at all the predictor of success is not time"." is a satisfactory conclusion, given the limited data, but your conclusion on the main post page is slightly different - you've no reference to raiders only.

But as we've illustrated, we're using a proxy (guild XP per char) of a proxy (XP generated) of the actual time spent playing (i.e. measurable in hours per day). It's not the most reliable of datasets.

"The subscribers played more (even 1 hour/day over 2 years are 600 hours) and killed 3 bosses on average. Than you could conclude that one boss needs 200 hours playing."

In reality, it's more likely that one boss needs 450-500 hours, and the next two bosses need 50 hours each. Once you're able to start raiding, it's about your skill / ability to adapt that determines your progression. But there is a much larger amount of time needed to prepare for the first boss of an expansion.

@ Sthenno: "Obviously if someone only plays for 10 minutes a week then they will never down a raid boss. I think that focusing only on those guilds that actually have downed raid bosses makes sense."

I agree, but it's not quite represented by the conclusion in the main post, as I illustrated to Gevlon.

It's worth noting, however, that HMs appear to require much more devotion than bosses 6-12. So it still does demonstrate somewhat that HM raiders are 'no-lifers', relative to the raiders farming 8/12.