Going a little deeper into logs and we are seeing something amiss. I'll update when I have more information.

## Rune of Re-Origination bugged

### (Locked)

I attacked a training dummy for 6 hours. Here is the log:

http://www.worldoflogs.com/reports/rt-knmw3x5inkgxqwgt/xe/?x=sourcename%3D%22Stenhaldi%22+and+spell%3D%22Re-Origination%22+and+fulltype%3DSPELL_AURA_APPLIED

I had 15.02% haste for that test. I have the 522 Rune of Re-Origination, which has a nominal RPPM proc rate of 0.92 [base value] * 1.15^((528-522)/15) [scaling for item level] * 1.1 [10% buff a month ago] = 0.957 . So the haste-scaled proc rate should be 1.101 procs/minute.

The mean proc interval in the above log was 64.95 seconds with standard deviation (of the mean) 2.59 seconds.

Without the bad luck streak prevention mechanic, the mean proc interval should be 22 seconds [icd] - 10 seconds [10 seconds of the icd count toward the proc chance of the first attack thereafter] + 60 sec/min * 1 proc/ (1.101 procs/minute) [the haste-scaled proc rate above] = 66.51 seconds.

With the bad luck streak prevention mechanic, the mean proc interval should be scaled by about 88.44%. This is a result of my math at <http://us.battle.net/wow/en/forum/topic/8197741003?page=4#79>; I and others have also verified by simulations that it is approximately correct. This means the mean proc interval would be 66.51 seconds * 0.8844 = 58.82 seconds.

In summary, the mean proc interval -

Measured: 64.95±2.59 seconds

Prediction assuming no luck streak prevention: 66.51 seconds (matches the measured value)

Prediction assuming luck streak prevention: 58.82 seconds (discrepant with the measured value by 2.4 standard deviations)

Without luck streak prevention, the cumulative distribution for proc interval t is

exp(-k*t)

with k = 1/AverageProcInterval = 1/(66.51 sec), i.e. an exponential distribution.

However, with luck streak prevention, the cumulative distribution function looks like

exp(-k*t*(1+(k*t/2-1.5)*3)-3.375)

for k*t > 1.5 .

The largest proc interval in that sample was t = 312 seconds. Assuming no luck streak prevention, that has associated probability exp(-312/66.51) = 0.00918. That's a one-in-109 chance -- not particularly unlikely.

However, with luck streak prevention, the associated probability changes to exp(-312/66.51*(1+(312/66.51/2-1.5)*3)-3.375) = 2.13*10^-9. That's one in 468 million -- far more unlikely.

I made a quick histogram of the the above data. It resides here:

http://i.imgur.com/w5darGD.png

(Error bars are approximate)

Superposed on the data are two prediction lines: one (blue) without luck streak prevention and the other (red) with that mechanic. These are just the probability density functions

f(t) = k*exp(-k*t)

for no luck streak prevention and

g(t) = k*exp(-k*t) , if k*t <= 1.5

g(t) = k*(1+(k*t-1.5)*3)*exp(-k*t*(1+(k*t/2-1.5)*3)-3.375) , if k*t > 1.5

with luck streak prevention.

Without doing any further numerical analysis, it's pretty evident that that the data is following the standard exponential distribution of the "no luck streak prevention" prediction line, rather than the distribution with luck streak prevention.

http://www.worldoflogs.com/reports/rt-knmw3x5inkgxqwgt/xe/?x=sourcename%3D%22Stenhaldi%22+and+spell%3D%22Re-Origination%22+and+fulltype%3DSPELL_AURA_APPLIED

I had 15.02% haste for that test. I have the 522 Rune of Re-Origination, which has a nominal RPPM proc rate of 0.92 [base value] * 1.15^((528-522)/15) [scaling for item level] * 1.1 [10% buff a month ago] = 0.957 . So the haste-scaled proc rate should be 1.101 procs/minute.

**First check: mean proc interval.**The mean proc interval in the above log was 64.95 seconds with standard deviation (of the mean) 2.59 seconds.

Without the bad luck streak prevention mechanic, the mean proc interval should be 22 seconds [icd] - 10 seconds [10 seconds of the icd count toward the proc chance of the first attack thereafter] + 60 sec/min * 1 proc/ (1.101 procs/minute) [the haste-scaled proc rate above] = 66.51 seconds.

With the bad luck streak prevention mechanic, the mean proc interval should be scaled by about 88.44%. This is a result of my math at <http://us.battle.net/wow/en/forum/topic/8197741003?page=4#79>; I and others have also verified by simulations that it is approximately correct. This means the mean proc interval would be 66.51 seconds * 0.8844 = 58.82 seconds.

In summary, the mean proc interval -

Measured: 64.95±2.59 seconds

Prediction assuming no luck streak prevention: 66.51 seconds (matches the measured value)

Prediction assuming luck streak prevention: 58.82 seconds (discrepant with the measured value by 2.4 standard deviations)

**Second check: largest interval**Without luck streak prevention, the cumulative distribution for proc interval t is

exp(-k*t)

with k = 1/AverageProcInterval = 1/(66.51 sec), i.e. an exponential distribution.

However, with luck streak prevention, the cumulative distribution function looks like

exp(-k*t*(1+(k*t/2-1.5)*3)-3.375)

for k*t > 1.5 .

The largest proc interval in that sample was t = 312 seconds. Assuming no luck streak prevention, that has associated probability exp(-312/66.51) = 0.00918. That's a one-in-109 chance -- not particularly unlikely.

However, with luck streak prevention, the associated probability changes to exp(-312/66.51*(1+(312/66.51/2-1.5)*3)-3.375) = 2.13*10^-9. That's one in 468 million -- far more unlikely.

**Final check: the distribution of proc intervals**I made a quick histogram of the the above data. It resides here:

http://i.imgur.com/w5darGD.png

(Error bars are approximate)

Superposed on the data are two prediction lines: one (blue) without luck streak prevention and the other (red) with that mechanic. These are just the probability density functions

f(t) = k*exp(-k*t)

for no luck streak prevention and

g(t) = k*exp(-k*t) , if k*t <= 1.5

g(t) = k*(1+(k*t-1.5)*3)*exp(-k*t*(1+(k*t/2-1.5)*3)-3.375) , if k*t > 1.5

with luck streak prevention.

Without doing any further numerical analysis, it's pretty evident that that the data is following the standard exponential distribution of the "no luck streak prevention" prediction line, rather than the distribution with luck streak prevention.

#44

4/17/2013

Quality Assurance

We're currently working on a new fix for the Rune of Re-Origination's unlucky streaks.

Thanks!

Not fun getting my first rune proc at the end of my second superliminal proc of the fight :(

Not fun getting my first rune proc at the end of my second superliminal proc of the fight :(

Edited by Dysheki on 4/17/2013 6:53 PM PDT

#47

4/17/2013

good to know that something is broke and a fix is in the works. i won it last night and after testing it i pretty much just put it back in my bag. got like 2 procs in 15 mins.... was pretty disappointing.

#53

4/18/2013

Greg Street @Ghostcrawler 4h

@Tinderhoof This is just a bug. We fixed it but it broke again. It will be fixed again next restarts.

https://twitter.com/Ghostcrawler/status/325031111816261632

#54

4/18/2013

Please restart the servers before Tuesday. I don't want to have to regem and reforge just so I can use Renataki's for a few days of progression.

#56

4/18/2013

Bad juju also appears to be bugged and not procing enough. Not proccing on pull very often (even when out of combat for awhile), and procced less on every single fight this week compared to the one before. Proced twice during a 5:30 Dumuru kill and 3 times during a 7:33 Consorts. Really bad luck or is anyone else having this problem?

Edited by Moncoko on 4/18/2013 11:27 PM PDT

#58

4/18/2013

Bad juju also appears to be bugged and not procing enough. Not proccing on pull very often (even when out of combat for awhile), and procced less on every single fight this week compared to the one before. Proced twice during a 5:30 Dumuru kill and 3 times during a 7:33 Consorts. Really bad luck or is anyone else having this problem?

Yeah. I'm seeing the same problem as well.

I have a couple of the Agi trinkets and I don't have any problems with their proc rates except for Bad Juju and Rune of Re-Origination.

#59

4/19/2013

### Please report any Code of Conduct violations, including:

Threats of violence. **We take these seriously and will alert the proper authorities.**

Posts containing personal information about other players. **This includes physical addresses, e-mail addresses, phone numbers, and inappropriate photos and/or videos.**

Harassing or discriminatory language. **This will not be tolerated.**