Sunday, February 26, 2012

A DPS Meter Fallacy

In a comment, Kring posted something that I would like to address:
Besides that it, again, encourages all the tricks to "cheat" Recount like
- not target switching
- dotting unimportant targets
- using AoE spells where not useful
- not moving our of fire
- not using non-dps abilities like decurse or spellsteal
In my view, the idea that one can "cheat" or "pad" the meters is something of a fallacy. It's technically, or hypothetically, possible. But in practice, it's extraordinarily unlikely.

If you are good enough to push yourself near the top of the meter by padding, you are good enough to top the meters through normal play.  

Topping the meters requires knowing and executing your rotation, using cooldowns appropriately, not dying, and casting as many spells as possible. Padding the meters is essentially executing the same steps, only substituting the "padding rotation" in place of your regular rotation. If you can't execute your normal rotation correctly, you won't execute your padding rotation correctly. If you aren't casting enough normal spells, you won't cast enough padding spells. If you aren't popping enough cooldowns in normal play, you won't pop enough cooldowns in padded play.

To reiterate, playing to pad the meters on a boss fight is the same process as playing normally. A player who is having trouble playing normally is going to have the same trouble when playing to pad.

The harsh truth is that DPS meters work!

The people who are consistently at the top of the DPS meters are your best DPS players. They will be the ones who are better at interrupting, at using utility spells, at moving successfully. The players at the bottom of the meters are the ones who need the most improvement, who will also be less likely to interrupt correctly, to use utility spells correctly, to move out of the fire.

There is a point where this is not true, where class and spec determines your position on the meters. That point is known as Paragon and the other Top 10 guilds.  For everyone else, position on the meters is determined by the quality of your play.

21 comments:

  1. I take issue with the concept that it all gets boiled down to one number that can be "padded", when anyone of at least average intelligence can easily see through those sort of tricks.

    You can't trick recount. You can only trick the person using it.

    ReplyDelete
  2. The problem is, people who ARE good enough to interrupt, etc., don't always have the inclination to do so. As Tamarind of Righteous Orbs said before, "Nobody ever says to you after a raid, 'Sick utility, man, sick utility.' "

    People exist who will go the extra mile, on their own, to pad meters and have the capability to do so. But only when specifically requested by name to cast so and so CD (that hurts their meter ranking) will they do so to help the raid.

    ReplyDelete
  3. The fallacy of DPS meters is not "padding*", but the false impression it gives people as to what's important. Is the person topping the meters really the one who did the most damage, or was it simply the one who ignored the important mechanics? Or was it the one that fight favoured? You can't look at the final tally of the top 10 DPS and say "Yep, the guy at the top sure is the best", because that's not what it's saying at all.

    Unfortunately, I have yet to really come across anyone who uses a full group comparison in the right way, so in the end all a DPS meter does in those situations is generate drama.


    (* The use of "padding" in this case is to follow your intention of a player who goes out of his way to deal extra damage purely to top the meter. This does not include situations where a player legitimately does a ton more damage because he used those abilities as part of fight mechanics; IE AoE classes getting massive burst DPS on fights with a lot of adds, or DoT classes that can spread damage across two targets to increase their total DPS.)

    ReplyDelete
  4. It isn't easily possible for anyone to see through people not debuffing so they can throw some extra dps instead, or ignoring the kill order so they can sneak in some more AE attacks.

    Possible to check via something like recount? Yes. Easy to check? Not really, no.

    Not to mention people whining about being asked to CC or off tank because it's going to affect their dps numbers.

    And the one thing with damage meters I never did quite understand is why the first couple of seconds seem to make such a huge difference to the final numbers. I remember all the dps emphasising this to me when I was tanking and asking them to hold off a couple of seconds so I could get solid aggro. Apparently this would have made such a massive difference to dps meters that it wasn't a fair ask. And I still wonder what's with damage meters that a second or two at the start can somehow translate into tens of thousands of dps difference at the end of a fight.

    I don't entirely trust meters.

    ReplyDelete
  5. Padding the meters really only happens during pugs or farm bosses where people are just screwing around or being lazy. Basically when the meter actually doesn't matter people "cheat" it. When you are working on progression raids it is rare to see people cheating the meters and they are soon removed if they do.

    @spinks: when I raided there was never any worry about the tank's threat as long as there was at least 2 people who could do tricks or misdirection. We let the tank get the first hit, maybe 2 hits before we opened up. The exclusion to this was low geared tanks which we gave 5 seconds to. Rule of thumb was if you are gaining to much threat as a dps player you were doing something wrong regardless of the numbers.

    ReplyDelete
  6. In normals it's more importanat that DD survive than make an OMG dps.

    ReplyDelete
  7. Sorry to "necro" the original topic (about the red, yellow, green performance dot). I don't think that Blizzard should spend any time implementing something like that. Blizz should be focused on delivering new, interesting, and challenging content. It should be up to the users -- in the form of add-ons -- to develop such a tool if it's really useful. The "market" thus far hasn't dictated the usefulness of such a tool. Recount and WoL are already pretty powerful; the only issue is ever configuration to an informative view. If you want a rotation analyzer, there's already stuff like CLCret.

    I'd rather Blizz develop a couple new raid bosses or new features or almost anything than build in a zumba type tool that's only ripe for abuse.

    ReplyDelete
  8. I have to disagree that those at the top of the meters "are better at interrupting". Where do you get that idea?

    Seems to me those most often at the top of the meter in a pug are precisely the players who ignore interrupts,CCs etc the most, and you won't see "Mr Meter Spam" running back to Rewind Time on Murazond for example, because that is a sure-fire dps loss.

    ReplyDelete
  9. Perhaps if there were a way to calculate "DPS per GCD", a "DPS equivalency" could be assigned to those GCDs spent on utility casts and rolled into an "effective DPS" figure.

    I mean, someone figured out how to roll absorption into healing figures, right? Why not this as well?

    ReplyDelete
  10. I could not disagree with you more.

    I think that someone that strives to be the top DPS at the expense of the kill is really not a good player.

    I have experienced it over and over. People that will not switch to adds and cause wipes simply because they could not be bothered to have their DPS be number 2 or 3. Mind you everyone is withing 1 or 2K of each other, but some people care more about that top spot than they do a kill.

    The excuse I hear over and over is that the other DPS should be able to deal with the adds/interrupts etc.

    That kind of DPS mentality simply states that DPS is what kills the boss, when in reality teamwork is what kills the boss.

    I proved it to our team on our second raid the other day on an alt run. They thought that having DPS was all that mattered and we went with a 4 healer (we had 2 very very undergeared healers on the group) 4 DPS 2 tanks comp. We executed, still beat the enrage timer and did it working as a team, not having all DPS.

    I see teams trying to do this all the time with Zerg strats to eliminate mechanics. Sure it is a numbers game at the end but I still hold the idea that teamwork and utility are always more important than raw DPS numbers.

    ReplyDelete
  11. I could not disagree with you more.

    I think that someone that strives to be the top DPS at the expense of the kill is really not a good player.

    I have experienced it over and over. People that will not switch to adds and cause wipes simply because they could not be bothered to have their DPS be number 2 or 3. Mind you everyone is withing 1 or 2K of each other, but some people care more about that top spot than they do a kill.

    The excuse I hear over and over is that the other DPS should be able to deal with the adds/interrupts etc.

    That kind of DPS mentality simply states that DPS is what kills the boss, when in reality teamwork is what kills the boss.

    I proved it to our team on our second raid the other day on an alt run. They thought that having DPS was all that mattered and we went with a 4 healer (we had 2 very very undergeared healers on the group) 4 DPS 2 tanks comp. We executed, still beat the enrage timer and did it working as a team, not having all DPS.

    I see teams trying to do this all the time with Zerg strats to eliminate mechanics. Sure it is a numbers game at the end but I still hold the idea that teamwork and utility are always more important than raw DPS numbers.

    ReplyDelete
  12. I agree that only a good player can top the meter. But we're not talking about topping the meter, we're talking about not getting the red light. Or probably even not getting the yellow light. Or even playing the meter because of fear you might get the yellow/red light.

    Let's assume your an average player who runs LFR. You get the green light and your happy. But you don't know if you still get the green light if you "waste" some time to interrupt a spell on the next boss. Let's assume you did some interrupting on the next boss and now you got the yellow light. And you promptly get insulted in raid chat by the mage in the group. What will you do on the next boss? Will you interrupt again? Maybe the mage will try to kick you if you're only yellow again.

    There are some easy ways to squeeze in some extra damage, like:
    - Not moving out of fire and continue casting. We all did that. The difference is that a good DD is able to judge if it can and will be healed through.
    - Start casting before the tank is on the mob. "He can use spot." We all did that. The difference is that a good DD knows when he can do that and when not.
    - Using AE when focusing the healer would be more appropriate.

    > The people who are consistently at the top of the DPS
    > meters are your best DPS players. They will be the ones
    > who are better at interrupting, at using utility spells,
    > at moving successfully. The players at the bottom of the
    > meters are the ones who need the most improvement, who
    > will also be less likely to interrupt correctly, to use
    > utility spells correctly, to move out of the fire.

    That's normally correct. But what if the game tells you that it doesn't matter if you interrupt. It's more important that you stay above x dps and you get the green light. Yes, you didn't interrupt a single cast but the game considers you a superstar (green light).

    ReplyDelete
  13. There are 4 good counter arguments against your post. 1) playing the meters 2) class/spec/glyph specific (dis)advantages in general 3) class/spec/glyph specific per fight or mechanic of fight 4) learning curve. All 4 are described in this post.

    Before I reply I want to make sure people understand the damage per second isn't affected after you or the boss is dead. The damage done stops only counting when the boss is dead or wipe. A dead DPS does no DPS. The amount of damage done is more important than DPS, everyone should remember this.

    "Topping the meters requires knowing and executing your rotation, using cooldowns appropriately, not dying, and casting as many spells as possible."

    No, sorry, it does not. Not _always_. There are legio exceptions; they are the problem.

    I would not want to argue the ability to press the button "5" for "swipe" or "mind sear" equals e.g. "knowing and executing your rotation". I don't believe a fury warrior is a worse player than a feral cat just because the fury warrior cannot AoE the small adds on Yor'Sahj. Maybe he has wrong spec, betting going arms, but 1) we cannot expect such for LFR and normal 2) if the adds go down because we got proper AoE DPS and he does good single target DPS he is doing his job.

    Also, activity does not say much about the damage done.

    You claim: "In my view, the idea that one can "cheat" or "pad" the meters is something of a fallacy. It's technically, or hypothetically, possible. But in practice, it's extraordinarily unlikely."

    Even though you are presented with examples. I don't understand how you can be so ignorant. Have you never seen it happening?

    There are far too many situations where you can "play" the DPS meters.

    ReplyDelete
  14. 1) You are a fire mage on Spine of Deathwing LFR. You cast TW, cast LB on 3 tentacles to get pyromaniac up. You then get a combustion up on the first amalgamation, and spread it around. You will be high if not top DPS, but you are not performing. Most likely, people have called you a retard in raid chat, and rightfully so. People are unable to perform a barrel roll in LFR (though in the first week I thought it was required) so it'd be a wipe. Now, I have done this very thing. Why? Because some other player did not wish to wait for my friend (who was tank) to be ressed. Instead, said specimen decided to start the fight right away. What happened was we wiped, I got vote kicked for this, while my friend was able to continue. I did not care, since I only needed gear from the previous 2 bosses.

    2) Another situation which arose was on my arms warrior. In the situation of my mage instead of DPSing anything I went AFK. If I recall correct they started the fight quickly after kill, and I went away to take a drink. When I came back I had AFK behind my name. my bossmod was making noises for the amalgamation exploding, and an other warrior was saying they should kick me afterwards as I was AFK. After Tendon was dead I looked at the DPS. I did high DPS on the Tendon. Now, here is the point: the DPS on all the other stuff is trivial and does not matter. In a normal or heroic attempt you only should care about the DPS on the Tendon. The entire talent tree and gear is optimized for this! As a side note I explained the warrior he could have his 2H from last boss as I was loot banned for it. After that he suddenly became silent. I won the shoulders with crit, wished the warrior good luck on acquiring the 2H, and left.

    3) Yet another situation on my warrior, this time I was slacking on Madness. I was AFK on the first platform, doing nothing. I was completely useless. If I recall I was on toilet. Ugh, details… from the 2nd platform and onwards I did perform, but I found myself low on the list. Then the regenerative came. I popped Recklessness and Bladestorm, lou and behind, I was in the top 5 of the list all thanks to the fact that A) my spec does insanely high AoE DPS with my CDs B) I do this if others don't do insane high DPS (have seen a mage going 90k here). Now now, was I skilled here!

    In the 3 cases above the DPS meter by default does not show any meaningful numbers. Their feedback is completely skewed, but if part of the default UI then people would see it as the authority of performance, reading the numbers wrong and draw the wrong conclusions. Why not leave meters to the educated few who are able to install and read them correct?

    ReplyDelete
  15. "The people who are consistently at the top of the DPS meters are your best DPS players. They will be the ones who are better at interrupting, at using utility spells, at moving successfully. The players at the bottom of the meters are the ones who need the most improvement, who will also be less likely to interrupt correctly, to use utility spells correctly, to move out of the fire."

    Sorry, wrong. I am a DPS and I find it more important that when learning a new fight the tactics are executed correct. You will find me going for survivability (which includes interrupting, moving) over my performance as DPS. The reason for this is simple: I know my rotation inside out, I just need to apply it on the boss fight. I have zero experience with the boss fight -tho from theory I know what to do, you gotta work it out your own group- therefore performing DPS wise will come easily after we got the tactics settled. The other reason you are wrong is that different builds, talent trees are good for different fights. A fire mage with pyromaniac and improved fire blast has 4 completely useless talent points for Morchok HC and Ultraxion HC, but they are useful for example on Hagara HC and Spine HC. Still, I'd much better prefer cauterize for these fights. Or even go arcane for some of these fights (and I hate arcane). Yet another example is Baleroc HC 10m. I was always one of the lowest, if not lowest DPS here. First of all I cannot multi dot here, second as a ranged without mobility like disengage and blink I am in disadvantage to others including the melee who barely move. Third, we cast heroism at start, and this is where I disperse. Four, I use TB trinket to help the healers. Was I the worst DPS player? No, I was actually doing our healers a great favor by following this tactic.

    "There is a point where this is not true, where class and spec determines your position on the meters. That point is known as Paragon and the other Top 10 guilds.  For everyone else, position on the meters is determined by the quality of your play."

    Once again, wrong. Morchok HC you want mitigation healers. Without it, you're doomed (resto druid for example is not good there) especially when gear levels were lower and people had less stamina and such. You also may want people to respec for this fight, for survivability. Mage with cauterize, feral druid with increased armor, priest with inspiration (I actually went disc after having holy since patch 4.0.1). On Zon'ozz, you don't want too many melee. On Ultraxion the feral and rogue cannot stand behind him, leading to a loss of DPS. On Madness, multi dot is ace. On Spine, resto shaman SLT is ace. We had great success with our DK DPS going tank on certain HC fights. On Hagara HC a hunter with aspect of the pack and nature resi are great additions. And so on, and so on. There are simply certain classes/specs which are better than other on certain fights. Dare I say, figuring these mechanics out is one of the most fun things in WoW raiding; making the fight work for YOUR team. You will never hear me (seriously) say wow I am the best DPS, cause I am the top DPS on Madness. That is because I realize it is the mechanics of my class which allow me to be that way, and second because every time I make a mistake in my rotation (as small as it may be) I realize I could've come up higher. My 4set also makes my rotation much more challenging.

    To sum it up. Well, I did so in 1st post 1st line.

    ReplyDelete
  16. "The people who are consistently at the top of the DPS meters are your best DPS players. They will be the ones who are better at interrupting, at using utility spells, at moving successfully."

    This may be true in a guild environment and where padding the meter is a black and white thing, in that the people who do the most dps are the people who are the best in dps'ing. However in a non-guild environment as in LFR and pugs, padding the meter is not what you said in your post. In that some of the people who may be at the top of the meter are not the people who are executing correctly, they may in fact be the people who don't switch on add thus PADDING their numbers.

    ReplyDelete
  17. I have never, nor will never fully trust a DPS and/or HPS meter. All of my MMO career people have been shoving them down my throat and as a guild/raid leader I've rarely found them useful in truly judging someone's skill.

    There are ways to cheat a meter. The healer that AoE heal's a ton and then tries to make the tank healer feel bad comes to mind.

    A meter is suppose to be a tool that helps a group get better. It is rarely used that way I'm afraid. That is why my guild doesn't take it too serious. At the end of the day, your DPS doesn't matter if the monster is dead! It doesn't surprise me that people try to cheat it though.

    ReplyDelete
  18. Meters mean very little by themselves. If you aren't taking in account the context of the situation, such as someone is kiting, or someone is on dispel duty so their healing output will be lower, then you're doing it wrong, plain and simple.

    Just like the stoplight meter is bad because it doesn't contain enough information, so do raw DPS/Heal meters without a breakdown of what they were attacking and their rotation (which both Skada and Recount give to a large degree, btw).

    I use meters all the time as a raid leader. In a guild situation, as mentioned before, they're absolutely useful and helpful and one can be guarenteed that I am taking context into account. In an LFR situation, meters are less helpful, because context is lost as people do things like stand in the (quite weak) fire.

    Basically, in cases where performance absolutely matters (Normal/Heroic modes with enrage timers, higher damage, and tasks such as dispels or target switching for context), Rohan's argument about needing meters still stands. One cannot improve without concrete feedback. In cases where performance frankly matters little, such as LFR, meters do nothing but feed people's egos.

    So the question goes, then, do we ruin the feedback mechanism for the people using it correctly and opt for reducing the amount of information so that people in LFR and Heroic 5 Mans can't tear their fellow party members a new one verbally for not being up to their standard of play, or do we allow the tools to exist for those who want to use them and use them appropriately to work on more difficult content, but this allows less savory players to use the tool as a bludgeon in content where performance is frankly not that big a deal?

    Damage meters are like the Triforce in A Link to the Past:

    "The Damage Meter will grant the wishes in the heart and mind of the person who touches it. If a person with a good heart touches it, it will make his good wishes come true... If an evil-hearted person touches it, it grants his evil wishes. The stronger the wish, the more powerful the Damage Meter's expression of that wish."

    ReplyDelete
  19. The way I've always seen it is that you really can't pad the meters, since typically the "padded" damage still needed to be done. You might complain because someone is using AoE or multi-DoTs, but the truth is that those adds probably needed to die anyway. I have plenty of people complain about AoE on the Madness of Deathwing encounter, because I'm using my AoE on the bloods while they're dealing single-target damage to the tentacle. The bloods need to die eventually, and so what's wrong with me trying to take that opportunity?

    ReplyDelete
  20. "The bloods need to die eventually, and so what's wrong with me trying to take that opportunity?"

    You'd have someone who is good at AoE do AoE, and someone who isn't you let single target. Called doing your role, doing what you're good and useful at. If someone cannot do either you replace them, and if you don't have enough AoE you replace some single target/burst with AoE and vice versa (respec, or do what you are less good at but still do it). As for Madness: Spellweave. If you are a lock you use RoF or SoC on it (not Hellfire) and proc spellweave. This is why locks, and DoT classes in general, are so insanely high on Madness (esp heroic). This is a DPS increase, and necessary, but it does not mean you are more skilled than the sub rogue who single targets and does during heavy AoE less damage than you. In fact, as hinted if you have too much AoE dmg and not enough single target that can be a serious problem on in this example Madness. It could mean you have no problem with getting Regenerative down quickly, but the Blistering on 4th platform just don't get enough DPS, healers cannot cope with it, and wipe. Or you are not able to nuke the corruption or bolt quickly enough because not enough burst.

    Plus some adds do not have to die. There is one tactic where on 4th platform Madness you ignore the Blistering. If someone does not follow the tactic while rest does he screws the raid. On phase 2, raid leader says you stay on Deathwing, but you ignore and start nuking the little tentacles, then terror (as if you could solo that all, but OK). AoE, high burst, fully HC geared it does not matter in that case you're as useless as someone who is DCed. Another example is chain lightning on Spine, to proc fulmination, to nuke Tendons. It does AoE dmg which is useful for you (for proc), but not effectively good. Shows on meters. Or if in some fight you are asked to single target 1 down and then AoE. You ignore, all stay up too long, oops wipe. Or if you take Ascendant Council for example. A high DPS who has no clue can really fuck up the fight. Been there, done it. Maloriak HC I remember doing as fire mage. So I finally got some nice procs and just when I got pyroblast and combustion up I want to spread it and splat, dead. There went my AoE dmg. Heck, on Spine HC there are tons and tons of adds. You really don't want to nuke them all down though. And finally, there is Alysrazor, where the tanks and the people who go up do more than the DPS who stay down and the healers. Does not mean they are more or less useful, it says fuck all about their skill. On heroic Lord Rhyolith, Al'Akir I also did insane AoE dmg. Did not mean I was better than the rest of the raid. Not even all AoE dmg was useful, and the use of it wasn't always in my control either.

    In this and other posts I gave so many examples, all from Cataclysm. I don't understand how people can be so stubborn and ignorant to ignore them. Every single example I gave is valid. They all skew the numbers in a way where without proper analysis it does not rate the highest "damage done" as "the best damage dealer".

    By the way, the same is true for healers. Healers also have assignements, and CDs. You can do 30k HPS on Madness, if you fail to pop your Aura Mastery or Hand of Sacrifice + bubble on the right moment (which by now you should know, if not ask) then you failed, you suck, and if you do it again I'd replace your useless ass and get a healer who is smart enough to pop CDs when required. I once did Madness on my resto druid. Me and the other druid healer popped Tranq on exactly the same moment. We stayed on boss. I don't know about him (he had 8/8 HC, good gear) but I was OOM, it was my last option. We did not communicate, no voice chat, and we wiped with something like 2%. If we did not pop it at the same time, we would both have healed more and it might've been a kill.

    ReplyDelete
  21. Rohan, I disagree with your assumption that 10M damage is as important as 10M damage. I think it's a fallacy; consider that the DPS purpose is not only to kill the boss (and adds) dead but also to prevent the DTPS on the raid and DTPS in most encounters changes over time based on the encounter timers and tactics employed by the raid. Hence, a DPS that does the same damage during the encounter but does less damage at times when DPS contribute little or even negatively towards damage taken, contributes less towards the kill.

    Consider a DPS on Madness that does not switch from the tentacles/head. While she decreases the damage taken thanks to killing them sooner compared to lower DPS that switches, she also increases the damage as she makes the mobs stay on low health (when they deal more damage) longer.

    While she might be good enough to keep her position on meters even if she played normally, she definitely contributes less than she would if she did the same DPS but switched targets. That means that DPS who do (slightly) lower damage but follow the tactics contribute more but show as worse on the DPS meters.

    The difference might not be large - actually I have no idea how to compute the DPS modifier that has to be applied for each case of not following the tactics - but it is there.

    ReplyDelete