Please upgrade your browser for the best possible experience.

Chrome Firefox Internet Explorer
×

Reverse Engineering is not 20%

STAR WARS: The Old Republic > English > Crew Skills
Reverse Engineering is not 20%

Grayseven's Avatar


Grayseven
03.28.2014 , 10:14 PM | #151
Quote: Originally Posted by Hoshkar View Post
Far as I can tell it is just a 20% of something happening. Not a 20% chance of happening, there for after 5 times it should = 100% sort of thing. Each time it is just a straight up 20% chance.


That and RNG hates some people more then others
It doesn't work that way. It is a flat 20% chance. After 5 times, you still have a flat 20% chance regardless of what happened the previous 4 times. It doesn't build up credit each time you fail and increase your probability of a crit until you get one.

The dice have no memory. Each roll is a new probability and sometimes Fate decrees that you are on the wrong end of the bell curve.
Sometimes you narfle the Garthok, sometimes the Garthok narfles you.

Khevar's Avatar


Khevar
03.29.2014 , 12:19 AM | #152
Quote: Originally Posted by Grayseven View Post
It doesn't work that way. It is a flat 20% chance. After 5 times, you still have a flat 20% chance regardless of what happened the previous 4 times. It doesn't build up credit each time you fail and increase your probability of a crit until you get one.

The dice have no memory. Each roll is a new probability and sometimes Fate decrees that you are on the wrong end of the bell curve.
Isn't that what Hoshkar just said?

Grayseven's Avatar


Grayseven
03.29.2014 , 05:40 AM | #153
Quote: Originally Posted by Khevar View Post
Isn't that what Hoshkar just said?
I have no idea. If it is, great, maybe people will finally understand how the system really works instead of working themselves into a furor because they think it should or does work another way.

Edit: Ah, okay, I see it now. I misread the post and thought that was how HE thought it worked. Instead it appears he was giving the example of how it doesn't work.

Still, reiterating the fact that it doesn't work that way in a myriad of ways may finally get people to understand how it really works.
Sometimes you narfle the Garthok, sometimes the Garthok narfles you.

HelinCarnate's Avatar


HelinCarnate
03.29.2014 , 06:48 AM | #154
My REs since last night.

201 attempts on all green items, 56 RE success
Quellryloth earlier stated he ran 125 with 26 success
The op had 402 with 54 success

Total for all is 728 attempts and 136 being successful for a total average of 18.68%

But no it couldn't be that the OP used a sample size that was too small..... /sarcasm
Dear BW,
Please nerf paper and buff rock. Scissors is fine as is.
Signed, Rock.

HelinCarnate's Avatar


HelinCarnate
04.03.2014 , 02:10 AM | #155
Ok over the past few days I also did 402 REs just like the OP. They were all greens and I was learning them as I was leveling up my crafting so I was not 450 doing an RE on a level 10 item just in case that makes some kind of difference.

I ended up with 115 successes for an average of 28.6%. So if the OP is correct and 402 is enough of a sample size to prove that the system is broken, then by my results he is correct. The RNG is broken and needs to be reduced because obviously it is giving RE crits more often than advertised.
Dear BW,
Please nerf paper and buff rock. Scissors is fine as is.
Signed, Rock.

Fidelicatessen's Avatar


Fidelicatessen
04.03.2014 , 02:18 AM | #156
Quote: Originally Posted by HelinCarnate View Post
Ok over the past few days I also did 402 REs just like the OP. They were all greens and I was learning them as I was leveling up my crafting so I was not 450 doing an RE on a level 10 item just in case that makes some kind of difference.

I ended up with 115 successes for an average of 28.6%. So if the OP is correct and 402 is enough of a sample size to prove that the system is broken, then by my results he is correct. The RNG is broken and needs to be reduced because obviously it is giving RE crits more often than advertised.
Well played!

Also I love your name.

znihilist's Avatar


znihilist
04.03.2014 , 12:19 PM | #157
Quote: Originally Posted by HelinCarnate View Post
Ok over the past few days I also did 402 REs just like the OP. They were all greens and I was learning them as I was leveling up my crafting so I was not 450 doing an RE on a level 10 item just in case that makes some kind of difference.

I ended up with 115 successes for an average of 28.6%. So if the OP is correct and 402 is enough of a sample size to prove that the system is broken, then by my results he is correct. The RNG is broken and needs to be reduced because obviously it is giving RE crits more often than advertised.
I know it is a joke, but people often forget that there is an uncertainty margin on these statistics. Someone doing 100 RE and finding a 15% success rate, then the theoretical average of 20% would still be within the uncertainty margin bounds.

psandak's Avatar


psandak
04.03.2014 , 02:52 PM | #158
Doing some further research into probability, margins of error, and confidence level, the OP's results do fall outside the norm:
  • For a sample size of 400 and a confidence level of 95% (standard), the margin of error is +/-5%. That means any results between 15% and 25% is representative of what probability says should happen.
  • The OP's sample resulted in a 13.4% RE success rate.
  • But all the that means is that the OP's data set is in that 5% outside the confidence level; run a 400 unit test 100 times and five of those tests will fall outside the 15% to 25% (acceptable/expected) range.

And HellinCarnate's test also falls outside the norm...that's two.

And just so we're clear, a sample of 50 or less has a 14% MoE; so any results between 6% and 34% RE success is expected.

znihilist's Avatar


znihilist
04.03.2014 , 06:13 PM | #159
Quote: Originally Posted by psandak View Post
Doing some further research into probability, margins of error, and confidence level, the OP's results do fall outside the norm:
  • For a sample size of 400 and a confidence level of 95% (standard), the margin of error is +/-5%. That means any results between 15% and 25% is representative of what probability says should happen.
  • The OP's sample resulted in a 13.4% RE success rate.
  • But all the that means is that the OP's data set is in that 5% outside the confidence level; run a 400 unit test 100 times and five of those tests will fall outside the 15% to 25% (acceptable/expected) range.

And HellinCarnate's test also falls outside the norm...that's two.

And just so we're clear, a sample of 50 or less has a 14% MoE; so any results between 6% and 34% RE success is expected.
I used to have a Professor in my Master studies who didn't like when we said margins of error, he kept saying if it is an error it can be corrected (fun times!)

Anyway, I have not been following the whole thread, so I can't really comment on what exactly being discussed.

However, a test showing a value being outside of the uncertainty margin doesn't mean something is wrong.
In my field (Physics), we usually require at least 5 sigma deviation before we claim we have made a discovery (in this case it is a messed up RNG). I wouldn't jump on a 13.4% chance with a theoretical value of 20% as an indication of anything is wrong.

psandak's Avatar


psandak
04.03.2014 , 07:48 PM | #160
Quote: Originally Posted by znihilist View Post
I used to have a Professor in my Master studies who didn't like when we said margins of error, he kept saying if it is an error it can be corrected (fun times!)

Anyway, I have not been following the whole thread, so I can't really comment on what exactly being discussed.

However, a test showing a value being outside of the uncertainty margin doesn't mean something is wrong.
In my field (Physics), we usually require at least 5 sigma deviation before we claim we have made a discovery (in this case it is a messed up RNG). I wouldn't jump on a 13.4% chance with a theoretical value of 20% as an indication of anything is wrong.
Your professor makes sense

I think you missed my point: while 13.4% and the 28% are outside the "uncertainty margin" for the sample sizes, they are only two tests; with a 95% confidence level there are bound to be tests that fall outside the "uncertainty margin," and those are two of them.