| Pages: [1] 2 :: one page |
|
|
| Author |
Thread Statistics | Show CCP posts - 0 post(s) |
KVR
   |
Posted - 2008.04.25 10:39:00 -
[1]
Copy and Paste of petition to CCP:
Dear CCP,
I think I have an issue related to the invention. I have launched 14 jobs of invention. First 9 were on my former corp. Last 5 were on my own. I have Minmatar ship, minmatar encryption at lvl4, mechanical at level 5. same with caldari. from those 14 i just got ONE BPC 1 RUN OF A KITSUNE, USING THE AVERAGE DATACORES. I Never saw something like that before., usually it were something lie 50 % success but i'm not getting any. Can you please check that issue?
I have been inventing for the last 8 months or so. I had an average of 50% more or less Invention success. Once I got mechanical engineering % i thought I was going to have a superior chance of invention success. Then I found out that using the average decryptors my success rate has lowered to a 5%. What the hell is happening here? I have launched 3 caldari dictor, 6 kitsune, 4 vegabond and one mammoth invention and i just got 1 copy 1 run of a kitsune from all. I have actually 8m sp on science skills, almost all on invention. I'm supposed to have being wasting my time tor 6 months or more?
|
Kazuo Ishiguro
House of Marbles
Zzz
|
Posted - 2008.04.25 10:44:00 -
[2]
Your sample is statistically insignificant. Please come back when you've done several hundred jobs (hint: the test server is a good place to try this sort of thing).
My research services
Spreadsheets: Top speed calculation - Halo Implant stats |
KVR
|
Posted - 2008.04.25 10:45:00 -
[3]
I forgot to add I have used in 3 cases the 2nd level decryptor and that I have all races encryption method 4, all 4 starship engineering at 4 and mechanical engineering at 5. I have also used in 9 cases POS invention and in 5 empire NPC station invention
|
KVR
|
Posted - 2008.04.25 10:48:00 -
[4]
Insignificant? those are the last 14 jobs, i have runned hundreds of them before and have had success on 50% more or less. This is absolutely not the first time, I've been inventing stuff on last 6 months non-stop.
|
El'essar Viocragh
Minmatar
KULT Production
Atrum Tempestas Foedus
|
Posted - 2008.04.25 11:00:00 -
[5]
Then show us you books and argue why the last 14 of your several hundred runs are a trend and not a fluke. Thanks.
--
[17:47] <Mephysto> its dead, jim |
Venkul Mul
Gallente
|
Posted - 2008.04.25 11:05:00 -
[6]
It is insignificant, statistically speaking, as a streak of 13 success or failures is well within the normal probability range (about 5/10.000 for a mix of frigates and cruisers with max skills). Considering the large number of invention jobs run every day it is granted that someone will get that kind of result.
If after running another 100-200 jobs you have still that kind of failures you can think that something is wrong. Just to point it out I had several streaks of 9 failures or 9 success (doing 10 jobs at a time) with modules and, as the chances of failure on module are much lower, the chances of that happening are around 7/10.000. But considering several hundred jobs my success rate is a steady 50-60% (depending on the module).
|
Shintai
Gallente
Balad Naran Orbital Shipyards
|
Posted - 2008.04.25 11:07:00 -
[7]
50% invention rate? Then you ahve been lucky. In average you should be around 35% so on frigs so stop whining.
Abstraction and Transcendence: Nature, Shintai, and Geometry |
KVR
|
Posted - 2008.04.25 11:10:00 -
[8]
At my point of view, if you have been running usually invention and you got an average invention rate of almost 50% based on a 15 invention jobs and you suddenly notice that on the last 15 jobs you have a 5% success there is something that is not working. Are you saying that I have been incredibly lucky and got almost all success on all the previous months or are you just angry and want to show it here? If you get used to have an average rate you get shocked when you waste bpc and encryptors to get nothing. Anyway, thanks for your unbelieavably kindness, this is the kind of response that any player in eve searches when comes to ask something in Forums.
|
KVR
|
Posted - 2008.04.25 11:15:00 -
[9]
Thanks Venkul for your response. I will try on the next processes to get a wider overview of my reat, and then came "whine" again if is not so fruitful. 
|
Man Bewbs
|
Posted - 2008.04.25 11:18:00 -
[10]
 Originally by: KVR
this is the kind of response that any player in eve searches when comes to ask something in Forums.
There was a Question??
I thought you just cut/pasted a petition
|
|
|
KVR
|
Posted - 2008.04.25 11:29:00 -
[11]
Originally by: Man Bewbs
Originally by: KVR
this is the kind of response that any player in eve searches when comes to ask something in Forums.
There was a Question??
I thought you just cut/pasted a petition
When I cut/paste something here is because I wanted to know if it was just me or simply I did not had any idea of what was happening. On my daily Eve play I try to help and teach the newbies or the people that asks for help on the matter that I'm used to work with. This is the kind of stuff you like when you play that game, there is always people that tries to help you with their knowledge and experience, almost the same when you ask for help on RL. But as a mirror of RL too, there is always smartasses proud of themselves that shows their own level replying like if they were the fecking kings of universe. Make me a favor and STFU if not going to help with a minimum of class.
|
Valenthra
Caldari
School of Applied Knowledge
|
Posted - 2008.04.25 11:34:00 -
[12]
I am no inventor myself but... Welcome too EvE with odds and % :P You just had a unlucky streak nothing else.
|
Shintai
Gallente
Balad Naran Orbital Shipyards
|
Posted - 2008.04.25 11:41:00 -
[13]
Originally by: KVR
At my point of view, if you have been running usually invention and you got an average invention rate of almost 50% based on a 15 invention jobs and you suddenly notice that on the last 15 jobs you have a 5% success there is something that is not working. Are you saying that I have been incredibly lucky and got almost all success on all the previous months or are you just angry and want to show it here? If you get used to have an average rate you get shocked when you waste bpc and encryptors to get nothing. Anyway, thanks for your unbelieavably kindness, this is the kind of response that any player in eve searches when comes to ask something in Forums.
Why would I be angry? Its a game.
But its simple math. The invention rate on frigs is 35%. So whenever you do enough inventions your average will be 35%. Sometimes you can have 15 successes in a row, other times 25 failues in a row. Statistics 101.
Abstraction and Transcendence: Nature, Shintai, and Geometry |
Schneiderr
Asgard Schiffswerften
Ev0ke
|
Posted - 2008.04.25 11:41:00 -
[14]
waiting for your next thread when you moan about getting 13 out of 14 inventions successul, cause it doesnt fit into your books.
this:
Originally by: Venkul Mul
Considering the large number of invention jobs run every day it is granted that someone will get that kind of result.
|
Man Bewbs
|
Posted - 2008.04.25 12:04:00 -
[15]
Edited by: Man Bewbs on 25/04/2008 12:04:28
Originally by: KVR
But as a mirror of RL too, there is always smartasses proud of themselves that shows their own level replying like if they were the fecking kings of universe. Make me a favor and STFU if not going to help with a minimum of class.
Sorry if i offended in my reply.
It wasnt my intention.
I guess just didnt realise that the Petition was infact the question.
PLease do let us know what answer you get to that petition. Im sure many here would be interested in it.
FYI - it has been mentiond in other threads ( cant find a link atm) that the random number generator use in EvE to decide things like success or otherwise has abnormal streaking. in that you often get many Fails in a row or many Success in a row rather than a truly random result.
Perhaps you hit one of these streaks
|
Idxx
Amarr
Interstellar eXodus
R0ADKILL
|
Posted - 2008.04.25 15:20:00 -
[16]
We had a long discussion about this. Check out this link: http://oldforums.eveonline.com/?a=topic&threadID=705811
and specifically my posT:
Just ran a quick test. Given a sample size of 1 million, with a widely accepted rng, i had the following results:
1,000,000 sample size, 50% boundary
Run 1
10+ streaks 490
Longest streak 17
Run 2
10+ streaks 535
Longest streak 18
Run 3
10+ streaks 568
Longest Streak 19
This means that an invention chance of 50%, which is more than most of us have for any ship type (can best decryptor and frig even get this high?) then you'll see around 500 "unlucky" streaks of 10 in every million attempts, and some unlucky bastard will go 19 attempts before a success *at 50% success rate*
Look at a smaller number, like battleship chance. I think this is 20% with no decryptor:
1,000,000 Sample size, 80% boundary
Run 1
10+ streaks 22174
Longest streak 58
Run 2
10+ streaks 22241
Longest streak 58
Run 3
10+ streaks 22301
longest streak 58
This means, on average, we'll see 22301 streaks of 10 failures or more, when inventing something with 20% chance, in every 1,000,000 attempts. And ya, some unlucky bastard will fail 58 battleships in a row.
How realistic is 1,000,000? How long will it take to see 22301 failures? Well assume there are 1000 professional inventors in game (no idea how accurate this is). This includes multiple accounts for real people. Assume each inventor tries oh, 25 battleship inventions a week. It would take less than a year (40 weeks, to be exact) to see 22301 streaks of 10 failures or more in a row.
No wonder there are so many posts about it.
|
Lara Dantreb
New Horizons
|
Posted - 2008.04.25 17:25:00 -
[17]
It happened to me to run 30 ship invention jobs with no success and then to have 5 successfull jobs in a row.
you made CCP staff waste time with your petition
|
Zifrian
Gallente
GFB Scientific
Interstellar Corporate Alliance
|
Posted - 2008.04.25 18:05:00 -
[18]
Why in the world would you ever need a million runs to test something. You only need 1000 for a 99% confidence. A million is overkill gesh.
One question though, is the chance based on the installation or on the user? If it is just when the user hits deliver and it runs the rng with your skills and whatnot, then other users do not matter. I assume this is what it is because it makes the most sense and would be the easiest.
So, if you have a 35% invention success rate, then 35 out of 100 or 3 out of 10 would be what you see. If you saw 50% on 100 jobs, then 1 out of 14 is just in the cards for you.
I've seen it in other MMO's too and I doubt CCP made a overly difficult system to calculate success rates. I'm not sure why people have such a hard time with this type of stuff. It seems pretty simple to me. |
Idxx
Amarr
Interstellar eXodus
R0ADKILL
|
Posted - 2008.04.25 19:15:00 -
[19]
1000 is way too small. I'll look for the formula.
Its pretty much accepted that the result is calculated based on the person starting the job, not the person pressing deliver. |
Zifrian
Gallente
GFB Scientific
Interstellar Corporate Alliance
|
Posted - 2008.04.25 20:17:00 -
[20]
Edited by: Zifrian on 25/04/2008 20:18:43
You don't need a sample size any larger than 1000 for normal statistics to get a +/- 5% error. 10000 would get you to +/- 1% but it depends on what confidence interval you are looking for.
It really is irrelevant though, what is needed is the formula for determining the success rate. This could be done through regression but one would need 40+ data points and to know all the variables that affect the success of a job. Right now it seems like the skills, decryptors, meta level of an item, and the bp seem to have an effect. I'm not sure if there is anything else.
I'm sure someone has done this or CCP has hinted at it. If not, I could do one for ****s and giggles but I'd need the data. Might be fun anyway |
|
|
Salvis Tallan
Gallente
The Shadow Order
SMASH Alliance
|
Posted - 2008.04.25 20:21:00 -
[21]
Linky
Also the invention formula is known. |
Burchov
Lyrus Associates
|
Posted - 2008.04.25 20:30:00 -
[22]
Originally by: Salvis Tallan
Also the invention formula is known.
Source?
________
Invention profit calculator |
Salvis Tallan
Gallente
The Shadow Order
SMASH Alliance
|
Posted - 2008.04.25 20:33:00 -
[23]
Originally by: Burchov
Originally by: Salvis Tallan
Also the invention formula is known.
Source?
Someone who I trust enough to go and make this claim publicly.
------
|
Idxx
Amarr
Interstellar eXodus
R0ADKILL
|
Posted - 2008.04.25 20:36:00 -
[24]
Originally by: Zifrian
Edited by: Zifrian on 25/04/2008 20:18:43
You don't need a sample size any larger than 1000 for normal statistics to get a +/- 5% error. 10000 would get you to +/- 1% but it depends on what confidence interval you are looking for.
It really is irrelevant though, what is needed is the formula for determining the success rate. This could be done through regression but one would need 40+ data points and to know all the variables that affect the success of a job. Right now it seems like the skills, decryptors, meta level of an item, and the bp seem to have an effect. I'm not sure if there is anything else.
I'm sure someone has done this or CCP has hinted at it. If not, I could do one for ****s and giggles but I'd need the data. Might be fun anyway
Perhaps for just testing the reliability of the rng to generate 50% correctly, but not for testing streaks.
|
Burchov
Lyrus Associates
|
Posted - 2008.04.25 20:40:00 -
[25]
Originally by: Salvis Tallan
Someone who I trust enough to go and make this claim publicly.
Which he did (at least I assume you're referring to that). Not that it's dev-confirmed, but meh, that'll never happen anyway.
/me updates the invention calculator.
________
Invention profit calculator |
Salvis Tallan
Gallente
The Shadow Order
SMASH Alliance
|
Posted - 2008.04.25 20:46:00 -
[26]
Originally by: Burchov
Originally by: Salvis Tallan
Someone who I trust enough to go and make this claim publicly.
Which he did (at least I assume you're referring to that). Not that it's dev-confirmed, but meh, that'll never happen anyway.
/me updates the invention calculator.
Like I said, I trust him enough to take his word. So the real issue now isnt the equation for success, but rather that huge debate on whether or not the random number generator is random enough. |
Schani Kratnorr
x13
|
Posted - 2008.04.25 21:25:00 -
[27]
Broken or not, Invention is a waste of online time.
I have spoken with many people who do invention on all levels, and despite their success or lack thereof, they all agree that it's booring as hell.
For the sake of everyone's sanity, please redesign the whole thing.
|
Verrain Ravenlark
|
Posted - 2008.04.25 21:26:00 -
[28]
Originally by: Salvis Tallan
Originally by: Burchov
Originally by: Salvis Tallan
Someone who I trust enough to go and make this claim publicly.
Which he did (at least I assume you're referring to that). Not that it's dev-confirmed, but meh, that'll never happen anyway.
/me updates the invention calculator.
Like I said, I trust him enough to take his word. So the real issue now isnt the equation for success, but rather that huge debate on whether or not the random number generator is random enough.
Which is probably isn't. If they are using the standard Python (or any language) randomization libraries you will eventually see statistically anomalous number streaks over large enough samples. With a massive sample size these streaks are actually somewhat predictable because of the seeds used to generate the data.
You need to use atmospheric noise or radioactive decay to get a truly random value.
You can see http://www.random.org/randomness/ for a good description of randomness in CS.
|
Dread Varog
|
Posted - 2008.04.27 01:01:00 -
[29]
wow allot of great info here never would have imagined runs of successes or losses so high thx cuz i was about to whine about my 50% streak, that went into 16 failures in a row 
|
Shintai
Gallente
Balad Naran Orbital Shipyards
|
Posted - 2008.04.27 08:16:00 -
[30]
Originally by: Dread Varog
wow allot of great info here never would have imagined runs of successes or losses so high thx cuz i was about to whine about my 50% streak, that went into 16 failures in a row
Hehe, im up on 15 failed ishtars in a row now and maybe still counting. But I did have 6 in a row with success earlier.
Abstraction and Transcendence: Nature, Shintai, and Geometry |
|
|
|
|
| |
|
| Pages: [1] 2 :: one page |
| First page | Previous page | Next page | Last page |