Which Fairy IS better?

[Teykos]

*Sigh*

@cainejw: If you include the additional information that you previously omitted, all your information is telling me is that you are much more comfortable playing Eos over Selene. You want to run tests. I doubt anyone has a problem with that. But if you're going to get defensive when you've left out information that can totally influence the conclusion drawn... yeah, I don't know what to tell you there.

[Rainsford]

*Sigh*

I think cainejw's issue is that he has a fundamental misunderstanding of how the Fairies and their buffs work. He previously did a "test" on Selene v Eos's Embrace and to see if there was a potency difference between Place and Heel. Of course, the difference between the two is casting frequency, not potency.

For what it's worth, having a trial with 2x the HPS of the others seems like an odd thing to keep in.

At the end of the day, not even using Fey Light seems to illustrate the misunderstanding of the discussion. Eos's skills, with their long CD's and relative weakness, don't have any major impact on a fight. While Selene's buffs aren't super effectual either, any increase in DPS that gives you an HPS increase on a lower CD is superior to Eos's buffs.

[cainejw]

*Sigh*

No information was omitted purposefully. Any defensiveness is in reaction to a post which seemed to point to the logs themselves being an error instead of elements in those logs being an error. You've also assumed my approach. I have no partiality to Eos or Selene. That should be evident in the fact that I have used the logs to conclude that initial assessment shows both faeries are useful in different ways (Eos for heals, Selene for DPS).



Eos seems to give a larger benefit to healers than Selene to healers. Selene gives a larger benefit to DPS than Selene to DPS.

I think cainejw's issue is that he has a fundamental misunderstanding of how the Fairies and their buffs work. He previously did a "test" on Selene v Eos's Embrace and to see if there was a potency difference between Place and Heel. Of course, the difference between the two is casting frequency, not potency.

For what it's worth, having a trial with 2x the HPS of the others seems like an odd thing to keep in.

At the end of the day, not even using Fey Light seems to illustrate the misunderstanding of the discussion. Eos's skills, with their long CD's and relative weakness, don't have any major impact on a fight. While Selene's buffs aren't super effectual either, any increase in DPS that gives you an HPS increase on a lower CD is superior to Eos's buffs.

Oh, look, it's you again. I identified why I worked my preliminary examination, identified as such, in the way I did. No logger was known, at that time, to log the effects of HOTs on Eos. As such, the only possible comparison was Embrace. That has since changed. You've ignored that multiple times. Why? Because you want to state, without exclusion, that Selene is the best in every single situation.

You are incorrect now once again, Rainsford. Selene does not benefit spell casters in a statistically significant way over Eos as outcomes are mixed. I can actually give you the spreadsheets to back that up. Or, if you'd like:

t-tests per DPS, independent single-sample t-test (testing each sample, the DPS, to a mean) with significance .05 and a t-crit of +/- 1.753. Columns 1 and 3 are Selene DPS, 2 and 4 are Eos DPS.

1.285714286 3.489795918 1.857142857 -3.612244898
4.06122449 8.387755102 19.97959184 2.102040816
-10.79591837 -10.2244898 -5.979591837 -7.775510204
0.795918367 2.183673469 -0.510204082 -5.163265306


4.994641549 -1.912411187 1.201982507 2.417712387
5.789513344 -3.616215671 3.738036929 4.460237304
-4.279822083 -8.061869489 -3.723705863 -1.759031799
4.309398221 -2.982705971 -0.834428247 0.261281443


Notice that both Eos and Selene returns statistically significant results meaning that both DPS are beyond the mean values for this group.

Same for healers:

-1.384615385 -2.540659341 -2.529230769 4.189010989
0.957066144 0.662857143 -1.988571429 4.379340659

t-crit is +/- 2.3 showing again mixed results.

Both showing mixed results means that the variance is likely accounted for elsewhere than the faerie itself. I'm still operationalizing correlation so I can then see how much variance is accounted for by the faerie, but that's a dichotomous variable, so I'll have to work that out at a later date. This is how this stuff is supposed to work when you attempt to hypothesize that one faerie is superior to another.

Conclusion of statistical test: There's no clear outcome. The use of faeries is mixed. This means that no one faerie is universally superior to another.

EDIT: Adding more in, comparing a Selene run from tonight with Eos/Selene above, using both of Selene's buffs. Comparing to Selene A, the DPS is about the same:

Selene A:

BLM - 267.97
MNK - 277.71
BRD - 154.36
BLM - 259.58

Selene C:

BLM - 287.93
MNK - 253.80
BRD - 189.51
BLM - 287.47

Eos B:

BLM - 236.40
MNK - 261.42
BRD - 185.24
BLM - 209.33

It would appear as if the group that loses the largest from Selene's absence is Black Mage. Monks may lose DPS due to Selene, but this could again be a confound. Bard appears to be relatively unfazed. Keep in mind that this is with both buffs in rotation.

Of course, more testing will be necessary. However, both healers lost 100+ HPS. This may be due to any number of things including faerie. Again, hard to isolate. The data seems to continue to show that the DPS benefit from Selene but healing loses quite a bit of healing from Eos.

The real issue here is the log reliability. Frankly, I'm not convinced they're reliable.

[Rainsford]

*snip*

But their Embraces have nothing to do with it.

Again, the discussion is focused on the buffs, Fey Illumination and Whispering Dawn vs. Fey Light and Fey Glow.

If you DPSed straight on a dummy, the class that most benefits from Spell Speed is BLM and the DPS increase from Fey Glow, again, is 6%. I'd assume that the increase in healing is the same if you spammed heals for the duration, but I don't know why you'd do that.

This means that, at most, Selene gives a 3% increase in overall DPS. You'll never be able to reliably test that in a real-world scenario with a parser, especially since some of the benefit comes from stuff that's non-measurable (being able to avoid mechanics easier, etc).

So again, I feel like you're just wasting your time. Eos's buffs are focused around healing, but considering that the content is easily healable with two WHM's or Selene, Eos doesn't serve a function. Even if Selene's average DPS and HPS increase was 1%, it's superior to Eos. From a mechanics perspective, Eos doesn't bring enough to the table to be more attractive compared to Selene.

[cainejw]

I'm not wasting my time. I enjoyed logging and examining to find there's really no answer about which faerie is superior to another conclusively. It allowed me to examine your claim and debunk it based on statistics. It allowed me to gain a better knowledge of the faeries and how parsing works. I actually enjoy gaining knowledge about Scholar, and it would appear we have a small amount of Scholars actually investigating before opining.

I've no doubt that I haven't convinced you. You're clearly someone who gets an opinion and refuses, in any circumstance, to change that opinion. I could present the perfect logs to you to show no real significant outcome, and you'd just bring up the very first, very limited test I did. You'd ignore that I stipulated that it was limited. You'd ignore every single disclaimer and declare yourself right because you think nobody can prove you wrong. In the course of all of that, you've forgotten that you're the one who presented the extraordinary claim. You're the one who has to back up that claim. You've yet to do so in a convincing fashion or, well, any fashion other than opinion. You're presenting yourself as an expert for Scholars, and your backing of that is your opinion. If you don't see the problem there, I don't know if there's even a point replying to someone who is so set in their opinion.

You claimed that Selene is better uniformly at everything. You were wrong. Selene appears better for Black Mages, likely Bards, and possibly Dragoons, and I've not seen the Monk benefit from her. Eos appears to help healers the most, may help Monk maintain combo rotation, and could present a "safety blanket" allowing DPS to feel safer--but that last part is my personal opinion. It should be taken with a grain of salt.

Finally, just to add, I ran just selene...it took 7 minutes on Coil 4. We took 3 hits of enrage. With Eos, we took 2 hits of enrage. Against a set time limit, it appears once again that the faeries had little effect on the fight in this near-end-game scenario. On smaller pulls and dungeons, I haven't tested. That may give me a fun thing to do when my FC goes to do cap runs.

[Trystan]

Guys... too many words, dont think so complicated. I would like to make this discussion simple:

Selene does "buff" the damage
Eos does "buff" healing

If you can win with normal healing you are better off with selene.

If you cannot win without buffed healing, you are better off with eos.

Since most normal content (including AK/WP, Ifriit/Garuda HM) can be done without a buffed healing (unless you are under-geared or the group is getting unproportional damage due to lack of skill or lack of gear or just bad luck) - you are mostly better off with Selene in "normal" content on level 50.

(I have no exp in BC yet, so if you only talk about endgame-raid there: just ignore me)

[Rainsford]

I'm not wasting my time. .

You're not giving any meaningful statistics and you never will. The fact that you're able to clear the instance with no issue regardless of Fairy used shows that Eos isn't serving a function. You keep demanding statistical proof of two things that are unrelated. It's not like you're discussing rotation where you could hit a dummy for 3 minutes and see which produces higher DPS, because Selene always will. HPS parsering is inherently useless since HPS that disregards things like overheals doesn't help the discussion.

So, again, the point you keep ignoring: Eos has buffs that enhance healing, Selene has buffs that enhance healing and DPS. However marginal the increase with Selene, why would you ignore it when the buffs Eos provides don't have any impact on the fight?

You also seem to be giving a lot of anecdotal evidence for someone whose trying to be "data driven". Ignoring the fact that you can only run Turn 4 once a week (and yet you have data on at least 5 clears?), saying "we got less hits of enrage with Eos therefore Selene's DPS increase is nonexistent" is silly on it's face and, again, shows a basic misunderstanding of the games mechanics. I don't bring up the earlier "trials" for any reason other than to show this basic detachment from how the game functions.

A 150+ increase in Spell Speed increases DPS, period. Attacking faster means you get in more attacks. Will this mean that your damage from one encounter to another will always remain static? Of course not. A BLM with a good rotation and a Relic staff will out dps a BLM with a bad rotation and a Relic + 1. Does that mean that we claim a Relic + 1 has no impact on DPS? If you don't need Eos for healing (which you never will in any endgame content), then you are always doing your party a disservice by not having Selene out. Always.

*snip*

Selene does buff healing. Fey Glow increases the healers and fairys cast time, meaning you get more HPS.

Eos's buffs to healing are weak at best. Whispering Dawn is only useful if the majority of your party is <80%, do not need to be quickly topped off, and you are unable to cast Succor. In most cases, almost the entirety of the heal will go into overheal. Fey Illumination, with it's low uptime, means you get an extra 1500-1600~ healing if you cast Physick non-stop for 20 seconds.

Like you pointed out, if you don't need extra healing, Eos loses her utility. In Coil, her buffs and their long CD's don't serve any type of game-changing roll, so not opting for the increased DPS of Selene is foolish.

Finally, @cainejw, you need to think on what you're arguing here. You seem to be making a big point out of "I am data driven" while seeming like a stick in the mud as far as the choice goes. Can you clear Turn 4 with Selene? Awesome, why would you ever use Eos? Do you still have a fundamental misunderstanding of how the mechanics work for healers?

[Teykos]

@Rainsford: He clarified that all but 1 of the Eos runs was a failed attempt.

[IceSpear]

You claimed that Selene is better uniformly at everything. You were wrong. Selene appears better for Black Mages, likely Bards, and possibly Dragoons, and I've not seen the Monk benefit from her.

Monks are different to the other classes with the fact that their rotation is significantly faster than other classes when they have the Greased Lighting buff so, because they're using their moves faster their TP doesn't regenerate fast enough for them to keep the rotation going for long as they'll eventually need to slow down their DPS or they'll need to stop all together if that makes sense.

[YuriRamona]

data

Please review my previous post that crunches some of these numbers. I took extreme liberties to get nice, round numbers so it doesn't stand up to exact scrutiny. But I think the principle applies.

I've not seen any sound argument here that refutes my analysis, and so I'm of the opinion that data is not necessary in this evaluation, when the theory provides a sufficient means. Clearly, you disagree.

If you mean to suggest that your handful of parses are relevant, then... I don't know what to say.

I'm not a statistician but even if you were to log and record every single encounter you engaged in from this point forward, you wouldn't have a very reliable sample.

If you do think that is is possible to settle this debate by parsing a larger sample, here are the problems. There are many reasons why parsing full encounters produces many intervening factors:
>> players will vary and improve their rotation bit by bit every single time (this is most noticeable for paladins)
>> players can be distracted or less alert on any given day
>> delay and lag in this game are quite variable even for the same player on the same day and have a tremendous impact on how comfortable they feel (and therefore their performance) or purely, on their ability to send commands to the server
>> players' equipment might change, or you might sub an entirely different person

Hopefully, this makes sense, at least conceptually. Parsers are only useful when you have nearly identical data sets. For example, a single DPS can figure out kinks in their rotation, or a healer can calculate the formula for their spells. Even if all 8 players have pinpointed their performance to a near-mechanical level of precision, there is simply too much variation built into us (humans), the connection (which sucks), and the game (whose randomness generator operates beyond the ranges at which we are presently discussing). Using a parser in this kind of fashion is risky, and at best, only qualitatively useful.

Maybe this addresses what you're really trying to get at, which is, does any of this theorycrafting matter in light of real-world conditions? Does it really apply or have noticeable, reliable effects? I think the answer to this is no.

The effects of fairies is at least an order of magnitude below the effects of natural randomness.

We would be lying to ourselves if we said that we felt the effects of fairy buffs, or were even able to see it in our parses (unless we had some ridiculously large sample). Per my linked post, the effectiveness of Selene's speed is, at best, a little over a second in reduction to GCD every 30 seconds. Eos' buffs are on cooldown for so long that we have to adapt for mechanics without them anyway. We're dealing with boosts of 2-4% when our effectiveness as Scholars varies by much more from instance to instance.

**Of note in the present discussion is that my analysis falls apart if you assume two things: Whispering Dawn and Fey Covenant will rarely be used to maximum effect every single time it is activated, while Fey Glow and Light will almost always be used to their maximum effect. The MP savings of Eos is negligible in a majority of instances. Selene almost always wins.