Greedy Goblin

Monday, December 20, 2010

Faction size vs progression

PvP update: it seems Blizzard completely untested the PvP systems before Cataclysm launch. Not only Tol Barad is unplayable, but the rated BGs are bugged too. We joined a rated BG, and got a trade pug opponent, obliterated them. They left the battle as trade pugs do. Then the battle ended because of not enough players and we got a loss. I really wish Blizzard would return to the "we publish when it's done" philosophy and don't publish unplayable content. I'm not talking about minor glitches but game-breaking problems. I don't play arenas, so I don't know if they are playable or not, but one of our main focus was large-group PvP and now both TB and rated BGs are broken.

It is a widespread belief that larger faction population is generally good. It's easier to find guilds or pre-made groups, easier to recruit from the server, and such healthy guild life allows better progression. Lot of people warned me that I seriously risk the success of The PuG by placing it to a "low pop, backwater server".

I always thought that more people is just that: more warm bodies, but anyone with a tiny bit of sociality somehow feels that a server with more people is better. However I found some numbers now. WoWprogress tracks raid boss kills, but give more data than just toplists of elite guilds. One of its chart display realms. And one of the data field here is "population". This is raider population, the number of people in the server who killed a raidboss in that tier (currently Cataclysm). It also gives faction ratios, so it's easy to calculate how many hordies and allies killed anything in Cataclysm:

Warcraftrealms have a list of faction populations. There is little data from some factions so I discarded all factions that has less than 2000 players. I merged this list with the WoWprogress list in an Excel table, calculating the raider % of every factions in the US servers. For example Aggamar has 474 raiders with 0.2:1 ratio, so 392 ally raiders. There are 4593 allies in Aggamar (all data Dec 17 morning), so on Aggamar-ally 8.5% of the players are raiders. Pretty good. Then I made a chart where I shown the raider % versus faction population:

Ouch! While the chart is pretty noisy, the trend is obvious: you have 4x more chance to bump into a raider on a 2000 pop faction (2%) than on a 20000 pop faction (0.5%).

Large factions are like huge cities. There are more exceptional people there than in an average city or town. They are visible and give the impression that the city is full of such people, while actually they are just a tiny percentage of the population. On the other hand there is a huge welfare-leech population in cities, making the average wealth lower than it is in a middle-sized city.

Like the poor person from a village moving to the big city, in WoW the M&S slowly migrate to larger servers. In a middle-size server if he tried out all 5 failguilds he has nowhere to go. On the other hand on a 20K population faction in the moment he quits xXdetkillazXx, he can get into DarkLulzSquad. "Ferindly heplfull gulds" spawn on these servers like mushrooms, as there is always a moron who thinks that he can fill a guild with /trade filth and it can carry him to raids. The M&S can blend into the crowd, can remain anonymous which is a bliss for him.

So no, the large servers are not elite places. They are places with some elite players and socials tend to bathe in their sunshine. But actually the server is just crowded with M&S. It is much harder to find good players in such a server than in a smaller one.

PS: Did I mention that The PuG is on a low population server?

Athéssoé sent a beautiful moron. The item he is selling is Jeweler's setting, 1.5G from vendor without discount in unlimited amounts:


Squishalot said...

I'm not sure how many people you'll convince to join the PuG's server, considering you've just demonstrated that intelligent people have more people (as a proportion of people on server) to screw over on a larger realm.

Re: opening about PvP, consider that Blizzard tested with volunteers and people who want to keep playing, not with trade puggers who drop from the battle too early. As you've found in the past, it's hard to predict a lot of behaviours.

Anonymous said...

i'm curious whether this would apply to any "new" server.

I don't know whether you noticed, but EU-Magtheridon (site of your ganking project?) was recently changed to PvE and is the current "New Players" server - so basically a "population reboot", coinciding with the Catacylsm influx.

What will EU-Mag turn out to be in 3-6 months time?

Gevlon said...

@Squishalot: "losers ragequitting a match" is unpredictable? Are you kidding?

Anonymous said...

If wowprogress does indeed only count raiders that have actually downed any bosses then your statistic is largely irrelevant right now, as most guilds (including the pug) hasn't started raiding or downing any of the bosses. Let's look at it again in 2 months time.

Squishalot said...

@ Gevlon - No, but losers are unlikely to be a significant part of the beta testing group, because they have no incentive to provide their time for no reward. Blizzard wouldn't have really encountered such behaviour in the beta. Certainly, if they did, the proportion would be significantly less than on live realms.

chewy said...

"Warcraftrealms have a list of faction populations."

Not entirely true. What it has is a collection of snapshots of people online when the addon was run. While this data is the best there is outside of Blizzard, it's already somewhat flawed. I'm sure you can see it, but this is like counting the population of a city by receiving photographs of busy streets, it's only indicative.

I'm not saying you're wrong, I'm saying the data is not comprehensive enough to make absolute conclusions.

Ygg said...

@Squishalot: part of any software development is to test for all cases, even the most unlikely that you could think of; if you fail to test what happens when one team leaves a BG, which is not a scenario no one could think of, then obviously the testing has been very slack.

Ðesolate said...

Rated BGs should have the same system as Arena. Disbanding team loses.

@the raider ratio: since at our server 60% of all serious raid guilds have a raidstop for one moth after the cataclysm start. (x-mas raidholidays)
Anothe point will be made when the new content is coming up (i guess in 3-4 month) and random hc´s will give you t11.
Then lowbobraidslacker will be able to clear some raidcontent with his friends. As in BC fast low raids will be interesting if the pointsystem stays at all raids give you Valor Points.

If old raids turn into Justice Points they will be abandonned as naxxramas in WotLK and will only spend achievements.

PS and off topic: testet 2 Healer setup in Lost City of the Tol'vir and Halls of Origination slowing us down ~15% of our standart setup-time. 2 healer 2 tanks slowed down by ~25% but no wipes no matter how fail the actions where. We had 2 wipes at the 2 healer setup both player based hard failures. Regular run had 5 wipes in Lost City of the Tol'vir and 8 at Halls of Origination (first time run).

Riptor said...

Sorry about the double Post last Friday. My Browser told me my initial Post was to long...

@Anonymus2: All serious Raiding Guilds have started to Raid at least last Wednesday. I guess with the somewhat increased difficulty of Cataclysm we will now be able to distinguish between Raiders and Casuals again. While in wotlk every drooling Lolkid could kill 11/12 in ICC and call himself and his Band of equally drooling Headcases “Raiders” (even according to the Stats on wow Progress) the same group is now stuck in Stonecore hc (or even better in Blackrock at the Second Boss because neither their Rogue nor the Prot Warrior are able to interrupt a Boss properly) and will in a couple of weeks be one-shooted by a fully charged Arcanotron while asking themselves if the green Mist on the Floor is to be considered equally dangerous as Fire (to which one only reacts if its red as we all know that blue Flames are good)
So, if you are a “Raider” and your Guild/Group has not started to raid yet you should seriously consider looking for a proper Raiding Guild or another Group.

OT: As far as I have understood Gevlon, the PuG is not supposed to be a Raiding Guild but more a sanctuary for the asocial Casual with whatever fancy in WoW (PvE, PvP, Goblinism, etc).

Aljabra said...

""losers ragequitting a match" is unpredictable? Are you kidding?"

He has a point. While you can predict it, it's pretty hard to do proper testing of it, as most M&S, who will behave this way on live servers won't get far enough in the beta to actually give enough data. And, as there are a lot of testing to do, it's pretty easy to overlook such things, especially when you focus on PvE more, than on PvP (which appears to be the case here).

"your statistic is largely irrelevant right now"

On the contrary, it is mostly relevant right now, and exactly because very little number of guilds has they raids going. That guilds are the hardcore top raiders, so if you want to count them when you can be sure, that more casual hadn't yet mixed in, now it's the best time.

Anonymous said...

> it seems Blizzard completely untested the PvP systems before Cataclysm launch

They could probably have tested it better, but calling it "completely untested" is uncalled for. I'm sure you can imagine how it would look if they did no testing at all. Your blog is generally very reasonable and well informed, but this comment falls in the same category as the argument stripped whiner forum posts about how Blizzard ruined the game or nerfed their class or whatever.

Too often when people complain about some product, I can't help but think that those people never actually worked in any projects where a product was developed. Stuff goes wrong all the time, deadlines are never met, noone is ever fully satisfied with anything, yet we still love doing it and still churn out great products. And every once in a while, it slips. Mistakes are human.

Andru said...

I am speechless on how you take the right data, properly work with it and arrive to the wrong conclusion, Gevlon.

NO good guild randomly invites people from the server so the fact that you're statistically 4 times as likely to bump into a raider than a non-raider on a low-pop server is largely irellevant.

Good guilds are a concentration of good raiders. More raiders=more good guilds. It doesn't matter that for 5000 good raiders there's 20000 M&S. They're NOT in my guild, therefore they're not in my game. I can just safely ignore that part of the server and I'd be none the wiser.

However, if there's only 300 raiders on a low-pop server and only 600 M&S, it does not matter to me, personally that the ratio is better.

From a guild perspective, I would think that they'd love to be on a high-pop server.

So they'd have to decline 4 times more applications than on a low-pop server. Big friking deal.

It's entirely one thing to find a specific raider-position in 5000 raiders than find the same in 300 raiders. The fact that the application board would be spammed by "lul inv peeps" doesn't bother anyone.

It's a tragedy, however, when you need a disc priest, and you can't find one in those 300 raiders on your low-pop server.

Yagamoth said...

I'd be very interested in the development of the graph over time. However I guess it would be way too time consuming doing this by hand every single time..
Rated BGs:
To be honest, I wouldn't have thought of this case. However, I'm not a WoW developer. And thinking about it, rated BGs can be seen as large arenas with objectives. And most people leave the arena when they don't see a chance to win anymore (e.g. when one of their mates dies in arena)
Something unrelated to this post:
It's often mentioned here, that The PuG is "asocial". I personally agree with quiet some things written here, however I'd consider myself to be simply "not social", as I do not have any intentions to hurt the socials. Does The PuG have this intention?

Gevlon said...

@Andru: your main mistake is assuming that those 5000 raiders are all waiting for your inv. On a high population server there are more raiding guilds, all poaching for the same raiders. Imagine that you just clone a server, every player gets a clone, every guild would get a clone-guild. Would it make your guild's recruitment easier?

Anonymous said...

I can't believe people are defending game breaking bugs like the BG one because its 'pretty easy to overlook such things' and whatever. Of course no amount of testing is going to pick up all obscure bugs but, honestly, the amount of severe bugs in 4.0.1 was ridiculous, and still more are coming to light.

The only reasons such bugs don't get picked up is that Blizzard don't know how to test software, Blizzard can't be bothered to test properly, or much more likely, is that deadlines meant there wasn't time for sufficient testing. Whichever it was, Blizzard (or whoever is responsible for setting their deadlines) are at fault.

Anyway, its much more likely the bug was picked up in testing but given less of a priority than other more serious bugs and due to deadlines and other more important work hasn't been fixed yet. Again, suggesting the expansion was rushed out.

Ulsaki said...

One thing that should be considered is that some of the more dedicated raiders will have transferred to lower population realms to avoid instance capacity problems, and server latency.


Business update: I've made about 80K profit this week so far, mostly from selling Truegold and Vials of the Sand.

Even using trade with my crafting offer has been profitable. Whilst there have been a lot of time wasters, it's created market awareness of the product, in addition to causing players to stockpile truegold (several have let me know they're saving the mats for it).

So far I've made about 9K (crafting them for 3K each) risk free. I've made much higher profits by crafting them myself and placing them on the AH, in the realm of 6-13K, but this comes at a much higher risk, since it's a ~50K investment.

Several players are also gaining me business via word of mouth.

Trade advertising is normally pretty useless, and generates low gold per hour. In a new market though, it still serves a useful purpose.

Dàchéng said...

Just a comment on Blizzard's testing: testing isn't the same as fixing.

The release was a rush job, with new features, late to or not available in beta-testing, making it to the test lab in the last few months, weeks and even days.

I can imagine the month before release would have been hectic. Many late features would have got little testing because more important features had to be tested first; other features that were tested would have been changed ("fixed") at the last minute with no time left for proper retesting (last-minute bug-fixing is a well known source of new bugs); and finally tested features in which bugs were found were released anyway, because there was no time to fix them.

None of this is the fault of the test team, nor the development team. It's a consequence of setting an aggressive deadline. Situation normal for any software project.

Seta said...

Is there any sign to Blizzard making a PvE server from Aggamar? I really like the idea of the PUG and the only reason for not joining is the PvP server. I really do not like the idea of giving total power to M&S to ass-fuck me whenever they just feel the urge for it.

Anonymous said...

Gevlon says:

"you have 4x more chance to bump into a raider on a 2000 pop faction (2%) than on a 20000 pop faction (0.5%)."

This conclusion does not follow from the data. Who you will bump into is also affected by activity level, which was not considered in the data.

In any case, I don't think you're going to get accurate results right now because the expansion is still very new. In two months I think the numbers will settle and give you something more interesting to work with.

Riptor said...

Anonymus said:

"This conclusion does not follow from the data. Who you will bump into is also affected by activity level, which was not considered in the data."

According to Blizz (and TB) the Top 11% of PvE and PvE Players log 50% of Servertime. The Chance to "bump" into a Raider should therefore be much higher then normal. Never the less in my Opinion Gevlons analysis shows that well populated Servers are more of a sanctuary to M&S as the Raiding Guilds are more likely to not keep track of each others experience with Bad Players.
The smaller the Raid Population on a Server (except for hc Guilds who recruit mostly Cross Realm) the more likely the M&S are to not find PuGs anymore as soon as a couple of Guilds know them to be abysmal.

I read in one of the Comments about Guilds going to low Pop Servers for better Latency. I remember Ensidia doing so as they blamed Lag in their Realmpool when they lost a lot of Progress Races. However, Lags in Instance Servers are not necessarily related to your home Server but mostly to the Battlegroups. Even if you are on a Low Pop Server you can have laggy Raids if you share your Servers with high Population Realms and too many Guilds log into the Dungeons at the same time.

Orgaansint said...

I'm playing on Spinebreaker Horde (EU) and Ravencrest Alliance (EU). The first is flagged as New Players, low-medium populated, the other one is full/locked.

It may be just me, but on the low populated realm It's way worse for me to find a decent pug. For example a goldrun will most likely never happen because people don't even have the gold for it.

While on the one with full population I manage to join raids that kill LK on heroic, people that come enchanted/flasked et cetera.

It could be a coincedence (spelling?) though. The Alliance server has several guilds with toplayers/alt from topplayers, while the Horde server only has one guild that is somewhat capable of creating a decent raid.

Anonymous said...

"@Anonymus2: All serious Raiding Guilds have started to Raid at least last Wednesday. "

Existing Raiding Guilds yes. But not all serious raiders that have come back with the release have. There's a lot of new serious raiding guilds forming, too. Finding and testing players takes a lot of time too (if you're serious about it, anyways.).

I myself have been lucky enough to find a good group of people relatively quickly and we started raiding last tuesday, despite having an incomplete lineup and still recruiting for our 10man team.

Ax said...

@Orgaansint - what you are seeing is more likely the spillover of raiders. Assuming you only have so much raid content your guild does (say 3 times a week), this leaves talented raiders up to their own devices during their off time. As Riptor says, the top 11% of PvE log 50% of Servertime (which we'll assume is true). This means you're much more likely to encounter these people in pugs.

That possibly counters Gevlon's point about progression, but I believe he's over simplifying many aspects which makes it hard to make any conclusions.

Alrenous said...

I knew rated BGs would be useless as soon as they announced it.

It'll probably be workable a couple patches from now.

Also the rating system is hidden. You can pretty much guarantee it's so broken it's almost not funny. If it's leaked or reverse-engineered, you'll find the system is sublimely game-able.

Unknown said...

*sorry for possible double comment-posting error*

Hello Gevlon,

I really find it great that you have gone to such trouble to collect and plot the pop-size/raider% data. I applaud your effort! I'm just not certain if one could draw any reliable conclusions from such noisy data.

Scientifically speaking, there is so much noise that the null hypothesis ('there is no relationship') could probably not be rejected with sufficient reliability (p<0.05)

Worse, I fear that the downward slope may be a natural result of combining two plots of which one is probably Gaussian (raider%), and the other Poisson-like, or Gaussian on a log scale. Simply put: extreme values don't often co-occur, like it's hard to be struck by lightning and an earthquake simultaneously. If you would plot the length of all Nobel-prize winners, you'd probably see that very few are above 2m10, and quite a lot between 1m70 and 1m90. You'd get a downward line too, and might conclude that long people don't get enough blood to their brains.

A good statistician (which I am NOT) may be able to tell you how to find out how reliable that downward line really is. As it stands, the best I can recommend is to divide your dataset in bins/slices of say 10 servers, and calculate the median (NOT average, as averages are much more influenced by accidental extreme values) of each bin. If you still see a consistently downward line, the relationship you found is probably correct; if not, the downward trend would probably just be the artifact of combining two natural distributions.

Wilson said...

There's a fundamental flaw in your analysis.

Most servers are low population, so the majority of servers with a high raiding percentage are going to be low population ones. This produces a slope like the one you have drawn.

However, if you look at the data points, you can see that most of the servers with very low raider percentages are also low population. This just doesn't show up in your slope because no server can have a percentage below zero. The reality is, you could pick any statistic you want - percentage of tauren, percentage of plate-wearing classes, percentage of characters whose names start with the letter "m", and you would get a similar looking graph.

Squishalot said...

The thought crosses my mind that the population data may be indicative of decent guilds transferring to low-population realms in the search of Realm First raiding achievements.

Anonymous said...

I think the guild leveling social engineering will change things.

Once some guilds have rank 25 perks, I think there will be considerable momentum to be in larger guilds. And that seems easier to achieve on more populous realms.

If you are a software developer and your beta did not predict / test the behavior of the customers, then you ran an incompetent Beta.