Question for Freedom Vyn priests.

[Dairuka 888]

Handy tips for people doing this:

 

1: Exercise diligence. Failure to record data is devastating, and an issue persists that people have terrible memories and only start recording once they notice a bad streak - which can taint the data. Controls are also a must; if you're chaincasting, keep doing it. If you're wearing certain clothes, keep doing it. Keep that nutrition good and keep your village's enchanting bonus steady.

 

2: Statistical tests. If you're comparing two datasets on a pass/fail basis, this is a godsend; it'll tell you if your stats are due to chance (for instance, Thor's data - if accurate - is statistically significant. Dairuka's, however, is not).

Note: I've been privvy to a few tests like these and I'd advise against using data from anything other than 0.01 significance, lest we suffer the wrath of jelly beans). As well, the z-test is a little questionable on sample sizes below 100.

Welcome to statistics.

you're gonna have a bad time

 

I don't believe there is any difference between Fresh Casts and Recasts. A lack of a statistical difference correlates with my previous (But unscientific) data. This whole test stems from the current rumor of Dispel + Fresh Cast > Recasts.

 

As for how I'm doing my test:

 

There doesn't appear to be any way to accurately count for recasts if pass/fail is the only thing accounted for. (70+ is my Pass/Fail mark now) As a failure on a recast with a 59 Wind of Ages cast on it, could have just as easily been a -1 cast as it would've been a 58 cast. (Both are determined as frowns/failures by the game.) Every "Fresh Cast" is actually an item that has first been dispelled, and then Fresh Casted as per the rumor's requirements.

 

Average casts are also impossible to do, even using estimations, if you're using Recasts. (Which is why all my averaging in my earlier tests is done with Fresh Casts only.)

 

Also, I'm actually logging what type of 70+ cast is being made, because it'll also give me a reasonable idea of what my average 80+ per 1000, 90+ per 1000 and 100+ per 1000 is.

 

If the way I'm testing for a difference between Fresh Casts vs Recasts bothers anybody, and they have a better way of testing it, or have suggestions on how I can improve my current testing method - please share and I'll give it a shot.

 

If the sample size is the problem, that's being rectified as we speak, and all variables have been included. (Assuming that QL and Item type don't matter in regards to the end result/power of the enchantment. Especially since I count all Shatters/Damaged as failures, and only count the total number of 70+ casts. 70+ casts being an important cutoff point for most Enchanters/Merchants. My second batch of casts was purposely done as equal as possible, to give the best results. All future casts will be done the same way.)

 

Edited: Sample is now 100 for each side. Still shows no statistical advantage for either one (p>0.01 at 0.101). Proportionally, Fresh Casts are 0.14 and Recasts are 0.23. Interesting results, but RNG is still a major factor. More testing is needed.

Edited September 24, 2015 by Dairuka

[Raybarg 442]

Only point I have to bring up about measuring Fresh Cast vs Recast is this loosely observed difficulty factor. This is actually interesting me enough to remotely consider running a test on.

 

Theoretically Dairukas analysis should have an answer surface to this, or confirmation atleast.

 

It seems casting is more challenging if casted enchant already exist on item. This increases the number of "failures to cast". (not "failures to improve power"). There is not much failures to cast with my priest anymore, but one thing is for certain; when enchanting 1ql pickaxes with Coc there just is not failures/shatters on Fresh Casts, but relatively high probability to fail/shatter on Recasts.

 

As Stanlee pointed out, this could be due to my memory bias(es). Even though, I do pay attention to every Fresh Cast on low QL items. Once I even run a skiller pickaxe enchant by dispelling all "too low" powers just so I wouldnt lose "that many" items (due to making more felt like more effort than the dispel casts in between.)

 

 

So when measuring Fresh Cast vs Recast, "failure to cast" should not be collected in same group with "failures to improve power". Those casts should be excluded from dataset as the test is supposed to measure succesful casts and power result distribution. Or am I wrong about it?

[Stanlee 3,434]

"This whole test stems from the current rumor of Dispel + Fresh Cast > Recasts."

 

I found that one tough to believe when I first heard it. Not because it doesn't sound like it could be a thing, but because laziness seems to be the MO of that side of the coding; doing a check against "already enchanted" is a step of effort for not much effect on gameplay. If it does have an effect, I'd think it's from the overheating mechanic, rather than recasts itself. Though I've never really seen that mechanic in action.

 

Sure, there's a lot of examples within the game that are contrary to what I just said (gender-specific valeri items, anyone?), but that one just doesn't match the feel of other areas.

[Dairuka 888]

"This whole test stems from the current rumor of Dispel + Fresh Cast > Recasts."

 

I found that one tough to believe when I first heard it. Not because it doesn't sound like it could be a thing, but because laziness seems to be the MO of that side of the coding; doing a check against "already enchanted" is a step of effort for not much effect on gameplay. If it does have an effect, I'd think it's from the overheating mechanic, rather than recasts itself. Though I've never really seen that mechanic in action.

 

Sure, there's a lot of examples within the game that are contrary to what I just said (gender-specific valeri items, anyone?), but that one just doesn't match the feel of other areas.

 

My latest batch showed the RNG showing no favoritism towards either side - at all.

 

The third batch is quite possibly the most accurate to the method, rather than jumping from item to item, I stayed with an item until it shattered or received a 70+ cast. I'm not sure if it's enough to draw a conclusion on, but I'm getting a reasonable idea that there isn't going to be a statistical difference between the two given the parameters I set. If there is at this point, I'd be floored.

 

I'll keep doing it this way up until I get sick of doing this test. I'll most likely stop at either 250 or 500 casts each. I still have other simple comparison tests I want to do that I feel are more important than some silly rumor. Including Hypothesis and Postulates such as:

• Does a battery's soul depth/channeling matter when linked to the main? (500 casts with a 50 Channeling Battery, and 500 casts with a 0 Channeling Battery)

How much does maximum faith bonus actually add to average casting power? (500 casts with a 90 Channeling Priest/50 Channeling Battery in Nahjo Faith Zone versus 500 casts outside of Nahjo Faith Zone)

Exactly what is the difficulty of every Vegetable? (100 tendings for each crop type with a 90 Farming Priest, logging all stat/characteristics gains, and 100 tendings for each crop type with a 30 Farming Priest.)

What is the average QL of Cordage Rope with 60 Ropemaking and 90ql Wemp, versus average QL of Door Locks with 60 Locksmithing and 90ql Iron. (500 each.)

Edited September 26, 2015 by Dairuka

[Odynn 2,248]

It seems casting is more challenging if casted enchant already exist on item. This increases the number of "failures to cast". (not "failures to improve power"). There is not much failures to cast with my priest anymore, but one thing is for certain; when enchanting 1ql pickaxes with Coc there just is not failures/shatters on Fresh Casts, but relatively high probability to fail/shatter on Recasts.

 

 

seen that myself.

 

Another point worth noting is the fact you get skill ticks on channeling on casts bellow 40 power - you can link that to your frown to improve to have an idea of the casts powers. I'd still like to see the value added in the event tabs though. You frown to improve the power [0] - You succeed to improve the power [103].

 

Would help a lot to spot a trouble somewhere or not.

 

Something else worth noting is the trouble on creation skills at high level. Were you seems to have a lower success chance than someone with lesser skill. Some are confirmed by the % creation chance but others like channeling are totally in the dark with the lack of tooltip / feedback.

[Gaeron 551]

I'll keep doing it this way up until I get sick of doing this test. I'll most likely stop at either 250 or 500 casts each. I still have other simple comparison tests I want to do that I feel are more important than some silly rumor. Including Hypothesis and Postulates such as:

• What is the average QL of Cordage Rope with 60 Ropemaking and 90ql Wemp, versus average QL of Door Locks with 60 Locksmithing and 90ql Iron. (500 each.)

Good luck on your testing, I can offer some data for comparison, this is a recent run on cordage making though not at the level you specified

 

QL          P1    P2    P3    P4

00 - 10     6     8     13    7

10 - 20     10    4     13    10

20 - 30     14    13    16    14

30 - 40     14    20    13    16

40 - 50     18    18    17    15

50 - 60     16    15    9     14

60 - 70     11    8     14    16

70 - 80     8     6     3     6

80 - 90     5     6     2     3

90 - 100    0     4     1     1

total       102   102   102   102

Total actions 460, fails 52 crafts 408 with an 88% chance of success so that seems in the right area.

88.38 rope tool down to 88.30, RM 85.48 up to 85.52 using 99.25 wemp.

 

I only repaired the rope tool to roughly offset the small RM skill gain.

 

Do the 60 skill locksmithing run with 90ql lump and compare to that, you'll soon see that you are a winner....  

 

I'd like to see the comparison between your cordage run and mine just to see how much, or little, improvement the extra skill grind has net me.

[Gumbo 1,563]

My high skilled Vyn priest cast a 5 Nimbleness on a weapon, no big deal, low casts happen. The next TWO casts to improve it, both failed to do so. That would appear to be a problem. Getting 3 x 5 or under Nimbleness casts in a row on a good priests isn't right. That's rookie type casting.

[Dairuka 888]

For the fourth time out of 5 batches of 50, Recasts have had a slight advantage over Fresh Casts. That's 250 casts for each side, 500 in total.

 

Currently, proportionally we're looking at 0.188 (47/250 Recasts) vs 0.132 (33/250 Dispel + Fresh Casts)

 

I'm actually now very curious. I want to see if this trend will continue, so I'm going to continue these tests for for another 5 batches.

 

My high skilled Vyn priest cast a 5 Nimbleness on a weapon, no big deal, low casts happen. The next TWO casts to improve it, both failed to do so. That would appear to be a problem. Getting 3 x 5 or under Nimbleness casts in a row on a good priests isn't right. That's rookie type casting.

Edited September 28, 2015 by Dairuka

[Heboric 808]

 

I'm actually now very curious. I want to see if this trend will continue, so I'm going to continue these tests for for another 5 batches.

 

 

I think the word, "improve" actually skews expectations in this case. If you made a 40 Nimbleness Cast, and your next two casts are 33 (sic) and 38 (sic) respectively - you didn't really fail to improve it, you just had three low casts in a row, which is well within the RNG's parameters. (My best case scenario in these tests is on the recast side with 47/250 casts of 70 or higher. I've had streaks of 12 where I didn't get a single good cast.)

[Dairuka 888]

Which might suggest that casts on "virgin" items (maybe even not dispelled ones), might be the way to go.

Edited September 28, 2015 by Dairuka