Guild Wars 2

I am posting the following for Broadicea because he has no access to internet while he is in prison
—————————————————————————————————————————————————————————————————————-

Hey folks. A lot of pure theory crafting gets thrown around in this forum and others, unfortunately in the PvE side we don’t have much actual testing going on. Some of you think we don’t need it. May those of you who believe that be confined for all eternity to the abyss of Tartarus.

For the rest of us, designing a good field test brings several challenges. Firstly, we obviously have to do this in a PvE mode, since limitations on builds in PvP (crit chance/damage mainly) rule out using the golems. Within that setting, we need to be as isolated as possible from interference. We don’t need John and Jane Q. Puggers running over and ruining what we had set up halfway through. Since many of our builds and calculations rely on full team bonuses being provided, we need the rest of our team there. How many others we need will be discussed later on. We also need enemies that won’t really fight back, since the amount of time you spend dodging is a huge variable that really can’t be adequately controlled. We all know we will dodge during most fights, but due to aggro mechanics, how much can vary from enemy to enemy, group to group, even build to build. I would suggest larger health pools as well, since more time spent on the trial will smooth out inconsistencies in rotations and other RNG factors.

There are of course other aspects to consider, the above is by no means meant to be an exhaustive list. Due to some of those issues, I believe moving on from Veteran Risen Giants is in order. The last time we had this conversation, some good ideas for targets were put forth. I think we should discuss those, as well as determine the optimal setup for the rest of the team that will be providing buffs.

The Target

Keeping in mind some of those challenges I mentioned above, I believe a target should fit these criteria

-large health pool (champions or legendaries only)
-isolated (within an instance)
-meatbag (able to do the whole fight with minimal or NO dodging and not worrying about damage mitigation)
-level 80 (no downscaling)

Right off the bat, being level 80 confines us to HOTW, CoE, or Arah. Examining these:

HOTW
None of the bosses I know of in story or the three explorables are total meatbags. Any fight against them will involve lots of downtime due to kiting/dodging/healing. Also, their health pools are excessively large, which leads to fatigue and overall not giving a kitten anymore before you finish.

CoE
The golem boss is a potential candidate, but it’s a bit of a messy fight, with the possibility of getting lasered or fisted (huehuehue) by the boss. The protection buff also drags this out longer than is necessary. You could assign your team to destroy turrets, I suppose. Food for thought. Subject Alpha obviously is not ideal, due to the dodges needed. The Evolved Destroyer though…now we’re talking. He’s a total punching bag. All you need is a guardian giving you aegis or just WALK out of the tooth area and you’re safe. The only downside to this target would be the need to leave and reapply the lasers to drop his shield. With only one person attacking you would likely need several shield drops to finish him off. One other downside would be that he’s so near the end of the path. The ability to get to a target quickly is a big draw. Probably the best candidate in CoE though.

Arah
Rather than draw this out I’ll just jump to spoj’s suggestion of the Ooze boss in path 1. By utilizing the coral safe spot, he becomes the ultimate meatbag. It’s right in the beginning of the path- needs only a few minutes from entering the dungeon to starting the fight. The rest of the team is very safe here as well, which makes it easy for them to standby and provide might and fury buffs. As far as other candidates in Arah, I believe Magecrusher and the Abomination are also potentials. Anyway, unless someone else can be shown to surpass the Ooze, that’s my pick. Perhaps the only downside would be not being able to use proper sigils. Does it share an undead classification? If not, Force becomes the only damage bonus gear, which hurts builds that stack multipliers (and thus rely on Force/Night+Slaying).

The Team

So let’s discuss the ideal team setup as well. We need a way to provide 25 stacks of might, perma-fury, and ideally 25 stacks of vulnerability on enemies as well. There are also the unique team buffs to consider. Due to that last bit, I believe there team should, at a minimum, have 1 ranger and 1 warrior. Considering the cooldowns involved, 2 warriors might be ideal in order to alternate usage of signet of fury and/or healing surge for Longbow F1. Blasting fire fields will be our main source of might generation- a thief could be a good pick to use SB2 on fire fields provided by warriors and ranger (torch 5). Fury is taken care of by the warriors using warbanner/FGJ, and the ranger using warhorn/red moa. The Ranger is obvsiouly using Spotter and Frost Spirit as well.

Vulnerability is the one area that might be difficult to maintain for the duration of a solo fight. How well could one engineer and two warriors maintain it? They could perhaps equip white weapons and no armor to reduce their damage as much as possible while still allowing them to “attack” and apply vulnerability. Is that feasible at all? If so, the engineer should take the thief’s spot.

Obviously boon duration (condition duration for the vuln stackers) would help here. The whole team should probably consider going naked with white weapons anyway, just to be sure there is minimal interference. Don’t forget to turn off auto-attack!

So anyway, these are just some starting ideas to get some good controlled environments to test out different builds. We can (and should) also discuss other aspects of good experimental method, such as using multiple trials to average times, proper recording techniques, etc. The point is, let’s come to an agreement for a standard and start using it.

Or Anet could just put in a kittening Master of Damage and we’d be done with the whole mess.

I would like to know the numbers too if you find some . Some people say that thief has highest single target DPS in organized groups . Even if i never been in the super mega zerk group i’m pretty sure it’s a lie . Because you can’t spam backstab over and over and the autoatack on daggers is garbage to And the sword.is too slow compared to warrior’s axe.

might want to do an actual test and see the numbers before jumping to conclusions

Two Engis, one elementalist and one warrior. Noone needs ranger buffs.
Simin is probably better than ooze, able to use rush/wa then.

a t s e.9614:

I would like to know the numbers too if you find some . Some people say that thief has highest single target DPS in organized groups . Even if i never been in the super mega zerk group i’m pretty sure it’s a lie . Because you can’t spam backstab over and over and the autoatack on daggers is garbage to And the sword.is too slow compared to warrior’s axe.

well with thief you don’t just spam backstab, it’s based on positioning (be behind them), and work backstab into your rotation. I believe above 50% target HP the proper rotation is C&D, backstab, then a full dagger auto chain, then just the FIRST hit of dagger auto, then C&D, etc. Below 50% I’m pretty sure it’s just heartseeker while above 6 initiative. Correct me if I’m wrong, pro-thieves.

Bringing up thief brings to mind a good point though, and a potential issue with the Ooze. Positioning is necessary to get proper benefit of DPS traits for both thief and ranger. Not sure how that can be overcome.

We used to replace one ele with a thief in our AC runs until that guy went off to play some other games. That is mainly because thieves have better damage modifiers though.

Broadicea.8294:

well with thief you don’t just spam backstab, it’s based on positioning (be behind them), and work backstab into your rotation. I believe above 50% target HP the proper rotation is C&D, backstab, then a full dagger auto chain, then just the FIRST hit of dagger auto, then C&D, etc. Below 50% I’m pretty sure it’s just heartseeker while above 6 initiative. Correct me if I’m wrong, pro-thieves.

That is the sPVP rotation, where revealed lasts 4 seconds. In PVE, reveal lasts 3 seconds, so you can fire off C&D immediately after one full auto attack chain. Though I haven’t done a full DPS check between HS and BS…

Anyway, I’m surprised the OP dismissed HotW so quickly.

In path 3 of HotW there is a champion so easy that the strategy for him has been “don’t bother” since release: The Champion Icebrood Goliath. His attacks are slow and incredibly low damage for their speed, meaning that nearly anyone in any gear setup can facetank him for incredible lengths of time. Certainly long enough to get an accurate baseline for DPS, even if he does have quite a lot of HP.

Though if the ooze is as good as you say, then there is only one advantage the Goliath provides, and that is being an Icebrood. You can use the Powerful Potion of Icebrood slaying to get a full 10% damage bonus (and 10% off of his attacks). Unfortunately it is still a day dungeon, though.

a t s e.9614:

Heartseeker is below 25% . Backstab still does higher damage at 50% . The Thief DPS is high yeaa , but i wonder if people would take a thief in their speedrun if they don’t need to skip anything ?

Attack speed comes into play a lot aswell. Thief has highest single target DPS. Thieves are something that should be utilized more…They will be in dungeon running guilds once eles get a nerf instead of a buff. Thief, can CnD Backstab combo for 20-35k every 4 seconds, and can HS span below 50% for around 15-25k. Thief is the highest single target. That all said the only thing they can provide a team other than stealth is vuln on crits…but they would have to drop some of its burst with Mug to get that…
A t s e I’m not gonna make snide remarks at you this time, but before you start jumping to conclusions like earlier posts get the numbers. Warrior is a good DPS class but gets beat by many in team situations, but the reason warrior DPS is so powerful is they self buff like a mad man, where Necro and Thief dont.

Dub.1273:

Two Engis, one elementalist and one warrior. Noone needs ranger buffs.
Simin is probably better than ooze, able to use rush/wa then.

It would be nice if contributions to this conversation carried a little more depth than “do what Dub says because he said so”.

Blood Red Arachnid.2493:

Anyway, I’m surprised the OP dismissed HotW so quickly.

In path 3 of HotW there is a champion so easy that the strategy for him has been “don’t bother” since release: The Champion Icebrood Goliath. His attacks are slow and incredibly low damage for their speed, meaning that nearly anyone in any gear setup can facetank him for incredible lengths of time. Certainly long enough to get an accurate baseline for DPS, even if he does have quite a lot of HP.

Though if the ooze is as good as you say, then there is only one advantage the Goliath provides, and that is being an Icebrood. You can use the Powerful Potion of Icebrood slaying to get a full 10% damage bonus (and 10% off of his attacks). Unfortunately it is still a day dungeon, though.

Thanks for the suggestion. I haven’t run p3 outside of getting the DM title so I honestly didn’t remember. That sounds like a good suggestion though, and the fact that it’s daytime is outweighed by being able to use slaying sigil and potion.

How long does it take to get to that champ?

Harrahou.4518:

Attack speed comes into play a lot aswell. Thief has highest single target DPS. Thieves are something that should be utilized more…They will be in dungeon running guilds once eles get a nerf instead of a buff. Thief, can CnD Backstab combo for 20-35k every 4 seconds, and can HS span below 50% for around 15-25k. Thief is the highest single target. That all said the only thing they can provide a team other than stealth is vuln on crits…but they would have to drop some of its burst with Mug to get that…

The thief can provide a few more things other than stealth, DPS, and vul on crit. My main dungeon runner is the thief, and since I usually pug, I have found excellent use in most of the thieves abilities:

Permanent Blind Field: disables all non-champion enemies quite well.
Cleaving Weakness in sword Auto Attack
Defiant stripping Pistol Whip/Headshot
Spammable blast finisher w/ Cluster Bomb
Boom Stripping* with Larcenous Strike + Bountiful Theft
AoE endurance regen with agility signet + Bountiful Theft giving long duration AoE vigor
Great rezzing abilities with Shadow Refuge
Projectile reflection w/ Dagger storm (also out-heals enemies with signet of malice)
Projectile stopping + blind with Smokescreen
Targeted Pulls with Scorpion Wire

All in all, a good thief is someone you want around. There are some disadvantages with the thief, though:

#1: They don’t directly increase DPS that well. Most of their utilities are defensive, with the exception being that boon stealing can steal might away, as well as remove protection. Otherwise, they’ll have to blow initiative on cluster bomb in fire fields to give everyone might.

#2: Thieves become redundant really fast. All that stuff I listed above can be accomplished by one thief. Having two thieves doesn’t do it any better.

Because of this, in any sort of DPS test, unless you are testing the thief, you won’t see a thief.

*Note: boon stealing has recently been cut in half. Because of this, I am seriously considering going back to a 15/30/0/0/25 build to get Bountiful Theft again.

(Not going to bother with quoting)

The champ goliath is the first boss of that map. You have to kill the initial wave of svanir (which takes, like, 10 seconds), destroy two ice totems, then you can just skip your way past everything else. The biggest obstacle to the goliath is that there’s a pack of icebrood wolves guarding the bridge to him, and they can tear a group up if they attack from behind, so before the real “DPS” test can begin you have to kill them and the adds around the goliath.

Blood Red Arachnid.2493:

Harrahou.4518:

Attack speed comes into play a lot aswell. Thief has highest single target DPS. Thieves are something that should be utilized more…They will be in dungeon running guilds once eles get a nerf instead of a buff. Thief, can CnD Backstab combo for 20-35k every 4 seconds, and can HS span below 50% for around 15-25k. Thief is the highest single target. That all said the only thing they can provide a team other than stealth is vuln on crits…but they would have to drop some of its burst with Mug to get that…

The thief can provide a few more things other than stealth, DPS, and vul on crit. My main dungeon runner is the thief, and since I usually pug, I have found excellent use in most of the thieves abilities:

Permanent Blind Field: disables all non-champion enemies quite well.
Cleaving Weakness in sword Auto Attack
Defiant stripping Pistol Whip/Headshot
Spammable blast finisher w/ Cluster Bomb
Boom Stripping* with Larcenous Strike + Bountiful Theft
AoE endurance regen with agility signet + Bountiful Theft giving long duration AoE vigor
Great rezzing abilities with Shadow Refuge
Projectile reflection w/ Dagger storm (also out-heals enemies with signet of malice)
Projectile stopping + blind with Smokescreen
Targeted Pulls with Scorpion Wire

All in all, a good thief is someone you want around. There are some disadvantages with the thief, though:

#1: They don’t directly increase DPS that well. Most of their utilities are defensive, with the exception being that boon stealing can steal might away, as well as remove protection. Otherwise, they’ll have to blow initiative on cluster bomb in fire fields to give everyone might.

#2: Thieves become redundant really fast. All that stuff I listed above can be accomplished by one thief. Having two thieves doesn’t do it any better.

Because of this, in any sort of DPS test, unless you are testing the thief, you won’t see a thief.

*Note: boon stealing has recently been cut in half. Because of this, I am seriously considering going back to a 15/30/0/0/25 build to get Bountiful Theft again.

Fair points, but I was going based on perfect group situations. Thief utility is trash apart from stealth. But when buffs required are covered and there is no point bringer more than one Ele, thief is the next best thing. Yes they bring blind fields and projectile mitigation. But in a group situation like I run in, thief would only be there to make up for lost DPS. Running one guardian is almost always enough blinds/reflects and a mesmer if one comes on the run. 25,30,0,0,15 is the ideal thief dungeon spec.

Wukunlin.8461:

Perhaps the only downside would be not being able to use proper sigils. Does it share an undead classification? If not, Force becomes the only damage bonus gear, which hurts builds that stack multipliers (and thus rely on Force/Night+Slaying).

The Ooze is considered undead but once again, using wrong sigils does not hurt builds stack multipliers at all.

Broadicea.8294:

Dub.1273:

Two Engis, one elementalist and one warrior. Noone needs ranger buffs.
Simin is probably better than ooze, able to use rush/wa then.

It would be nice if contributions to this conversation carried a little more depth than “do what Dub says because he said so”.

You get perma fury, perma 25 vuln, perma 25 might and warrior buffs. Although this obviously has a downside: You have to count attacks in your combat log, add their damage up and devide it by time spent. Engis and eles are doing damage while grantling might/applying vuln.

But anyway: Isn’t it just enough to find an afk fight (simin) and then do the whole thing without any vuln/might/buffs? The relation stays about the same, counterside is that bleed becomes way stronger than norman and that some professions like warrior apply vuln.

Need to control buffs really, otherwise we can just go back to killing vet giants in cursed shore xD.

Vuln is irrelevant if talking about direct dmg only. Benefit from leaving it out would be the ability to buff without causing dmg to target while still getting precise results.

Konu.1826:

Vuln is irrelevant if talking about direct dmg only. Benefit from leaving it out would be the ability to buff without causing dmg to target while still getting precise results.

Like I said few times before, there’s no reason to test with vulnerability, sigils, potions or ranger spirit unless you want to check what given build can supply.

Yes, and I’ve never disagreed with you about it, but for some unknown reason not all seem to agree they are irrelevant. While some of the tested builds would probably cause vulnerability in their standard rotations which would end up being a problem, I can’t come up with a reason why other dmg modifiers would be needed.

Well the thing is that sigils can be relevant too: a weapon set made of two one-handed weapons can have two sigils, so let’s say night and slaying (two 10% modifiers) while a two-handed weapon set only has one sigil; however two-handed weapon sets have 10% more weapon damage in general (greatsword) which compensates for the lack of second sigil. If we took all sigils out then two-handed weapon sets would have an advantage.

You should be able to remove the vuln from some of the builds without reducing the builds base dps. Doesnt matter about sigils and pots aslong as you use the same for everything. But for some builds you might use 2 1h weapons and have 2 modifier sigils. So you would have to take that into account. Its probably best to just leave sigils in so people dont have to go buy a weapon without a sigil.

Yes, I should have been more specific. Obviously the modifiers left out would have to be the ones shared by all the tested builds. Vuln might not be a problem for most of the builds but still that is something one might need to take into account when testing.

Dub.1273:

Broadicea.8294:

Dub.1273:

Two Engis, one elementalist and one warrior. Noone needs ranger buffs.
Simin is probably better than ooze, able to use rush/wa then.

It would be nice if contributions to this conversation carried a little more depth than “do what Dub says because he said so”.

You get perma fury, perma 25 vuln, perma 25 might and warrior buffs. Although this obviously has a downside: You have to count attacks in your combat log, add their damage up and devide it by time spent. Engis and eles are doing damage while grantling might/applying vuln.

But anyway: Isn’t it just enough to find an afk fight (simin) and then do the whole thing without any vuln/might/buffs? The relation stays about the same, counterside is that bleed becomes way stronger than norman and that some professions like warrior apply vuln.

Here’s why I don’t like Simin. To make the test not a pain in the kitten , you’ll have to bug her, and: 1) not everyone knows how to do it, 2) once knowledge of how to do it becomes more common, it’s extremely likely that it’ll get patched quickly 3) getting there in a reasonable amount of time requires really huge exploits, 4) see #2 again.

The coral at the Ooze will probably be patched eventually as well, but that’s been public for a long time and nothing yet has been done.

Also I think it’d be better to test builds out as close to actual conditions as possible without making the test difficult to set up. Hence, all buffs applied.

I only read through a bit of this so sorry if I’m off but engineers using white weapons will make no difference other than the stats off the weapons, kits are already modified to exotic pistol damage, this of course is based on using grenades/bombs to apply vulnerability, rather than something like the utility goggles toolkit. This may change your results.

Max might and max vuln is not so obvious.

If you test with max might and vuln: you punish builds that invest traits into creating might and vuln. Whats the point of investing points into a might generating trait when your control test is going to get you to 25 anyway? Works the same for vuln.

If you test without max might and vuln: you will find builds that generate their own might and vuln are vastly biased.

Example: With Might and Vuln, Thief single target dps will be absolutely insane since it doesnt waste any trait points unnecessarily into generating buffs.

Without Might and Vuln, a warrior build with Rending strikes and Forceful greatsword will be the top dps in the game, along with s/d ele and 100nade engineer.

So the question becomes, “which scenario most closely replicates real world gameplay conditions?”

Should start calculating team dps rather than class dps. And then we can simply test by making the composition fight a select meatshield boss and compare to the calculations.

spoj.9672:

Should start calculating team dps rather than class dps. And then we can simply test by making the composition fight a select meatshield boss and compare to the calculations.

That won’t be popular with the types of people who aren’t interested in hearing about the dps superiority of teams with a ranger in them.

Hmm.

You know, there’s something I remember from my time playing WoW that might help. At the time Elitist Jerks (the min/max collective) had a similar set-up to this using the meatbag Patchwerk. Patchwerk was much like your target, in that it didn’t require you to move around and dodge things, so you could just sit there and DPS race to your heart’s content.

But rather than just sort of fluffing off the differences that might exist between patchwerk and the whole rest of the game and the difference players may have on this outcome, they acknowledged their existence by creating a threshold. They did some field testing on ‘normal’ bosses with the exact same build across multiple people, and found a performance variance of (I think?) 14%. It wasn’t statistically significant or anything, nobody’s got that kind of time, but it just gave them a general benchmark they could use internally. Whenever a discussion was becoming more specific than a 14% difference between doing one thing and doing another, they acknowledged the difference was so minute it might not be repeatable from one person to another/one boss to another.

This logic tends to follow basic statistics. It’s very similar to the concept of an alpha level, or your willingness to accept you may be incorrectly assuming your data is more widely applicable than it actually is.

They avoided alot of chasing their own tails that way.
They also avoided having to prove their findings were applicable to the general public, by stopping themselves from engaging in tests or backing results that might have a high risk of being impractical in the first place.

Elitist Jerks <3

It’s a clever practice but it only applies to empirical tests, doesn’kitten

I always thought of it more like using empirical testing to avoid wasting your time experimentally testing a hypothesis nobody cares about. Is there another way to look at it?

Either approach is fine, actually so long as you remember that observable phenomenons always trump theory.

An example is economics where the current paradigm are microfoundations. Meaning, any macro theory you have is only considered valid if you can provide micro-level causes to explain the macro effect. Unfortunately, there are plenty of macro phenomena that we observe in reality that cannot be explained with micofoundations. What happens in those cases is, half the economists in the world say “that can’t happen” while the others say “…but it is happening.”

The difference is whether you use theory to describe how the world is versus to describe how the world ought to be. Once you venture into the ‘ought to be’ realm you set yourself (or whomever’s money you manage) up for literal disaster.

So in regards to guild wars 2, in an ideal situation you would measure kill times/clear times with the same players using different compositions and analyze the data to figure out why one composition was more or less effective than another.

Vox Hollow.2736:

They did some field testing on ‘normal’ bosses with the exact same build across multiple people, and found a performance variance of (I think?) 14%. It wasn’t statistically significant or anything, nobody’s got that kind of time, but it just gave them a general benchmark they could use internally. Whenever a discussion was becoming more specific than a 14% difference between doing one thing and doing another, they acknowledged the difference was so minute it might not be repeatable from one person to another/one boss to another.

I’m not completely understanding you I think, but are you saying that they let other people use the same build on a target and they found that there was a variance of 14% DPS?

I’m sorry, but if the error on your DPS calculations is 14%, in any statistical setup, that’s enormous. Statistically speaking, errors of <5% are acceptable, anything over 5% and your data is a bit iffy.

I can’t remember the number but it may have been more around 1.4% or something like this. WoW was a game where you could solve a pvp fight on paper, so in comparison PvE was like reading a book!

Nike Porphyrogenita.8137:

Either approach is fine, actually so long as you remember that observable phenomenons always trump theory.

An example is economics where the current paradigm are microfoundations. Meaning, any macro theory you have is only considered valid if you can provide micro-level causes to explain the macro effect. Unfortunately, there are plenty of macro phenomena that we observe in reality that cannot be explained with micofoundations. What happens in those cases is, half the economists in the world say “that can’t happen” while the others say “…but it is happening.”

The difference is whether you use theory to describe how the world is versus to describe how the world ought to be. Once you venture into the ‘ought to be’ realm you set yourself (or whomever’s money you manage) up for literal disaster.

So in regards to guild wars 2, in an ideal situation you would measure kill times/clear times with the same players using different compositions and analyze the data to figure out why one composition was more or less effective than another.

In a full path clear with all 5 members contributing, there are just too many variables to approach a useful analysis. I mean, the difference between 5 bearbows versus an organized team the differences are clear, but the differences between a 5:24 clear with W/W/E/M/G and a 5:17 with W/R/Eng/M/T are too subtle to be dissected by that kind of 30,000 ft analysis. That’s why I’d like to home in on the individual contributions. Of course a 2% difference in clear time doesn’t really matter for casual runs but they do for records (not to mention this provide great data for pvt losers to get upset about).

Zelyhn.8069:

I can’t remember the number but it may have been more around 1.4% or something like this. WoW was a game where you could solve a pvp fight on paper, so in comparison PvE was like reading a book!

A statistical variance of 1.4% is acceptable, and indeed negligible.

14% on the other hand…

Broadicea.8294:

Nike Porphyrogenita.8137:

Either approach is fine, actually so long as you remember that observable phenomenons always trump theory.

An example is economics where the current paradigm are microfoundations. Meaning, any macro theory you have is only considered valid if you can provide micro-level causes to explain the macro effect. Unfortunately, there are plenty of macro phenomena that we observe in reality that cannot be explained with micofoundations. What happens in those cases is, half the economists in the world say “that can’t happen” while the others say “…but it is happening.”

The difference is whether you use theory to describe how the world is versus to describe how the world ought to be. Once you venture into the ‘ought to be’ realm you set yourself (or whomever’s money you manage) up for literal disaster.

So in regards to guild wars 2, in an ideal situation you would measure kill times/clear times with the same players using different compositions and analyze the data to figure out why one composition was more or less effective than another.

In a full path clear with all 5 members contributing, there are just too many variables to approach a useful analysis. I mean, the difference between 5 bearbows versus an organized team the differences are clear, but the differences between a 5:24 clear with W/W/E/M/G and a 5:17 with W/R/Eng/M/T are too subtle to be dissected by that kind of 30,000 ft analysis. That’s why I’d like to home in on the individual contributions. Of course a 2% difference in clear time doesn’t really matter for casual runs but they do for records (not to mention this provide great data for pvt losers to get upset about).

I was talking pie in the sky ideal. In this fantasy we can run the path 1000 times with each build and get reliable data.

Yeah in WoW we were fighting to get 0.5% dps more than other groupmates so a 14% difference wouldn’t walk by untouched

Well, highest team dps possible would likely include a ranger and at most one warrior. In praxis, such a team would be a really stupid choice if you want to efficiently run dungeons.

Arah P1 Ooze may be a bad test subject for professions with rotations that require movement or positioning (elementalist and thief) or have pets (mesmers, necros, rangers). Maybe Subject Alpha 1 before the path is picked? I’ve never seen what happens if he’s taken past 70% health because by then I’m seeing Mount Maelstrom’s loading screen. The champ before that might be alright for a quick test, too, since the only attacks that need to be dodged only occur every 30 seconds or so. Simin would indeed be an ideal subject if it weren’t so annoying to get to her.

For DnT’s big number shots, we use 1 ele, 2 rangers, and 1 warr. The rangers can burst vuln for the time to get one or two skills off, but would be able to sustain approximately 10 vuln by just spamming rapid fire and hunter’s shot if traited fully into marksmanship and having eaten a koi cake. Warrior, if traited fully into strength and with 10 in arms and with a koi cake, can probably keep up 7 vuln easy. Ele just auto attacking with a scepter in water attunement with at least a 30/25/0/0/0 build and a koi cake can add another 5-7. Honestly, I’d like to see a guard instead of a second ranger for the variety of boons and easy burning that it can provide, and that would get us up to 25 vuln and give all the fire fields and blast finishers needed for 25 might for a little while. Of course, this is all just mental math on my part, and an engineer could easily replace the ele. Not sure on the CD on its might rotation because my engi hasn’t been seeing much combat recently.

How about a low level Bloomhunger? If you bring a guard, it’s pretty much a lolstomp anyway. Stationary target, low damage (as long as you walk a little), and a decent HP pool. It’s just not up against a wall.

DEKeyzToChaos.7381:

How about a low level Bloomhunger? If you bring a guard, it’s pretty much a lolstomp anyway. Stationary target, low damage (as long as you walk a little), and a decent HP pool. It’s just not up against a wall.

Isn’t his hitbox big enough to make every WW hit proc?

I mean, for FGS, he’s a bit too small, but still relatively big enough to get 80% of procs, but I think you should be fine for WW, no?

What’s the downside with using the Indestructible target golems in Heart of the Mists for controlled tests?

Well, just to give you some context, DPS in WoW at the time was self-focused instead of group-focused. So testing singular builds by themselves makes alot of sense in that game. I don’t really know what the GW2 equivalent of this testing would be, and wasn’t really accommodating for that.

Admittedly, I don’t quite remember the number. It has been quite a few years.
But I do know it wasn’t your typical .05 alpha, because it wasn’t an an actual alpha, it was just variance.

Basically they got an idea of how different people and events were from eachother, and used that as a reality-check to their calculations.

I only meant to draw comparisons to an alpha, because it shared similar qualities to being a confidence level you check against. I didn’t mean to imply they actually used variance as their alpha level in a real statistics analysis. You can’t get actual statistical significance in the Meta, nobody has the time to go mucking around with hundreds of people just to make a post about a single build.

But I always thought this clever little kludge did a pretty good job of acting in the spirit of it.

frifox.5283:

What’s the downside with using the Indestructible target golems in Heart of the Mists for controlled tests?

Crit stats are capped on the low side in PvP. Not to mention skills working differently due to splits.

Deks: Level 1 Bloomhunger might be a good choice, though ideally I’d rather not even have the chance of dropping below 90% HP. Is there a slaying sigil/potion?

Honestly, I just wonder what the point of a lot of this is, since it seems like many of our woes can be solved with math. I can see the use in many of these tests if you are going to figure out stuff like

Damage rotation: sum of attacks with higher damage/second than auto attack so that their total activation time is greater than their recharge. Or, basically every attack with higher DPS than auto attack if it doesn’t meet the first criteria. Buffs/debuffs handled as multipliers to the average DPS of the rotation for comparison.

Might sustain: base level of how many stacks of might are sustained given a certain rotation.

Vulnerability sustain: base level of how many stacks of vulnerability are sustained given a certain rotation.

Damage boosting thresholds: stuff like scholar runes or first strikes.

Condition sustain: uptime of burning, stacks of bleeding, etc.

And combining those to find an optimum attack set for any build. Once you figure out this baseline, then you can just apply math to figure out the other stuff:

Max might: It’s just additional power, which can be considered a percentage increase by dividing power at max might by standing power the build has otherwise. Something similar can be done with conditions, but it is highly class specific so I wont’ go into all the details.

Max vulnerability: it’s a 25% flat increase from nothing to max. For any build, it is 25-sustained vulnerability, and that is the percentage increase of your damage before applying vulnerability in the first place.

Sigils/potions: These are usually a flat percentage, which can be added on after figuring out everything above, if you use these sigils and potions.

Procs: These are already pure RNG, but follow a simplistic method: having twice the crit chance means double the procs, which means that non-cooldown procs are twice as frequent / twice as long, short cooldown procs get their additional time after cooldown reduced by a similar frequency, and long cooldown procs are nigh negligible anyway.

If you want to compare class to class, then after getting all of the sustains up there, you might as well go off of the wiki’s listed damage x 1.416 (2600/1836, or the armor adjustment for PVE) when looking at how much damage a class can do when compared to another. Compare their effective power, and then whomever is higher gets higher/lower multiplied to the wiki’s listed damage, then add on the DPS difference from sustained conditions rates, and you have your comparison.

The hardest part of it all is the DPS rotation and what condis/buffs it sustains, and once you’ve figured those out, then every other problem you encounter is simple multiplication.

Blood Red Arachnid.2493:

Honestly, I just wonder what the point of a lot of this is, since it seems like many of our woes can be solved with math. I can see the use in many of these tests if you are going to figure out stuff like

the point is doing it for real

While I am a complete maths addict I would agree that in-game testing is vital

Why don’t you do this in WvW? Just get someone from an opposing server to be your test dummy. The only problem is the low health pool.

A full toughness, vitality warrior with massive health-signet regeneration, perhaps? Maybe you can have others giving him perma protection. Maybe someone (or many someones) healing the test dummy as well with water field blasts — I don’t think there’s RNG in healing, so you just need to count how many times he was healed.

I don’t understand either why this is really needed. On static target math should get the same result. If it doesn’t then your math sucks and you should work on that. Once you know that your math works you can generate any kind of situation and try to simulate reality.

I understand wanting check if animations are timed correctly or that skills actually do what they say. Like calculate average DPS for Axe chain and then just Axe chain the Ooze to death. And see if they match.

But you forgot about the para-virtual force skewing the rules of maths in-game!

One thing that would be cool is, once “perfect” dps is known in an ideal situation with perfect rotations, you can measure your own deviations from that. If, for example, you kill a boss 30 times, and your dps is on average 10% below what attacking a static dummy, and other players report similar 10% loss, you could create a “difficulty scale” or rating or something for each boss. Might not be very useful, but it would be somewhat cool.