The risk of relying on Combat Log Parsers

[Syrano]

MMO communities have always amused me.  Primarily because of all the affection, association and ego we put behind the abilities of our toons.  Based soundly on the premise that all toons of the same class are created equal with little concept for the player behind the toon.  We tend to stack ourselves into our class categories, wave the flags of "Nerf!" or "Uber!" around.  Last night my group and I were playing around with combat parsers and there were some serious lessons learned.First, my chanter friend and I were using the same parser, the advanced combat tracker (found at: http://act.webhop.net) .  It's very cool, has some amazing graphs, the ability to paste results into your game and it even works over a network so you can run it on a second machine. However, it became obvious that results are a mixed bag and dependant on the viewpoint of the parsing user.  Often we'd see a %20 swing in DPS between our two parses of the same fight, sometimes as much as %100.  There would be tmes where they were both identical but it was wise to keep in mind that DPS is very very very relative from the perspective of the parser.Additionally parsers are extremely in accurate in multi-encounter situations.  It has no way to know which mobs applied to what encounters, essentially the timer stops when it sees an encounter over message.  This can throw off all the numbers depending on who is doing what to which mobs.  Very very dangerous to rely on data from the parser in such situations.But the big issue the parsers highlighted is the amazingly, almost absurdly, different results for two players of the same class.I present the first case:Warlock A.  Wearing very good t5 fabled/t6 crafted gear. 57th level.  Not sure what ability is skills were at.  Safe to assume some adept III's, probably a master or two. He is a well known player and well equipped.  Averaged nearly 600-700 dps through most fights at Giants in PoF.  I'm not kidding you.  He was nuke happy.  Yes, he often got aggro early on, but things died so fast it didn't matter.  You didn't need a parser to see this guy was something special, he never stopped casting.  When we had a pally join and put his aggro transfer on the warlock, the pally always had aggro, warlock even complained of not being able to pull aggro.Warlock B.  Later Warlock A left, a new warlock showed up.  59th level, mostly crafted t6 gear.  In a respectable uber guild on my server.  Averaged no better than 180-200 dps, even after being told to just let fly.  There was a marked difference in mob kill rate.  This is the average behavior of most warlocks I've played with.  I can out dps this kind of warlock left and right.Conclusion of first case:A) Warlocks are machines when played well.  B)  Overnuking is not always a bad thing.  C)  Just because you play a dps class does not mean you actually achieve fantastic results.I present the second case:Illusionist A.  Spends a lot of time studying his spells, knowing their quirks and playing his class.  He can do some seriously amazing things with crowd control, stuns, stifles.  He averages nearly 220dps.  In an AE fight he can average 400dps and outdamage me.  How?  Simple, they get piggy back spells, he plops a few things on my that make him do more damage, along with spells like illusory allies and suddenly he's a damage king.Illusionist B. Barely averaged more than 100dps.  More along the lines of what I expect to see from illusionists focusing on group support, CC, buffs, etc.Conclusion:It's all in how you play, but DPS is not king, it is not the only measure of ability.  Just because a class is not generating major dps does not mean it is not being useful.Other notes:Back to the first case.  When we had Warlock A in the group, my own person DPS was exceptionally pathetic.  On my own parser I average about 250-350dps.   It can vary dramatically at times.  When warlock A was in the group I was barely doing better than 100-150.  Why?  I couldn't get a bloody target.  by the time I changed mobs and went to attack, they were dead.  My AE's would barely trigger before the mobs would die, often I'd have to rely on Lucky Ruse alone because the cast time on Swathe was too long.  This warlocks presence made me look like a [Removed for Content].It was a classic example of how a dps parser and a given night, circumstances, mobs, and group make up can radically change results.  The simple truth is that DPS results are so varried and so reliant on other factors that at most you can only take general trends and specifics from them.   They are not absolutes.  They are not the be-all-end-all of authoritative information. Things only get worse in large raids where 24 people may not all be in range of your parser, your logs.  All the damage a mob does or receives may not be detected.  especially in large AE fights where you're running in and out of AE range.  At the end of the day, DPS measures are so utterly subjective they are hard to trust.Add to that the simple fact that player behaviors can very so wildly that just because one person of class A is a damage kin does not mean another person is.  *How* you play your class is just as critical as what the class is. I see a lot of illusionists complaining about their damage output for example, yet I can point to one who routinely outdamages most people in the group.  I have heard a lot of warlocks complaining about not doing enough damage yet I can point to one that blew the doors off of things.  It was so eye opening how varried people who play their class was...Edit:  In re-reading this I covered at least three topics in no particular order.  Ah well, such is the internet, and posting from work <img src="/smilies/3b63d1616c5dfcf29f8a7a031aaa7cad.gif" border="0" alt="SMILEY" /><div></div>

[DanaDark]

<div></div><p>I agree! I am a swashy and have been parsed as doing more dps than a ranger 5 lvls above me... yet in the next group the guardian had 100 dps over me most encounters...</p><p>I dont trust those programs anymore lol</p>

[TwistedFaith]

Ok this is a VERY interesting post.I myself play a 60 ranger as my primary class, he was my first class not one of these wannabe rangers who seem to spring up.Anyway our guild raided Drathar last night, I parsed the encounter. I was placed into a DPS group with a troub for his hate reduction and another 60 ranger.Well to put it blunt unless the Advanced Combat Tracker was well off with it's numbers, the 2nd ranger should have been ashamed of himself. We were both using legendary posions and decked out in rare t6, I have all Adept3's and a few masters thrown in.I'm not going to post actual figures BUT like the original poster said, there are some people who know how to play their class and others who just get by. I know exactly when to hit Streams of Arrows, I know exactly what buffs to hit jsut before I use streams to maximise my dmg.

[Raveller]

<div></div>Good info. I play my Swashy to have fun, not to crunch numbers.

[Keldo]

if the parser is innaccurate, you need a new parser!But overall I agree with you, in a single group setting especially in eq2 a combat parser does not serve much use.However, its function will always be invaluable on raid targets, if not just for parsing DPS, but for figuring out AE timers, mob resistances/output and other nuances.<div></div>

[massem]

<span>I would be pretty sure that those who parse high, are those that  regularly parse themselves and take time to study their spells/combat arts. </span><div></div>

[Zygwen]

<span><blockquote><hr>masseman wrote:<span>I would be pretty sure that those who parse high, are those that  regularly parse themselves and take time to study their spells/combat arts.</span><div></div><hr></blockquote>Regular parsing is not necesary to parse high. Just need to understand the mechanics of the game and use your character to its fullest potential. Syrano is essentialy saying that the average player does not push the limit of what their character can do.Back in my EQ days as a beastlord we had members that didn't just parse they used statistical analysis to get as close an approximation of how the real mechanics works as possible. What people are doing now in EQ2 pales in comparison to the amount of work spent to understand the game mechanics in EQ. On the other hand, I'm sure many of the Top raiding guilds have done a lot of this analysis and are not releasing all their info because it gives them a competative edge.</span><div></div>

[massem]

<div></div><blockquote><hr>Zygwen wrote:<span><blockquote>.....</span><hr></blockquote></blockquote><p>Yes you are right - one must still think and try various things yourself if to improve - but parsing is a good way to confirm if your thinking was right :smileyhappy: </p><p> </p>

[Syrano]

<div></div><div>Interesting how the post was interpreted.  Yes in some ways I was definately pointing out that there is a dramatic difference between two players, but not so much to berate folks for not trying hard enough.  More so that you have to take someones "parse" with the grain of salt.  No two players are going to parse the same even if they are playing the same class. Whether or not they are skilled or not.</div><div> </div><div>Additionally, the accuracy of the parser is not in question.  It's the simple fact that range to the parsing player has EVERYTHING to do with it.  Your log is range based, if someone goes in and out of its range, your parse will be horribly inaccurate.  This is, actually, most noteable on large raid mobs where 24 people are scattered all around the mob, especially ranged fighters.  The logs maximum range is well within visible distance.  Just something to keep in mind.  I definately do NOT trust parsers on raids for my DPS unless I'm running the parser myself.  Nor do I use it to assess others performance on a raid.</div><div> </div><div>They are very useful for overtime analysis and attack pattern analysis on mobs, resist types etc.  Just well, I'm always leery of "I out dps'd a XXXX" or a "I got out dps'd by a XXXX" posts. Too many variables.  Way too many.</div>

[Keldo]

<div></div><div></div>Actually range of damage of the log is about 50m, which is about 10m father than you ever need to be from a raid mob.  As long as the person parsing the damage is in range of the mob, the range of any other player does not matter - the damage is reported from the mob, not the player.  So even if they are halfway across the zone, if you are near the mob as say one of their dots goes off, it will still register.In EQ1 this was a MUCH larger issue as log range was miniscule at best.Also I really don't want to disagree with you here, because most people who post those types of ' I was outdamaged by xxx in x situation', are new parse users that don't know about the variables.  But, in the hands of someone who is experienced  at analyzing a parse, information you can gather will prove useful in the long run.  So, just because some people don't know how to use it, doesn't mean they cannot, or should not be sourced if they are used correctly.  I guess the real moral is, just don't take an unknown parse as word, take some time to learn it yourself if you are that interested!<div></div><p>Message Edited by Keldoth on <span class="date_text">01-23-2006</span><span class="time_text">03:21 PM</span></p>

[AratornCalahn]

I say, your a rouge - that means you /flirt with everyone you can, talk to as many memebers of the opposite sex as possible and try to look really really cool ALL the time.Sure hit the mob a few times, but dont forget to run up to the healer and /flirt a bit. DPS? thats for rangers and assassins to worry about - cos they dont got style <img src="/smilies/3b63d1616c5dfcf29f8a7a031aaa7cad.gif" border="0" alt="SMILEY" /><div></div>

[DonSavan]

<blockquote><hr>Syrano wrote:Additionally parsers are extremely in accurate in multi-encounter situations.  It has no way to know which mobs applied to what encounters, essentially the timer stops when it sees an encounter over message.  This can throw off all the numbers depending on who is doing what to which mobs.  Very very dangerous to rely on data from the parser in such situations.<hr></blockquote>I have my parser set to end after about 6 seconds of no combat message just for these situations.