Automated Guild Progress Ranking - Page 12

Brissa · 08/03/07, 12:37 PM

Originally Posted by songster

1) Getting into SSC and TK
2) Killing Kael
3) Essence / Mother Shahraz

The list of upcoming nerfs looks like it's addressing 2 and 3 - Blizzard presumably intend for a large proportion of the people that got into SSC to progress into BT.

Nerfing the time it takes to complete phase 1 on Kael will do little to speed up progress. Sure you will have more time on your hands because you will be wasting less time on trivial things with each attempt but the impact is sure to be small.

songster · 08/03/07, 2:05 PM

OK, so I'm off work today and did the breakdown for the full list.

Bracket		Number
0 - 25		3300
25 - 50		2768
50 - 75		3144
75 - 100	3402
100 - 125	837
125 - 150	852
150 - 175	296
175 - 200	175
200 - 225	135
225 - 250	231  \___  Blocked at Vashj
250 - 275	135  /
275 - 300	87
300 - 325	199  <--- Blocked at Kael'thas
325 - 350	19
350 - 375	17
375 - 400	9
400 - 425	3
425 - 450	4
450 - 475	11  \___  Blocked at Bloodboil
475 - 500	10  /
500 - 525	6
525 - 550	22  \
550 - 575	21  |---  Working through BT/Hyjal, but past Bloodboil
575 - 600	25  /   
600 - 625	0
625 - 650	13  
650 - 675	17  <--- Working on Souls, or working on Shahraz+Archimonde
675 - 700	0
700 - 725	13  <--- Working on Shahraz
725 - 750	4    <--- Working on Council
750 - 775	5    <--- Working on Illidan
775 - 800	0
800 - 825	12  <--- Killed Illidan

Looking at it at the resolution of individual guilds, the high end breaks down very neatly indeed (as indicated). The score per boss is large enough that the individual bins mostly correspond to specific bosses.

You can see large subsidiary peaks at ~325, and at ~250. These correspond to Kael'thas and Vashj. In fact. everyone up to ~275 is blocked on Vashj, the only difference is whether they're concentrating on finishing SSC, or whether they've taken down Al'ar as well

However, what of the lower range - the vast majority of the raiding population? Bear in mind that every single one of these guilds has Kara loot, so we are not talking about non-raiding guilds. A score of 25 approximates to Curator.

At this point, the bin size of 25 is insufficiently precise. Also, at this point the accumulation of loot via farming starts to be more important relative to killing new bosses, due to the way wowjutsu scores. Looking at the specifics, it's somewhat hard to draw a line where Kara stops and Gruul's lair begins. Now that the attunement is lifted, many guilds are killing Maulgar before Kara is clear. I note in passing that this absolutely gives the lie to the idea that people are happy with the 10-man game and don't want to do 25-man instances.

To get the clearest possible picture, I went and looked at the data for 5-point bins from 30 through to 180.

Bracket		Number
30 - 35		672  \
35 - 40		722  |
40 - 45		382  |--- Clearing through Kara
45 - 50		657  /
50 - 55		753  \
55 - 60		718  |--- Farming the majority of Karazhan
60 - 65		331  / 
65 - 70		1197 <--- Has killed Maulgar prior to clearing Kara
70 - 75		145
75 - 80		1200 <--- Has cleared Kara and killed Maulgar or Gruul (?)
80 - 85		221
85 - 90		1883  <--- Has killed Maulgar and Gruul 
90 - 95		38
95 - 100	60
100 - 105	37
105 - 110	720  <--- Has killed Maulgar, Gruul and either Magtheridon, Lurker or VR
110 - 115	24
115 - 120	13
120 - 125	43
125 - 130	440  <--- 4 25-man bosses down
130 - 135	9
135 - 140	7
140 - 145	138  \___ 5 25-man bosses down
145 - 150	258  /
150 - 155	13
155 - 160	2
160 - 165	99
165 - 170	174  <--- 6 25-man bosses down (Maulgar, Gruul, Magtheridon, VR, Lurker and Hydross is the typical pattern)
170 - 175	8
185 - 180	7

Note that past Karazhan, the worth of an individual boss is 15+ points, so the graph breaks down into individual mini-peaks corresponding to the number of 25-man bosses downed.

What we see from this is that:

1) A lot of people have completed Kara and not gone into 25-mans. This is the constituency that needs ZA quickly.

2) A lot of people have completed 10-man, are trying to get into 25-man, but have nothing more than a couple of Maulgar kills to their name

3) Of the guilds that are able to get into 25-man content, a huge number cannot kill anything outside Gruul's lair / Magtheridon, and a similarly huge number cannot kill anything beyond the so-called "loot pinatas" of VR and Lurker.

In these lower echelons, you don't progress, you farm the easy stuff and wipe a lot on the rest. These guilds have averaged one new 25-man boss every other month since TBC came out. And judging by the amount of gear they've farmed from Kara, Maulgar and Gruul, it's not for want of trying. There are still very significant barriers to progressing from the "one-boss" instances into the larger multi-boss instances. It's not clear whether they're social barriers or encounter complexity barriers, but they're unquestionably there.

To give some idea of the degree of the blockage, there are more guilds with scores between 85 and 90 than there are with scores between 150 and 825. And these are not just small, casual 10-man guilds that don't care about 25-man content, these are guilds that are trying to conquer the 25-man content and are unable to get beyond the first couple of bosses.

A confusing factor is that many guilds are recorded as killing Gruul but not Maulgar. This may reflect transfers, but is IMO much more likely to reflect guild alliances, where a consortium of 2-3 guilds has 1 or 2 Gruul kills to its name, and the loot happens to have divided unevenly. The vast majority of Gruul-killing guilds have at most 1 or 2 pieces of loot from him.

songster · 08/03/07, 2:09 PM

I'll come back and take another look in a couple of weeks - see how things have changed.

kaib · 08/03/07, 8:08 PM

Originally Posted by Brissa

Nerfing the time it takes to complete phase 1 on Kael will do little to speed up progress. Sure you will have more time on your hands because you will be wasting less time on trivial things with each attempt but the impact is sure to be small.

You do realize that 'the time needed to complete phase 1 will be substantially lowered' is a rephrase of 'we are going to reduce the HP of kael's advisors by a lot', right?
That means that essentially you have a lot more time in p2/p3 to clean up mobs before kael spawns, which is a pretty big nurf to the fight.
Right now the advisors have a total of about 2.5 million hp (I think the 3 males 700k and the mage 500k). If they remove 800k to 1 mil from that (which I think is realistic if he sais substanial reduction), that makes that phase a LOOOT easier.

Trouble · 08/03/07, 11:07 PM

Originally Posted by kaib

You do realize that 'the time needed to complete phase 1 will be substantially lowered' is a rephrase of 'we are going to reduce the HP of kael's advisors by a lot', right?
That means that essentially you have a lot more time in p2/p3 to clean up mobs before kael spawns, which is a pretty big nurf to the fight.
Right now the advisors have a total of about 2.5 million hp (I think the 3 males 700k and the mage 500k). If they remove 800k to 1 mil from that (which I think is realistic if he sais substanial reduction), that makes that phase a LOOOT easier.

They specifically said that phase 3 will remain unchanged.

WoW Forums -> Raid and Dungeon Encounter Update

Grogi · 08/04/07, 12:42 AM

Very interesting info, thanks Songster. I've always assumed that data would be something like this but it's great to see finally numbers about this. Hopefully Blizzard knows the data too and makes some important changes for their next expansion. For TBC it might be too late to do needed changes in order to allow more guilds to experience most of the 25 man content.

Brakar · 08/04/07, 3:15 AM

Originally Posted by songster

I'll come back and take another look in a couple of weeks - see how things have changed.

First, hats off for the data gathering. From my own point of view, I'm not too surprised with how it has broken out. Especially in it's current state, 25 man raiding can have a very snowball like effect. It can be hard and painful to get enough people consistently, get enough gear, learn good group composition, learn how to heal spikes, etc. etc. Once you get enough of all of these magical ingredients to down Gruul and Mag you're over the hump. Once you can do these two cleanly you will be able to progress through a good portion of the T5 content. With time you should be able to do any fight at least up to Leo in SSC and VR in TK. Similarly, by accounts Kael is a gate keeper. It's not that early BT and Mt. Hyjal are easier than SSC, it's they are easier relative to the content you just passed. The jump from Gruul/Mag to SSC is minimal and are about the same in my opinion. It may take you time to learn, but they're fights you should be able to. Conversely, after Kael, early BT is actually a step down in difficulty so guilds blaze through it. It would be like killing Vashj is required to go into TK at all. Once you get Vashj down, you should easily blow through Al'ar and VR and quite possibly Solarian.

The problem as I see it isn't that there is a spike at the early 25 man bosses, it's that the spike is too high. If the vast majority of your guild has little to no 25 or 40 man experience, there is a rediculous amount of things to learn just to finish off Gruul and Mag. Even once you get over the largest hurdle of a consistent 25 people that is at least somewhat balanced there's a lot of agony ahead. There needed to be a 25 man version of Karazhan. There's a fairly clean progression of difficulty and you can learn important things from various bosses. Even if you completely stall on curator, you still have a handful of other bosses you can kill and feel like you're being productive. It's infinitely better to have a "gatekeeper" boss after truly easy bosses. Even if you spend weeks never getting past the gatekeeper you feel productive as you're getting other loot for that size raid. The first boss being such a boss is completely demoralizing and destroys many a guild.

Looking ahead, I wonder how high the Gruul/Mag block may become and if we'll see a lower dip ala BT as those that can get Gruul/Mag move on while those stuck stay stuck. For your data, did you anything to try and tell the difference between those trying 25 man bosses and failing versus those not even trying. Guilds that just aren't big enough for a 25 man raid aren't entirely fair to include since there's no way to tell if they'd actually be stuck on Gruul/Mag for long.

songster · 08/04/07, 7:16 AM

Originally Posted by Brakar

For your data, did you anything to try and tell the difference between those trying 25 man bosses and failing versus those not even trying. Guilds that just aren't big enough for a 25 man raid aren't entirely fair to include since there's no way to tell if they'd actually be stuck on Gruul/Mag for long.

No, there's no way to tell that from wowjutsu data. However, I think the fact that we're seeing Maulgar kills from a lot of groups who don't have Nightbane or Netherspite down (or in some cases even more) suggests that a large prorportion of the 10-man groups are trying to get into 25-man raiding.

[Edited this post due to an error in my Excel sheet - see below]

songster · 08/04/07, 11:58 AM

Apologies to all, I had an error in my Excel sheet, now corrected. I've edited the analysis post above.

The results are actually somewhat more depressing following this analysis - the "block" is shifted down by 10 points. Thus the major spike corresponds to guilds that have killed Maulgar and Gruul but nothing else. It also reveals the very significant number of guilds at 65-70, who are starting 25-man raids prior to finishing Kara.

So yes, there is huge demand for 25-man content - basically any guild with scores from 65 up is trying to tackle this content. But the dropoff after Gruul is immense. This may reflect the artificial barrier of the attunements (now lifted), or may be a historical artefact reflecting disbanded guilds.

As I understand it, wowjutsu only removes a group if it actually /gdisbands. A large proportion of (1) may well be guilds that got Maulgar/Gruul/Magtheridon down and then died because of the attunements issue. Guilds with VR down are more likely to be "live" guilds, since he's only really been accessible for a short while. If part of the "three boss" spike is down to dead guilds, then that signal will hang around in the data set even if the live guilds make progress into SSC. I guess we'll see in the coming few weeks.

Antiarc - any sane way of addressing this issue? Flag up guilds that have no kills on their "top level" bosses over the last month, perhaps? Obviously if a BT guild stops killing SSC bosses, that's no biggy. If an Gruul-killing guild stops killing Gruul, that's clearly indicative of something else!

Aspir · 08/05/07, 12:12 AM

I do enjoy these posts you've been making Songster. It's an informative view of the data, and at some point I'd probably like to generate reports like this directly.

Originally Posted by songster

Antiarc - any sane way of addressing this issue? Flag up guilds that have no kills on their "top level" bosses over the last month, perhaps? Obviously if a BT guild stops killing SSC bosses, that's no biggy. If an Gruul-killing guild stops killing Gruul, that's clearly indicative of something else!

I'm assuming you mean 'Aspir', not 'Antiarc' -.-

Right now a lot of guild disbands end up getting caught by the "must have 10 70s" rule. So for example, the guild disbands but doesn't /gdisband -- however there's mass exodus of players drops the guild underneath the minimum required players.

I also have some reports showing member retention, which I've been using to help catch some guild disbands (very low member retention usually indicates an inactive guild). A number of people have also e-mailed in pointing out disbanded guilds and other errors (it's really helpful when people do this).

Regarding seeing guilds with Gruul kills but not mag: I check for this in one of my scripts to track down suspect guilds, but I haven't addressed all of them yet-- there are several hundred guilds that still fall under this category. Most of the time they are guild transfers that happened before wowjutsu started scanning, or server transfers. I believe the quantity of PUG/Guild Alliance situations is a lot smaller.

Note: I just finished making a first pass at updating the WowJutsu point values based on the statistical data, so this will affect any further interpretations that you do:

+---------------------------+----------+------------------+-----------+
| Name                      | OldValue | NewValue         | Delta     |
+---------------------------+----------+------------------+-----------+
| Doom Lord Kazzak          |        1 | 0.58059951663017 |   -0.4194 |
| Doomwalker                |        1 | 0.59144964814186 |  -0.40855 |
| High King Maulgar         |       10 |   15.20233297348 |   5.20233 |
| Gruul the Dragonkiller    |       12 |  20.906296730042 |    8.9063 |
| Rage Winterchill          |       30 |  31.266349077225 |   1.26635 |
| Anetheron                 |       30 |  31.313950777054 |   1.31395 |
| Kaz'rogal                 |       30 |  31.375796914101 |    1.3758 |
| Azgalor                   |       30 |  31.416212797165 |   1.41621 |
| Archimonde                |       35 |  36.224272608757 |   1.22427 |
| Attumen the Huntsman      |        2 |  1.0814146399498 | -0.918585 |
| Moroes                    |        3 |  1.5149976015091 |    -1.485 |
| Maiden of Virtue          |        5 |  2.0000939369202 |  -2.99991 |
| Romulo & Julianne         |        4 |  2.6785959005356 |   -1.3214 |
| The Big Bad Wolf          |        4 |  2.5038551092148 |  -1.49614 |
| The Crone                 |        4 |  2.4761879444122 |  -1.52381 |
| The Curator               |        5 |  3.3637437820435 |  -1.63626 |
| Terestian Illhoof         |        6 |  7.1580649614334 |   1.15806 |
| Shade of Aran             |        6 |  7.3459193706512 |   1.34592 |
| Netherspite               |        6 |  7.4231885671616 |   1.42319 |
| Chess Event               |        4 |  4.0989996343851 | 0.0989996 |
| Prince Malchezaar         |        8 |   8.552485704422 |  0.552486 |
| Nightbane                 |        8 |    14.8345079422 |   6.83451 |
| Magtheridon               |       20 |  28.020212173462 |   8.02021 |
| Hydross the Unstable      |       20 |  28.474243164062 |   8.47424 |
| The Lurker Below          |       18 |  27.880250930786 |   9.88025 |
| Leotheras the Blind       |       22 |   29.70819568634 |    7.7082 |
| Fathom-Lord Karathress    |       19 |  29.344209671021 |   10.3442 |
| Morogrim Tidewalker       |       19 |  29.168962478638 |    10.169 |
| Lady Vashj                |       25 |   33.73619556427 |    8.7362 |
| High Warlord Naj'entus    |       35 |  31.309191703796 |  -3.69081 |
| Supremus                  |       35 |  31.311571359634 |  -3.68843 |
| Gurtogg Bloodboil         |       35 |   31.48274230957 |  -3.51726 |
| Teron Gorefiend           |       35 |  31.392440795898 |  -3.60756 |
| Shade of Akama            |       35 |  31.330605745316 |  -3.66939 |
| Reliquary of Souls        |       37 |  31.544477939606 |  -5.45552 |
| Mother Shahraz            |       35 |  31.572957754135 |  -3.42704 |
| Illidari Council          |       35 |  31.577703475952 |   -3.4223 |
| Illidan Stormrage         |       40 |  39.489924252033 | -0.510076 |
| Al'ar                     |       25 |  30.309156894684 |   5.30916 |
| Void Reaver               |       18 |  26.475796699524 |    8.4758 |
| High Astromancer Solarian |       25 |  30.688435077667 |   5.68844 |
| Kael'Thas Sunstrider      |       30 |   34.34583568573 |   4.34584 |
+---------------------------+----------+------------------+-----------+

You can read the full update here: Development Notes

Solari · 08/05/07, 2:03 AM

Originally Posted by Moshne

I have browsed this thread a bit, and I am not seeing this, but is there a way to "report" guilds that are clearly out of line to assist in getting it straightened out? We just had a guild on our server break into the "Top 1000" US guilds with a single server transfer giving them 5-6 kills.

We've just had this problem too on Thaurissan, with a transfer with Mount Hyjal items joining a guild stuck on Vashj and bumping them up ~400 rankings. I had a little think about this issue, and a potential solution may be to wait until a guild has two items attributed to two different people before recording the kill and updating the rankings accordingly.

A more accurate but perhaps more troublesome option would be to simply include a "report error" button, but I can see that that might create a few troubles for the armory crawl, and also when that guild finally kills the boss that the transfer had brought loot from initially.

Aspir · 08/05/07, 3:21 AM

Originally Posted by Solari

We've just had this problem too on Thaurissan, with a transfer with Mount Hyjal items joining a guild stuck on Vashj and bumping them up ~400 rankings.

I'd actually addressed that specific situation earlier today. The guild showed up in my reports and I made some adjustments. You just don't see it in the rank listings yet because I haven't regenerated them since the change, but if you go to the guild's page you'll see Hyjal no longer shows up.

songster · 08/05/07, 3:41 AM

Originally Posted by Aspir

I'm assuming you mean 'Aspir', not 'Antiarc' -.-

Whoa, I'm sorry - mother of all brainfarts there! Thanks for the info on changing scores.

Edit: a quick look at the new scores shows that the absolute numbers have changed a bit, but the general principles are the same - the major drop-off in raid terms is moving past Maulgar/Gruul, and there are quite a lot of people getting Maulgar down before Netherspite and/or Nightbane. The new score also seem to reveal a further divide between those tackling lower Kara and those on upper Kara - again a fairly obvious conclusion.

Interestingly, when looking at realm scores rather than individual guild scores, the lowest peak has very much smoothed out. Loooks like that peak was partly 'due to realms with very low numbers of guilds - setting the bar at 20 means there aren't 10 to average together, and the numbers get artificially(?) lowered.

Fifty · 08/06/07, 4:21 PM

I'm sure i don't stand alone saying EU players are eagerly expecting WoWJutsu coverage, i know it's completely drawn out but any news from the frontlines?

Another thing that would make browsing WoWJutsu allmost perfect is the search option, i'm especially interested in how exactly do you plan to make the search engine work, and most importantly when would be my question?

tedv · 08/06/07, 4:32 PM

Originally Posted by songster

As I understand it, wowjutsu only removes a group if it actually /gdisbands. A large proportion of (1) may well be guilds that got Maulgar/Gruul/Magtheridon down and then died because of the attunements issue. Guilds with VR down are more likely to be "live" guilds, since he's only really been accessible for a short while. If part of the "three boss" spike is down to dead guilds, then that signal will hang around in the data set even if the live guilds make progress into SSC. I guess we'll see in the coming few weeks.

Antiarc - any sane way of addressing this issue? Flag up guilds that have no kills on their "top level" bosses over the last month, perhaps? Obviously if a BT guild stops killing SSC bosses, that's no biggy. If an Gruul-killing guild stops killing Gruul, that's clearly indicative of something else!

I've been thinking about this issue as well. The best idea I had was to remove a guild from the listings if they had no new loot in two weeks, naturally adding them back if they resume again. Determining the "highest tier" bosses for a guild is a difficult algorithmic problem, but checking for no new loot at all is easy, and should be a great indicator of most guilds that have stopped raiding.

Aspir · 08/06/07, 4:34 PM

Originally Posted by Fifty

I'm sure i don't stand alone saying EU players are eagerly expecting WoWJutsu coverage, i know it's completely drawn out but any news from the frontlines?

I was really wanting to be able to release it tomorrow, but I have a feeling I might need to push it back until Wednesday, so I can fix some more of the existing data errors.

Another thing that would make browsing WoWJutsu allmost perfect is the search option, i'm especially interested in how exactly do you plan to make the search engine work, and most importantly when would be my question?

Search is one of those features I really want to add, but I need to make sure not to add poorly so that it doesn't bring my server to its knees. It's relatively high on my list, but not the top of the stack right now.

My focus is first on finishing stuff so I can release the EU data, and then working on improving the rate at which wowjutsu can gather the guild data. Search is somewhere shortly after that.

tedv · 08/06/07, 4:40 PM

Originally Posted by Aspir

Note: I just finished making a first pass at updating the WowJutsu point values based on the statistical data, so this will affect any further interpretations that you do

...

You can read the full update here: Development Notes

Can you explain the actual algorithm you chose to determine relative point values? I'm guessing it has something to do with the relative proportion of guilds that have access to engage a boss versus the percent that have killed the boss, but I can't tell exactly from the data.

Cannings · 08/08/07, 5:20 AM

Originally Posted by Aspir

I was really wanting to be able to release it tomorrow, but I have a feeling I might need to push it back until Wednesday, so I can fix some more of the existing data errors.

ZOMG Its Wednesday!

ikillyouheal · 08/08/07, 7:58 AM

I've noticed a "bug", when you click a link to someones armory on EU, it tries to open their armory through the US armory site. try "http://armory.wow-europe.com/" instead of "armory.worldofwarcraft.com".

Nice going

Giske · 08/08/07, 8:19 AM

If you click on a european realm it shows "US rank" instead of "EU rank".

Egel · 08/08/07, 8:44 AM

Originally Posted by Aspir

Regarding ranking based solely on Tier X progression and time killed
I don't think this kind of approach would fit very well into the current system primarily because we have no indication from the Armory as to what time a piece of loot dropped, only when the Armory last updated a given guild/character, so when multiple guilds kill a boss on the same day I won't have an authoritative way to know which kill actually occurred first. Now perhaps this doesn't happen a lot on the high end of raiding, but I'm sure it will happen a lot at the lower end. Actually not even that long ago there were two US guilds who got their Illidan kills hours apart, and I had to refer to websites to determine which actually came first (since I lock ranks at Illidan). I'd like to limit these types of situations where I have to manually figure out which event comes first.

I still think number of bosses and then time of last kill is the best ranking system also for your site. If you can't distinguish two guilds just let them share place, or manually check the time, or use your item point ranking as last resort if you absolutely don't want guilds to share place.

To speed up the crawl you could search the level 70 top ranks in top guilds more often but maybe you do that already.

Last Resort killed Archimonde for the first time a day or so after we cleared everything btw

woo-haa · 08/08/07, 9:02 AM

Where should we report if there are some kills missing? The kills from two timers ago have not yet updated on our guild page.

Otherwise, I was very glad when I saw the EU numbers come today. It gives back the lost feelings of competition within a server.

Dehn · 08/08/07, 9:09 AM

Was very nice to see my own casual gruul-slaying guild ranked when I checked it today. Keep up the good work... (And I could BET you 10$ that Blizzard will have something similar within a year if it goes popular..)

Khamul · 08/08/07, 9:19 AM

With the european version comes a little bug.

I'm on the realm "La croisade écarlate", the realm link url is http://www.wowjutsu.com/eu/lacroisade�carlate/ , which page is not found.

Of course, it happens with every realm name containing an accentuated character.

(In that case, the url http://www.wowjutsu.com/eu/lacroisade%e9carlate/ works as intended)

Thanks anyway, we're pleased to see the EU support

alkis · 08/08/07, 11:46 AM

I have a small request:

Instead of indexing the top 500 guilds from the US and EU in the daily, can you maybe index 250 for each continent and then the top 10 or 20 from each realm? This will keep intra-realm competition at a higher level and it seems more important than the competition for the 250-500th position throughout EU or US.