The Real Problem with Parsers

[JackFross]

As for your guide, you mentioned you used "math". I think it's safe to assume you've used a program to help you do that. So how exactly would this be different from parsers? Parsers simply simplify it for this sole reason. It does add one benefit. Namely numbers. Numbers are much easier to grasp for the guide's effectiveness, rather than a "theoretical guide". Without the end result (the numbers), how could you prove your guide's better, if not just as effective, as another guide?

I forgot to address this point. The "program" I used is nothing other than pen-and-paper, automated by just running the calculations in excel rather than doing them by hand. I directly take the potency amounts in-game, use the GCD cooldown I have, and, using these numbers, calculated what would be the optimal rotation between a few different variants I came up with. Dragoon is simple in that it has 2 combos and 2 skills out-of-combo that are halfway useful. (Full Thrust/Chaos Thrust and Heavy Thrust/Phlebotomize, respectfully)

I reasoned out that Fracture is a waste of time and TP, because its potency per gcd is less than every other potential skill you could put in that spot (listed above) - INCLUDING Heavy Thrust, when you add up all those 15%'s that it gives on any of the other GCDs.

None of this required parsing - none even required verification by parsing. Numbers don't lie. Using simply the potency amounts, I also was able to calculate when it would be optimal to use buffs such as Life Surge, simply by mathing it out hard enough. I didn't have to parse using it in each place in a long fight to see. I just extrapolated using "perfect" rotations with "ideal" timings and decided which skills it would be useful to use it on. It was a loss to use it for Chaos Thrust (where it used to be a gain) but a gain to use it on either 4th hit OR Full Thrust, so long as you hit the positionals.

All of this has merit regardless of parsing or not, because I considered all other cases. At least 90% of other cases are thrown out before you even start because 90% of the skills on Dragoon are less potency per gcd than using either of those 4 blocks described above. From there it's just a matter of figuring out how to maximize the use of those 4 blocks. There's only so many permutations that exist when you only have 4 pieces to slide around, especially when 2 out of 4 of those are static damage regardless where you put them.


All of this described above comes with numbers.
Heavy Thrust is 170 potency per gcd, not counting buffs.
Phlebotomize is 170+30x? potency per gcd, not counting buffs.
Chaos Thrust combo is (180+220+250+290+35x?)/4 potency per gcd, not counting buffs.
Full Thrust combo is (150+200+360+290)/4 potency per gcd, not counting buffs.

The ?'s depend on your skill speed and the rotation you end up putting together.

In practice, we can calculate the Full Thrust combo's potency, including our base buffs as Dragoon (Heavy Thrust's +15% damage and Disembowel's 10% Pierce debuff resist drop):

150+200+360+290 = 1000 potency
1000 / 4 = 250 potency per gcd
250 x 1.15 = 287.5 potency per gcd (with Heavy Thrust)
287.5 / 0.9 = 319.44 potency per gcd (with HT+Dis)

That's a number I can compare to the others and see which combo is stronger. I didn't use a parser to get that number at all. All I "automated" was the last two calculations.

When you set up a rotation, it will give you a specific potency per gcd. You can compare these numbers to the numbers produced by other rotations to find out which one is strictly best without doing the fieldwork of testing each one a hundred times and averaging the results to give dps numbers spat out from a parser that estimates dot ticks.

How is it different from a parser? It's calculated using nothing but the numbers the game gives me in tool tips without "automating" anything aside from flat calculations (Though a calculator hardly automates anything). A parser scrapes the data from the game and compiles it, automating everything from data gathering to the calculations of damage amounts. These are two very different methods, though they look similar if you paint in broad strokes.

[Malevicton]

snip

Calculating your optimal potencies shows you your hypothetical ceiling. If you wanna know how well you're actually applying that knowledge then you need to analyze your combat logs.

[JackFross]

Calculating your optimal potencies shows you your hypothetical ceiling. If you wanna know how well you're actually applying that knowledge then you need to analyze your combat logs.

Sure. That's correct, to some extent.

However, once you have the base rotation and understand the base mechanics of the class, you've reached the skill floor. In general, someone at the skill floor isn't going to be noticed as being "bad," even with parsers. The skill gap is immense for most classes, and I can only speak for Dragoon on this, but...

If you just do the base rotation of Dragoon with absolutely no buffs (even ignoring GSK), I can still cruise at ~960 dps on a dummy. This compares to my 1350+ resting dps on dummy using buffs and ogcd attacks. If that 1350 translates to 1100-1200 in any given turn of Alexander normal mode, the minimum skill variant I outlined here would likely translate to ~815 dps. Granted, I'm i203, so my stats are quite higher/better than, say, an i180 player, but these numbers should still hold (somewhat) for someone who's been buying Eso gear weekly and has the rest 190. They'll be ~100 or so less, but meh. Same story as below.

No one will call him on his shit. It's low, yeah, but nobody is gonna say "BRO U SUCK" they'll just shrug it off and move on. And that's simply doing mechanics correctly and performing your base rotation while using literally 0 cooldowns for the entire duration. Nothing. No jumps, no Geirskoguls, no Life Surge, not even a Blood for Blood.


Basically, the mathematics I'm talking about isn't related to calculating a hypothetical skill ceiling, it's about calculating the very real skill FLOOR for a class. The base rotation you should be doing to perform the class correctly. So, uh, essentially - you don't need a parser to tell you that you're playing your class according to how the current meta tells you the optimal way to play it is. You'd need one to see yourself IMPROVE beyond the baseline, sure, but to know you're there? Not necessary, imo.

[Viridiana]

In practice, we can calculate the Full Thrust combo's potency, including our base buffs as Dragoon (Heavy Thrust's +15% damage and Disembowel's 10% Pierce debuff resist drop):

150+200+360+290 = 1000 potency
1000 / 4 = 250 potency per gcd
250 x 1.15 = 287.5 potency per gcd (with Heavy Thrust)
287.5 / 0.9 = 319.44 potency per gcd (with HT+Dis)

And then you get in game and realize that your calculations are off, because 1) potency scales weird and Full Thrust isn't 12x as much damage as Phlebotomize ticks, even without buffs, and 2) that Disembowel should probably be X*1.1, not X/.9. At least, that's what the community's been using.

[JackFross]

And then you get in game and realize that your calculations are off, because 1) potency scales weird and Full Thrust isn't 12x as much damage as Phlebotomize ticks, even without buffs, and 2) that Disembowel should probably be X*1.1, not X/.9. At least, that's what the community's been using.

I dunno what numbers you're looking at, but the ratio of Full Thrust to Phlebotomize ticks is ~12x in-game. If you're comparing HT+Dis Full Thrust v HT+Dis Phleb, then the difference will be larger, since the Phleb dot ticks aren't piercing damage and thus aren't boosted by Disembowel. Potency scales perfectly linearly. There's RNG, of course, but the ratios are consistent, according to potency amounts. You also still don't need a parser to see these relationships. I sure didn't use one to test it out and figure out that potencies scale linearly with damage. (Since I was playing on PS3 when I did these initial calculations back in 2.0-2.1)

It should be /.9, if the game works the way the tooltip explains. It decreases piercing resistance by 10%, it doesn't increase piercing damage by 10%. These are different statements. Just because other people do it one way doesn't necessarily make it correct.

For the record:
x1.1 = x11/10
/0.9 = x10/9

Those numbers are different by 1/90. Doesn't really make much difference, except that dividing by the 0.9 is more accurate, assuming the game works the way it states in the tooltip.


This is getting a bit off-topic. The point still stands I didn't use or need a parser to calculate the numbers that give you an optimal rotation for Dragoon and Paladin. I haven't bothered looking into other classes yet, since those are my only two 60s, so no other ones matter to me at the moment. You don't need a parser to discover the skill floor for your class. You do need one to see how you're progressing toward the skill ceiling. It's really really obvious if a player isn't at the skill floor yet for Dragoon, at least to my eyes and ears. I've been playing the class since 2.0 beta phase 4. I know the sound effects and animations for all the pre-heavensward skills like the back of my hand. If someone isn't reaching that baseline plateau it is *painfully* obvious to me.

And if someone is at or above the skill floor, there's no real reason to be a dick to them, because at least they're playing "properly." If they're below, that's cause for you to offer constructive criticism and back down if they get offended.

The problem isn't parsers being illegal, it's players being too pigheaded to admit that they're doing something wrong. Hell, I've even had players respond with "So? We still cleared." after I tried to offer advice on how they could play Dragoon better because they were dealing about 50% of their potential. Maybe that would be fixed if they could see their numbers. Who knows?

[Viridiana]

Potency scales perfectly linearly.

Hahahahahaha...No. No it doesn't. Except when it does, sometimes. But usually it doesn't.

For the record:
x1.1 = x11/10
/0.9 = x10/9

Those numbers are different by 1/90. Doesn't really make much difference, except that dividing by the 0.9 is more accurate, assuming the game works the way it states in the tooltip.

And the former is what all the guides go with, and is probably more accurate. And a difference of 1/90 will start giving you wrong numbers as soon as your damage exceeds 90 points, which was forever ago. Once you're up past 900 damage abilities, you're looking at 10 points off from expected value.

[JackFross]

Hahahahahaha...No. No it doesn't. Except when it does, sometimes. But usually it doesn't.

When does it not? I can't say I've ever seen such an example. I'd like for you to provide one, since I'm legitimately curious. It would be more helpful than acting smug and pretending to know something I don't know without providing anything to back you up.

And the former is what all the guides go with, and is probably more accurate. And a difference of 1/90 will start giving you wrong numbers as soon as your damage exceeds 90 points, which was forever ago. Once you're up past 900 damage abilities, you're looking at 10 points off from expected value.

It *should* be such that you'd divide by 0.9 rather than multiply by 1.1, according to the tooltips. I had never done testing of my own to prove it, that's just the way it should work, if the tooltips are actually accurate to the way the devs set up damage calculation. This is simply because:
No Disembowel:
100% dmg v 100% resistance = 100/100 = 100% damage
Disembowel:
100% dmg v 90% resistance = 100/90 = 111.1111% damage

However, I just ran tests on it hoping to confirm the way the tooltips say it. I was proven wrong. I didn't use a parser for this, since I feel like all of this was just an attempt to bait me into doing so. I scraped the data from my combat log by hand and entered it into excel to run the calculations.

http://puu.sh/koUcA/9108110013.png

There's more data points beyond the bottom, I just can't be arsed to clip multiple images together to show all of it. Came out to a 10% buff, so I retract my statements. x:

I included there a calculation using Sunny's damage formula to get an estimated potency for Heavy Thrust. That damage formula scales potency linearly according with damage. I feel like it would be far more disparate than 0.45% if this were a thing that doesn't scale linearly. To test THAT further, I ran another test with the whole Full Thrust combo:
http://puu.sh/koVh9/8b4d231584.png

The formula being used there is:

Damage Dealt = ((Potency/100)*(WD/25+1)*(STR/9)*(DET/7290+1)*BUFFS)-1

BUFFS in this situation is a flat 1, because the dummy has nothing special on it. But, yeah. Potency is weird and totally not linear at all? I'm really really confused as to why you would think that is the case.


And, again. All of this was done without the aid of a parsing plugin/program. A parser wouldn't have even sped things up, since the battle log as I set it up for my purposes is no less easy to read than a parse would be. Also important to note re: potecy scaling.

Vorpal Thrust:
Combo: 200
Not: 100
Diff: 100
Diff/Combo = 100/200 = 50%

Full Thrust:
Combo: 360
Not: 100
Diff: 260
Diff/Combo = 260/360 = 72%

These numbers are reflected in the battle log as the % by which damage is increased due to combo potencies. Everything seems to be backing that they scale linearly, so I'm gonna need a very convincing argument to show me that I'm wrong on this.

(how is this still on topic? I'm continuing to prove that you can calculate and discover the skill floor for your class *without* the aid of a parser, so the argument that you need one to know if you are even meeting bare minimum for your class is baseless)