Thursday, June 14, 2007

Ninja Talk.

Another not so good day yesterday. Couldn't keep a connection for very long and I was tired, so just XPed my WAR a little bit then just went to bed. Didn't even get a level on my poor neglected WAR. That is literally all I did yesterday, I know it sucks, but what is a man to do.

Anyway, I need something to talk about today so I did a bit of forum perusing and I found this interesting post about Unji/Unsho NIN's versus Senjuinrikio/Perdu Blade. I found their conclusions to be woefully inadequate. Here is the post and the information that they provided:

Originally posted by LinktheDeme on Blue Gartr:
Before this is derailed I'm copying and pasting what I posted in advance section.
These are my parse results after about 170 mobs. Redline has U+U and I am using Senj+ Perdu. we parsed about the same exact total damage except he had higher ws total.

Higher WS total was my concern with senj + perdu to begin with because I thought that Senj + Perdus total ws dmg would outweight the delay loss. This parse proves that since U+U ws dmg came so close that Unji + Perdu wouldn't vary too much from U +U meaning higher total ws dmg faster tp results.

Also I personally consider my choice of WS gear trumping his which is why im concerned that his ws dmg is doing more total than mine.

Both 8/8 katana 5/5 str he's elf though and i'm hume

Anyways to sum it all up Unji+Perdu (Unji main hand) should be better than Senj + Perdu. (BUY UNJIS WHILE THEY DROP IN PRICE GOGOGO)

edit: Also i'm Carus not the war

Anyway, I have more than a few problems not just with the conclusion but the testing itself. I am going to lift some of the objections that were pointed out in the test from other people in this thread first then move on to the more obvious problems with the conclusion. The first objection that was made concerned the equipment that was used for the test. I must say that I am not a big fan of Ninja Chainmail for TP build. I understand that many people like the Ninja Chainmail for the 5% reducation in weapon delay, but it requires giving up a whole lot in return. I can't see it generally being more effective than a Haub, Haub +1, or even Koga Chainmail. Now, I use Koga Chainmail for my own personal reasons, as it is close to a Haub, without all that reduction in evasion. This is a personal safety preference, probably not a maximizing decision, but I am ok with that and will accept that criticism if issued, but the point here is that it was suggested that U/U can justify the loss from a Haub because of the already low delay will be boosted even more by the Ninja Chainmail, while a Senj user should favor a Haub. This was not the case in this test as LinktheDeme pointed out later the difference in TP build gear.

Fire bomblet
Walhara Head
Nin chainmail + 1
dusk gloves
Swift belt
Magnetic earing?(which ever gives acc+3)


Senj + Perdu
Walhara head
Nin chainmail
amemet +1
Dusk Gloves
Swift belt
Fire bomblet
Byakko's pants

Camp was nyzule isle mamools.
Again, after see this, others had issues, as did I. It is not that they had bad gear, not at all, just that some of the gear selections were a bit weird. I am not one to general criticize others gear, but I will definitely point out flaws when people attempt to make conclusions without setting the proper perimeters for a test. There were definite flaws in the testing that is for sure, but more importantly lets look at the conclusion.

The conclusion that U/P is better than S/P because U/U "won" in this test is questionable at best just by looking at the total damage output, which is a difference of .02%. .02% is something I would say that should be considered negligible. I don't know the margin of error for this test, or how you would even calculate the margin of error for this test. For that matter I don't even remember how to calculate a margin of error ^.^/ but the point is, I think its pretty safe to say that it is not a significant difference. So lets look a little farther into the parse and see what else is going on. The U/U user is definitely getting more damage in WS, likely from the higher number of WS being used, but remember the delay difference between these two set-ups is not that large any more with the Perdu Blade in play; U/U (201+201) = 402 vs. S/P (227+190) = 417, which is a grand total difference of 15 delay. Yeah, thats not that much at all, and to say that 15 delay makes up for a difference in WS damage is again rather questionable, this difference would definitely increase if the set up was U/P but again is 26 delay that much difference? Yes, it is beginning to be a bit more of a difference, and the gap would definitely widen in this case, but is the difference in weapons really the cause of the unbalanced WS damage or is there another factor lurking?

Lets check the accuracy differences, and we might see what could be the real culprit. U/U's accuracy was 88.86% vs. S/P's accuracy of 84.59%, and now we have something a bit more concrete when it comes to differences. Considering the accuracy differences here, it is very hard to draw any conclusion at all. In a perfect test, you would want equal amounts of accuracy for each player, and actually you would want the same exact gear too, but that is unlikely to happen. But a greater that 4% difference in accuracy is going to have an effect on the results, and that will be amplified in when considering WS damage because each hit has the potential to count for more damage, so each miss in a WS is going to have a larger effect on the results there.

Ultimately, what does LinktheDeme's test tell us? Not much. Considering the variations in damage, accuracy and equipment, this is definitely not conclusive. But more importantly, the thing to remember is that you shouldn't always take parsers, especially parsers your didn't do yourself, with a grain of salt. And one parse is never conclusive. This is the reason that scientists test other scientists theories rather than just accept them. That is the purpose of scientific theory, to reinforce and substantiate claims based on testing, not to draw conclusions after one test without considering all of the variables.

This whole discussion has a corollary that has to do with how people make gear selections and how they come up with the gear that they use. I have an objection to the current belief that the perfect set up is the standard set up, meaning making claims about what people should do and what gear they should use based upon a perfect party set up is not a good thing to do. We should aim to give advice to those in the middle of the road and then learn to make adjustments based upon the changing environment of each unique party. This was also pointed out by Rykoshen on the BG forums in another thread when after many different people were making claims that people should not be using STR on WS's, and instead be using accuracy because each hit counts. I found this odd for two reasons. One, the person making the claim was a MNK, and they definitely need the accuracy for their 8 hit WS, while it may not be as important for a NIN on their 4 hit WS (this presents another issue of universalizing the specific, but that is a discussion for another day), not to mention the modifiers for Asuran Fists is incredibly low compared to Blade: Jin. Two, the assumption was that people would be using meat. Now for a MNK this is an option, but for a NIN to maintain a good haste build there are many different sacrifices that need to be made including giving up some accuracy. To me, and many other people, this makes too large of a cut into our accuracy to get to the point where meat is an option. And this brings us back to the point about perfection being the standard for evaluating common practice. As Rykoshen pointed out, we all do not have the ability carry about two bards for every party, we can't always have a perfect party set up, and in those all the more common cases, the ability to support meat for NIN's is not assured. Thus, many people use Sushi so that they can maintain a high rate of accuracy within the standard party set up. In these cases, many NIN are over-accurate and thus can give up that accuracy on WS's for more STR for WS modifiers. What is good for the MNK is not always good for the NIN, and vice versa.

My conclusion from those parses? I will be looking forward to my Perdu Blade and I will be using S/P. But to each their own, do what you like, because ultimately that is all you really need.

Well, I thought that was a good post today. Hopefully, I will have some more excitement for tomorrow, and some more pictures to post from what we have done for the day, and finally congrats to Kira for getting 68 on BRD finally, and keeping for at least a day. :)

No comments: