Parser ResultsFollow

Parse results aren't stored in a mass database, players who parse have them saved on thier individual computers. That being said, it's very difficult to accurately measure efficiency of gear A to gear B with a parser due to the large variations in player skill/aggression/buffs/debuffs/etc. Using a single parse as an "end all" for an argument is dumb.

Parses are good for measuring:
accuracy
avg dmg (melee/WS)
counters/avoids/etc

All of which can also be estimated via theorycrafting and normalized to "ideal" players for a much better indicator of general results.

I didn't assume they were stored on a large database. I do, however, recall seeing a decent amount of parses being displayed here a long time ago on a somewhat regular basis. Maybe it was just during a kick when lots of people were displaying results.

I also didn't want to use a single parse to be the end-all authority on the subject. I figured the best thing to do would be to dig up a few examples just to have as that: a few examples to refer to. They wouldn't be the basis for the claims completely, but they would certainly help as some bit of "back up" for those who believe the claims but don't actively parse themselves.

Edited, Jun 10th 2009 2:47pm by TheTang

parses can be interesting, but their greatest use is in confirming theorycrafting. "competitive parses" don't accomplish much, even if you're very controlled or parse against yourself. parsing is good to confirm sh*t like "theorycrafters say i should have 93%~ ACC at this camp. let's see about that..." and such. but when it comes to "which weapon/item is better?" theorycrafting and knowledge of what stats do decide that.

honestly, over something like 2-3 years of parsing the majority of my XP parties and a lot of events, i've probably done 5 parses or less that were actually enlightening to me in any way.

the main one i can think of concerns haste. for a long time, i (like anyone else who grasps what haste does to delay) acknowledged haste's increasing returns and its general value, but like a lot of people i refused to accept that small amounts of haste could make a difference in a short fight. starfox helpfully (and often rudely) explained it to me repeatedly on alla, but for some reason i didn't understand his explanation (if you're interested, the explanation is that for any variable you think will introduce bias and stop haste's benefits from being realized, there's at least 5 more variables that reintroduce enough randomness in fights for haste to be realized. for example, even if you and your friends always autotarget together, you won't always do the same number of JA or WS, and mobs will be different level and have different HP, so there's always that random situation where 3% more haste means 3%~ more chance of hitting when you wouldn't have, and you do, 3%~ of the time).

it took me a long time to grasp this explanation (which is a bit embarrassing), but i accepted starfox's (and other people who claimed % increase in attack rounds per minute translated directly into % increases and DoT) because of parsing. there's a guy on phoenix named 'gates' who is a long time MNK like me (also a really nice guy). we used to party in KRT, and i always remembered him as the only MNK who could outparse me, even though my gear was often inferior to my PT members (knowledge, aggressiveness, and proper gear/macro/food choices make a HUGE difference for DD). as we partied more often, i noticed that he and i were equally aggressive in WS, equally attentive to macros, equally mindful of positioning for autotarget, equally good and making quick choices to maximize damage. he also did HNM sh*t either before i did or with more successful groups than i did at the time, so he ended up with byakko's haidate before me, black belt before me, etc. when we would parse, he'd always have 1-2 superior pieces of gear to me, and would always win.

during the height of my arguing against the benefits of small amounts of haste, i remember parsing a few parties with him when he had black belt and i had brown belt. in every one, he outparsed me by about 5%-7% (depending on buffs), which is more or less exactly what the theorycrafting predicted. i'd even look at our total melee damage or number of hits, and he got just the amount more you'd expect. these are as controlled parsing conditions as you can get; identical playstyle, identical macros and gear, identical merits, identical attitude toward damage dealing. to me, the results meant that, whether i grasped why haste cleanly gave the results it did or not, haste just did work the way starfox said. that was the main parsing experience i had that made any difference at all to my thoughts about the game.

other things that interested me were seeing the DoT potential of PLD/NIN with joy/justice, getting outparsed by WAR/SAM and seeing where his damage came from, and maybe a couple others like that. the really interesting thing about parsing is the examples it gives of damage distributions (melee, WS, enspell, whatever). on the one hand, it informs theorycrafting by reeling it in and showing you what realistic parties are like, mathwise. on the other hand, it lets you see the problems in people's playstyle; for example, whenever i see people post parses proving that their hot shot DD "is better than BB monk," i almost uniformly notice that the MNK in question averages her/his asuran fists anywhere from 100-200 damage lower than mine. you see what people aren't doing, and what's holding them back.

but say you want to compare hagun and shinsoku. the role of parsing there is to show damage distribution and such. you'll never see the actual comparison from a parse. people play too differently. i could maybe bring out such differences by parsing with gates, but we're not studio gobli or ffxi scientists; we're just average players who want to have fun (and he and i don't even really know each other or interact too much because we're never in the same ls). even then, the parses would only serve to confirm the theorycrafting or show what may have been misguided about theorycrafting. the parse itself is not the deciding argument against a piece of gear.

so, if you want to help ls members or win gear arguments, learn what stats do. think of things in terms of % increases to DoT (that way you can usefully compare disparate damage sources like haste, ACC, STR, DA, etc), learn the kind of trends that operate with these things, and work on being able to lay out explanations of them concisely so people don't just tune you out. actual parses usually say more about player than gear.

Edited, Jun 11th 2009 2:40am by milich

exactly. unless you're actively involved in trying to refine theorycrafting itself (which isn't a huge discipline; there's only so much to know about this game), parsing offers two pretty simple benefits: seeing what's up with gear/food/etc choices, and having meaningless (but often fun) epeen comparisons. no dis on parses, but that's pretty much what they're good for.

if you find dynamic uses for them or things to investigate, definitely use kparser, and get in touch with kinematics for advice. he's endlessly helpful and friendly, and his parser is great.

here's 2 relevant links if you have some spare time (both involve a lot of chaff with the wheat).

concise explanation of jitter and other variables explaining small % haste getting realized by starfox in this thread.

lots of vitriolic arguments (but more important, many computer simulations of merit parties to illustrate starfox's point) in this thread. i may just be being arrogant, but i think that i raised all the most powerful anti-small amounts of haste arguments (generally more conceptual than expressly mathematical) in that thread, and still lost the argument.

(bonus thread with testing, which happens to be the thread i found the link to the old fumas/dunes thread in.