I recognize this is an ongoing study, but you really should set up some context each time you make one of these posts. It is isn't clear what MMO you're dealing with unless someone goes and looks back at the old posts to see what the objectives were. Aside from that small but noticeable issue, I believe the community truly appreciates the data you're providing here. Thanks.

Only 3.6% spend more than an hour in raid content in a MONTH? Haha, and I always thought the complaints of MMO companies pandering to the top 1% of players were slightly apocryphal.

Out of interest, what percentage ARE the typical thrice-weekly raiding guilds?

Prognosticator,

Excellent suggestion. Maybe a boilerplate sentence just above the details block at the bottom which gives minimal context and points off to a more extended description.

Let me make sure I understand, because the 3.6% number is cool and shocking, but it seems the less interesting number.

Unlike the "snapshots" Blizzard released back at BlizzCon, your sample set should catch both mains and alts, and I should think a person isn't too likely to raid on multiple characters within a month. Meaning that you could have 5 characters of which only 1 visited a raid, but they're all played by the same person.

So the 30% of all level 60s number combined with 28% of those bailing under an hour seems much more telling to me. That's, what, 21.6% of the level 60 characters. Sounds compelling to me.

Is that analysis right? Great work by the way, love what you're doing here.

I think etornal is right that there's an interesting question here about mains and alts that the drop number may say something about. In the raid guild I was in, basically there was a set of about 10-15 core players who had at least one alt at 60 as well as a main, and the alt was often a class that it was felt might be needed to make the raid happen (priests primarily; also warlocks, druids, mages and warriors). So you'd see people log in on one character, and then often drop as the raid assembled to go get the other one. Or even drop after someone else dropped in order to replace the person who dropped.

But otherwise, I think this is a hugely telling number (both of them) and a pretty good indication of how badly astray, in my view, Blizzard's development resources have been invested over the last six months.

Let' assume that an average top-level player do raids 2 times a week, and the average duration time on each raid be 5 h.

So, monthly minutes in raids are 4*2*5*60=2400M. In a diagram above, there is really few player(almost no one) who spent 2400M in doing raids. If this distribution of minutes be true, raid could be never high-level players' contents in western WOW players. It feel so strange to me.

In Korea, as far as I know, highest-level WoW players really love raid contents. To my personal rough survey through internet fan cafe, most of the top-level are enjoying raids minimum 3 times a week.

The arithmetic and presumptions in the preceding comment look a touch off to me, but there is a valid point here: the first thing that could be gleaned from the data you collected (although not in the way it is presented) is how many play-hours were sampled, and how many of those were spent in raids. That would give a figure of x% (.0x%?) of all playing hours were spent in raids.

That, for me, by itself would throw Bliazzard's strategy in a rather harsh light. Then you can normalise by gathering info on average number of characters played per account, average skew towards or away from main character in that, average hours played per level and applying that to approximate a figure for hours at level 60.

In all, though, i am sure there is some valuable info here to be extracted from the data.

I need more info, what was your method of sampling? Did you have a bot doing /who continuously and recording who was in what zone? How often did you check on where these people were. Also what servers did you use and how old are they? A newer server won't have as many established guilds. Where as an older server will have a larger number of established end-game guilds.

Please post your sampling method/procedure. I do like your sample size but if I check at noon each day, i'll get lots of people but not a lot of people raiding.

Some questions. You seem to be sampling a small number of servers from your post on methodology. This could skew the results. Also, are you sampling every 5-15 minutes, or only certain times? Do you account for guilds which raid at off-hours?

I have a very hard time believing there's that few '3-a-week' raiders. I'd like to see full numbers on how you did it, which servers, and where you think your errors could be. One I see obviously is that if it finds more than 49 of the same race/class/level it doesn't work properly. You said this is rare. I disagree. On some servers there can be up to 10 raids on one faction going on at the same time, which is likely ~30-50 of each class just for that. Add in non-raiders, and for common race/class combinations(ie. NE hunters, UD rogues, etc.) I would expect you'd hit the limit more frequently. Or you aren't hitting the true high pop servers.

Also, how do you account for alts, as a previous poster pointed out? Does your 3.6% number only take into account level 60 players, or all levels?

I personally have played up to 9 characters on the same server over the course of a month, and only raided on one. Yet I'd count as a 'raider' for all 9.

The methodology can be found here:
http://blogs.parc.com/playon/archives/methodology/index.html

The census bot takes a snapshot every 10-15 minutes. Online as much as possible.

Anthony - 3.6% is all characters. 30% is only level 60 characters. The latter number is more interesting I think.

Eric is currently working on a way to work around the 49 /who return limit. I think that cap is the largest potential source of error right now.

The other possible error I see Nick is that anytime it catches a raid during a wipe that could be 30-35 people outside of a raid zone that are actually raiding. These wipes can be quite frequent on cutting edge content. Just ask anyone who's ever learned vaelastrasz or razorgore. You spend more time outside the instance running back than attempting some nights.;)

This is still time raiding.

I think the 30% number is close, and possibly even somewhat accurate. I would like to see for all servers though, or at least which servers were being tracked.

Also, what did you define as raid zones? Did you include AQ20 and ZG as 'raids'?

One more question, what % of the 'less than an hour' raids can be accounted for by Onyxia?

Would you mind sharing the percentage of overall time (for all level 60 characters) spent raiding?

So just to recap. You ignore all overflow when the /who result is 49. Doesn't that make all your data wrong since your only counting the first 49 people and IGNORING all the other players? Guys come on, that is a huge screw up and slants the numbers drastically in favor of the casuals. Just copy the coding from wow census or another program that more effectively sorts data. Once you fix this problem, reset the data and take the sampling again you'll find a much larger percent of people do end-game content.

I have to echo was Zardas said. Your data is probably horribly, horribly skewed just on the basis of being capped at 49 responses to /who.

Further, I would guess that a very large percentage (probably at least 33%) of people who raid MC or farther have more than one level 60 - but they only raid with one of those level 60s. So if I'm on my 60 priest having fun all day but switch to my 60 warrior to tank a dungeon, I show up as two. This is going to be undeniable and uncorrectable source of error in surveys like this.

I hope you don't imagine that your survey told you anything useful whatsoever. The fact is that if you go on to almost all servers you will find a dozen guilds that are at least trying Molten Core multiple nights a week, if more aren't even farther.

"I have to echo was Zardas said. Your data is probably horribly, horribly skewed just on the basis of being capped at 49 responses to /who."

LRN2READ FTW! (From the link helpfully posted just above your post)

"The server is only willing to return 49 entries to us at a time, so if there are more than that, we restrict the levels (e.g. "Dwarf Paladin 1-25") until we're sure that we're seeing everybody."

"Caveat: We currently have no way to catch all players if there are more than 49 online gamers with the same race, class, and level."

We did read, from the link it says right here they are having a problem. On a large server this is a can be a lot of people.

Hah, thanks for picking out the LEAST problematic element of what I said and STILL being wrong about it. You lose at internets.

Just to put the /who issue away, if we can - Nick, can you confirm that you're still only having this issue with dwarf paladins, or have you started to see "/who overflow" on other combinations?

Also wondering if you've started to do the number crunching to detect alts: it seems like with the data you now have, you could create a name-overlap list of all the avatars that were simultaneously present in the same zone. Crunch that algorithm through all the samples and you'd get a list of all the possible alts for each avatar; if you get any complete overlaps between a set of avatars you probably have a hit. You could then use that to refine your estimates on how many actual players are raiding, for example.

The 49 cap is not a problem. This entire study is based on samplings and not ACTUALS. The 30% would still apply if 1000 dwarf paladins were found (I shudder to think of such a world). This is just basic statistics and probability. Of course there will be variations if you included more participants in your sample, but the same could be said if you changed servers too.
Good job boys. Keep up the good work.

Yes, we are seeing the cap hit for combinations other than level 60 Dwarf Paladins. We only see it for level 60s, but of course that's pretty much all that we're looking at in the raid situations. This is the reason that we're looking for other ways to subdivide the data, and get a more accurate read.

As far as we can tell, if there are more than 49 Dwarf Paladins, the 49 returned to us are essentially random. Of course, we can't know they're random, but we haven't been able to find a correlation. At peak raiding hours, we don't even know how many Dwarf Paladins are not being counted in the census. If we were trying to draw conclusions about the percentage of time spent in a raid dungeon, our conclusions would be all the more suspect. So our deduction that "28% spent less than an hour in raid content" should be taken with a grain of salt. They might have popped in later, and we missed it.

But when we talk about what fraction of level 60s spent time in a dungeon, any time in a month, our numbers are only skewed by people who did raid, but who only showed up at peak times, and were always unlucky enough to not be included in the sample. Statistically, they may exist, but I seriously doubt they have a significant impact on the conclusions.

So: Yes, we want to be able to take the census more accurately. But I suspect revised results will confirm our results, not invalidate them.

(Note: As well as facing an hours-in-the-day problem, the other reason we have not been eager to change the census bots is because we current average 15 minutes to scan through 20 race-class combos per faction, and then subdivide by level. There are substantially more zones than 20, although many of them are relatively unpopulated, so we fear that re-doing the scrapers to be zone-based could lengthen the census cycle considerably. Not to mention needing to dynamically add new zones as players enter them.)

The server issue is a real issue. We are only looking at 2 PvE realms, 2.5 PvP realms, and 1 RP realms. Sorry, but RPPvP is not even being looked at. These servers have been live seen at least April, 2004. (Ha! We labeled one of the PvE servers "low-pop", because it was at the time.) When we draw conclusions about PvE vs PvP, etc, we're talking about comparing 2 servers against 2 servers. I think our reply here has to be, "Data on 5 servers is better than data on no servers." Server/account donations welcome.

Mark,

I'd like to comment on one other point you made.

"Further, I would guess that a very large percentage (probably at least 33%) of people who raid MC or farther have more than one level 60 - but they only raid with one of those level 60s. So if I'm on my 60 priest having fun all day but switch to my 60 warrior to tank a dungeon, I show up as two. This is going to be undeniable and uncorrectable source of error in surveys like this."

Quite true. We have no way to detect alts. Making virtue of necessity, we only discuss this in terms of characters.

So. Level 60 characters who never, ever enter a raid dungeon, not even once during the entire month, but are played by the same player as a different level 60 who raids frequently, may lead some to interpret our numbers as meaning that the player playing the first characters is not enjoying the raid content. As you point out, this is not the case. Readers will have to assess for themselves how frequent they believe this occurrence to be.

"I hope you don't imagine that your survey told you anything useful whatsoever. The fact is that if you go on to almost all servers you will find a dozen guilds that are at least trying Molten Core multiple nights a week, if more aren't even farther."

I'm afraid that you're going to have to explain this to me a bit more. 12*40=480 people in MC might seem a little high to me (as a player, not "me" as a researcher), but not wholly unbelievable in a total server pop of about 6000 online at one time. Doing a quick calculation, 11098/5.5 is about 2000 players per server that we observed in raids during the month. I guess I'm not seeing where a dozen guilds in MC would contradict anything we've said. Please say more.

For the love of god I've never seen such a loser group of people in denial so bad as the hardcore raiders in WoW. Will you just please sit down and STFU and accept the fact you are a m-i-n-o-r-i-t-y, a very very small minority, albeit a vocal one that makes you just seem to appear as very large group.

You are like locusts to the average MMO playerbase. You come in, eat up everything in sight ten times quicker than any normal gamer could, and any dev crew wishes, then you complain cause you and your pack of no life internet buddies who play the game 12 hours a day have nothing to do.

You want something to do? Get a life. Play at a normal rate of speed for christ's sake, or gather up your colony of insects and move on to another MMO to eat up.

And if real life is not something you;d like to spend your time on, at least get it through your skull you're a minority group, always have and will be and you don't deserve jack.

I feel that the presentation confuses the matter. What you actually sample are data points of character + location + slice of time.

But the character view is misleading. We don't get a clue how long each character was actually played. So 5% of characters observed might be 5% of the time slice samples or 30% - we simply don't know from looking at the data.

As there's no good way to track accounts, a second, time based point of view would be very interesting here.

Also in this kind of study, I "guess" sampling time matters a lot. The ratio of raids should be explicitely higher during prime time than out of it - you can always find a pick up group when you've a day off, but for a 20-40 player raid, everybody must be free to play. A prejudice in the sampling time might kill any conclusions here.

To the "omg" comment above flaming raiders:

Actually you seem to be a bit in denial because most of the raiders I've come across (I'd say 80% of my guild at least) are very far from that stereotype. I have a full-time job and am a full-time student. Many of my guildmates (and we are a 'hardcore raiding guild' in BWL) are married and have children. Playing the game '12 hours a day' would get very boring for anyone, not just those that 'have a life', as you say.

The '12 hour a day' people are a minority within the minority of end-game raiders. Maybe I'm just lucky though and am in a good mature guild.

Just because I know how and where to level a character so I can get it done in two months only playing 3 hours a day on weekdays and longer on weekend mornings doesn't mean that I'm any better or worse than you. It just means my playstyle is different. Get over yourself.

Anyway...

This survey really shocked me. I expected the number to be closer to 20%. It seems nearly all of my friends who play WoW, both those that I have met online and those that I know from real-life activities, do raid at least once a month.

"omg" doesn't make sense to me. I understand the locust analogy - people who enjoy playing a game "efficiently" will probably look like they are zooming through the game - but its NOT the raiders complaining about nothing to do. Its the NON raiders, who either can't or wont raid who are the ones bitching. Having played in both raiding and non raiding grps, I can say anecdotely that my raiding guild spent less time online - They would pop on for a raid and usually log off within and hour of it ending - really using the time "efficiently," whereas my non raiding guild would simply waste a ton of time in IF trying to get a small grp going or just screwing around. This isn't a judgement on playstyle or who is having more fun; just what I observed.

As to who the developers are catering to, as someone who has worked on a (illfated) mmorpg, let me promise you that developers simply can't keep up with the rate that even a casual player can chew through content. Raid scenarios give them as good of "bang for the buck" as anything they can create (PVP notwithstanding). One raiding dungeon can keep 40*X people busy for 100's of hours while casual players want the indepth questing and progression that requires an expansion pack (which blizzard is happy to provide but it takes TIME).

Pete
www.theperfectdiet.net

Raiding has changed significantly in WOW's expansion with some factors encouraging it (such as a lower administrative hurdle thanks to smaller raid sizes) and others discouraging it (it's not as important to PVP success as before).

Has the percentage of level capped players that raid changed much?

Is Karazhan usage more like the larger raids (it has a lockout) or more like regular instances (it needs the same number of players as UBRS did)?