Dev Tracker - Discussion

It's just a kupon for naegling

I enjoyed the second Vana'bout so I'll definitely be doing it daily. But how could the carrot compare with Cornelia? I'd be worried that the tiers are going to be harder to reach with a worse reward.

Nothing compares to Cornelia. We'll appropriately hold a candlelight vigil when she is removed from the trust menu.

To be fair, Cornelia is great don't get me wrong... but it wasn't anything spectacular from a reward perspective.

The times we received Cornelia for a period of time in the past was for a period of 6 months or so. SE simply implemented that as a reward instead of just giving it to us as they had years past lol.

I am single character player and use trusts all the time and love having Cornelia as an option. But don't be blinded by a reward that was technically handed out for nothing prior.

The appropriate reward would have been a cipher scroll for her so that she was a permanent trust if as an individual you earned "X" Plaudits. Redeemable for plaudits significantly higher than the other tiers that would have required more than just casual/minimum effort.

A RoV style KI for stuff like Omen and newer could certainly have potential.

We also failed the first VanaBout because the rewards were unworthy. The moment it was Cornelia, it was an immediate win. Players don't necessarily require rewards to be "new" per se, so long as they are good. They gave us a good reward, so their planned event didn't fall flat on it's face twice in a row.

If they give us cipher for permanent Cornelia use, they can't dangle it in our faces ever again, and for SE, that's very important for them as they can continue to wheel her out and promote whatever campaign or participation.

I happen to go into a mild depression/break from game whenever she's unavailable. I'd love to have her permanently.

Had nothing to do with cornelia. The first vanabout required daily full clears just to hit max tier and the goals were not party friendly. I think you were given one grace day, two missed days and you were out. I dont believe the JPs were keeping pace on day one, and they were able to double dip.

It might be 2025 soon, but it's still FFXI.

The most welcomed would be KI that push cooldown for Imprimaturs restoration to like 1h.

Daily VanaBout 1 goals that gave 10 plaudits each:
Clear ambu 3 times
Heal 300+ HP
Deal 500+ ice dmg
Deal 500+ ltng dmg
Deal 500+ wind damage
Do 10 skillchains
Clear ambu 3x without dying
Do 20 weapon skills
Do 10 MB's
Clear Normal 3x

Those magic damage and WS/SC/MB goals were not party friendly. The mob barely had enough HP for ONE person to hit those goals within 3 fights. The one-time objective required 20 clears under 5 min, which meant an extra run where you werent trying to hit those above goals. That required doing Ambuscade FOUR OR FIVE times a day, EVERY DAY. This tied up the limited instance spots and congestion for the event happened.

Feel free to correct me because I want to be sure. This Old Case change is basically making it so I might actually see a DRK +1 in my lifetime? That would be insane. It's just about the job distribution, right?

They're adjusting the basis of the rand{} formula they use server side to determine what job you get from the case.

Which begs the question why there is any form of job weighting on the formula to begin with. Like, what does this mean?
"When we tested our random number generator 100 times, and we got an acceptable distribution of earrings, but when we ran it 10 million times, we noticed there were more earrings of base jobs than expected"

Is that what the statement means? Did they literally suggest they originally used a small sample size to determine the output when the ability to generate a meaningful sample size was well within their control to do so?

I think it is just even jobs distribution yes, BUT I wouldn't put it past them to have also messed up the HQ rate as well and have carefully worded their response to avoid taking ownership. Shinryu and +2 earrings really were examples of step changes in the drop rates in FFXI history - talking several RL decades for a perf aug'd ear +2 of choice and >1000 kills with th9 on VD for drop of choice on Shin (only other example is kclub but that's not content). I'm convinced they *** up and don't want to back track.

It sounds to me like they were saying they tested their RNG function on 1-1,000,000 but then the numbers actually used to generate the job are larger than that and the RNG isn't as random at that high of numbers.

Probably some rounding issue which doesn't cause a problem with small numbers, but once you get to larger numbers it can cause more numbers to end up on one side of your distribution instead of the other.

Seems like it was a failure in testing. They didn't have a large enough scope of testing or didn't run enough tests, something like that.


There wasn't supposed to be job weighting. The RNG was supposed to look like this: Make random number from 1-1,000,000,000.

Case 1-45,454,545: WHM earring
Case 45,454,546-90,909,091: WAR earring
...
etc.

Problem is that (for some reason) the RNG doesn't properly generate a random number from 1-1b, it accidently favors the lower numbers. This is the bug they're intending to fix.

Or... hear me out, they tested nothing. Adjusted nothing and people will be like "Oh ***, I got a summoner earring, this anecdotal evidence is all the proof I need! sortie is fixed!"

The easiest fix would be an exchange rate like Empy boots back in the day to get the ones you want, or even how they did with Omen cards. Which has been suggested since day 2 of the event existing

I also still have never seen a drk+1 earring. It's the only job I haven't gotten at least a +1 on.

Yep. I've had a few RDM +1, and I seem to get a MNK earring of some sort almost every run, but never DRK +1 and super rarely the NQ.

There are 22 jobs, why would you need such a large number to work with?

And before you say "To determine NQ or HQ earring from the Old Case"
Roll 1: from 1-20, 20 is HQ anything else is NQ (est 5% rate)
Roll 2: from 1-22: each number represents a job.

It takes 3 seconds to crack open a case, theres no reason it cant run these two rolls on the backend in that time.


Regardless, this is why I believe their quote of random number generator and small/large numbers alludes to sample size. If you open 100 cases, you're likely to wind up with 4 or 5 of each job. If you wind up with one 3 and one 6, thats just luck of the draw, you're not gonna ask questions.

short version is that computers cannot truly generate random numbers, the algorithms used can provide roughly equal distribution within a preset range and effectively any implementation is going to start with a 'large number' and narrow it down to the requested range using one method or another

SE loves using their own code, and they wrote ffxi at a time when there were considerably less RNG implementations available, so it is not completely pants-on-head-stupid or unexpected for this to happen

[why is there a word filter on r*tarded..?]

You can't just "roll a random number 1-22". You have to take some seed value (which is probably preposterously large to be sufficiently random) and then TURN THAT into a random result from 1-22.

Smaller scale, but this is an interesting video on the topic.

I'm a little confused with their RNG code. Is this something that's unique to opening old cases in the way they programed that specifically, or do they probably call their RNG function that's used for everything in the entire game? If their RNG results are not as accurate with number ranges larger in size, that would mean rarer item drops might be effected? Such as needing to roll a 1 out of 10,000(?) for k club, the 1-10,000 values weren't as equally distributed?

I know how RNG's work, but its irrelevant to what SE wrote. You're the one who came up with the 1-1billion.

To me this means that either:
A-They tested their RNG with a small sample size, and the expected results were within the acceptable deviation. If they can run a RNG test using "small values", implying generating a digit between 1-1000, why extend to "large multi million values"? Thats why I believe what you posted about 1-1billion isnt relevant here and you're misunderstanding what they mean by "small values" and "large values". This is a translation issue and "values" refers to "sample size".
B-Whatever formula they use to assign values (JOB and NQ/HQ) to the randomly generated digit was flawed to begin with. This is likely, but I'm struggling to find out how it could break down with significant distribution gaps of:
the 6 base jobs
15 of the 16 remaining jobs
drk

Maybe the dataset of drk was just bad luck.


Then again, the JP translation isnt helping anything:
「古びた小箱」を使用した際の抽選結果に偏りがあるのではないか、と公式フォーラムにお寄いただいた件へのご報告です。
お寄せいただいた内容を踏まえてテストルーチンを組み、数百万回の試行検証を行った結果、ご指摘いただいた通り1％程度が差分として現れました。これまで用いていた乱数アルゴリズムは、小さな値に対してはかなり平均的ではあったのですが、大きな値を乱数として発生させる場合に若干の偏りが顕在化する特性があった模様です。乱数アルゴリズムの調整を行うことで、計算結果の分布差が極小になるようにいたします。フィードバックをお寄せいただき、ありがとうございました。
This is what google is spitting out
I still dont understand why they would need to all of a sudden alter their RNG for "large numbers".

Presuming a 5% rate of +1 earring from an old case, you'd only need 440 random digits.
one job NQ/HQ = 20 * 22 jobs = 440

Even a 1% HQ rate would only need 2200 digits for all 22 jobs.

And before you talk about how RNG generates a large number, I know.
They are the one who stated they altered their Random Number Generator from generating "small numbers" (who knows how many digits this is) to "large numbers" (once again, who knows how many digits).

I didn't really understand the terms they used either. I suspect they used google translate or something for the Japanese-version of stats language lol.

I see two outcomes:
1. More +1/+2 earrings from NQ boxes.
2. Even worse drop rate for +1/+2 earring from NQ boxes.

Day of the update, all boxes will give full aug +2 100%

Don't miss your shot at the 2 hours its "broken" before emaint

I doubt they will dish out an even worse drop rate. It is either 1) the job distribution is not equal, which was already tested in this thread, or they feel we weren't getting a high enough rate of +1/+2 from NQ and +1 boxes. Either way, it is a win that they noticed and are updating... but it's not groundbreaking. It will still be a painful drop rate.

I wish SE would just make the Sortie RoE's monthly. Is there anything so bad about 1 +2 case every month?

I never hold out hope for SE. It's just been 20 years of underwhelming and back-biting of their customers.

If they give you the carrot you stop chasing the carrot.

Yes, that's bad. Until there's a new carrot to replace it.

Just a reminder that Eiryl doesnt do Sortie because theres no gil to be made in Sortie, and as usual he has no clue what he's talking about.

Giving one +2 case every month will not cause people to stop "chasing the carrot". There is a ~75% chance (22 jobs, realistically most people play about 5 or 6 on a regular basis) it will give you a job you dont use, and another 70% chance it will be a low roll. Such a move would be more likely to encourage people to come back as the incredibly elusive +2 earring becomes a lot less elusive.

You only need look at the amount of idiots clamoring for +2 earrings to know how bad an idea it would be to start passing them out monthly.

You don't give away your only leverage. Common sense.

I'm just saying vana'bout, month XI, or Christmas is a great time to drop a low effort +2 case and shut us all up.