These forums have been archived and are now read-only.

The new forums are live and can be found at https://forums.eveonline.com/

Science & Industry

 
  • Topic is locked indefinitely.
123Next page
 

T2 Invention oddity

First post
Author
Mister Falcons
NF Plains
#1 - 2014-08-22 06:00:44 UTC
Over the last 2 day's I tried for the first time my hand at T2 Invention. I had made several Multiple runs copies of Mining Crystals. I have Level 3 in the skills Amarr Encryption, Laser Physics, and Electronic Engineering. Giving me a probability of 46% chance to Invent T2 Crystal BPC. I had a few decrypts of various amounts to test addition that raised my average chance. I had following combination I used in the first 20 tries to invent: 2x +20% chance (got 0 with this combination), 3x +50% chance (got 1 here), and 2 +80% chance (got to T2 BPC's here). the other 13 was ran with no decrypts. Based on my base of 46% chance to get BPC we could expect 9 wins of a T2 BPC, considering I added improvements we should actually see about this factor by a bit. When it was all said and done I only got 5 BPC's.
The second time thinking maybe the adding decrypts maybe that messed with the results, so I ran another 20 with no decrypts giving me 46% chance flat out to get T2 BPC. Again I expect to receive 9 BPC's. What I got was 4 BPC's. This is half the expected results.
While I realize that its a Random Number Generator that is resulting in my win/lose chance, but the average over 40x attempts of half expected returns indicates a potential bug in how the random number generator is being used for the calculation.
Lady Naween
Ministry of War
Amarr Empire
#2 - 2014-08-22 06:11:58 UTC  |  Edited by: Lady Naween
this old chestnut.

it is random, and your sample size is TIIIIIIIIIIIIIIIIIIIIIIIIIIIIINY. Do many many many more and you will see the system actually works just fine.

see threads like https://forums.eveonline.com/default.aspx?g=posts&t=365901&find=unread
Elena Thiesant
The Scope
Gallente Federation
#3 - 2014-08-22 09:58:42 UTC
If you do some searching, you'll see this comes up again and again and again. Random is random. Specifically, each invention attempt has that 46% chance of success. The attempts are independent, so failing one does not make it more likely to succeed the next.

Yesterday I completed 10 ammo inventions (47% chance of success) and got two successes. Last week, I completed 10 ammo inventions (47% chance of success) and got nine successes. This is normal

45% chance of success does not mean 9 successes out of every 20. It means each one of the 20 has an independent 45% chance of success. They may all fail. They may all succeed (though the odds of either are really low). It's a bell curve, over time and with large numbers of invention attempts (thousands), your success rate will tend towards the average. That doesn't mean that every batch has to meet the average exactly. It won't.
Cyniac
Twilight Star Rangers
#4 - 2014-08-22 12:54:02 UTC
As has been said nothing odd about your results...

Except that you used decryptors to invent ammo... now I find that odd, though I freely admit it's been a very long time since I actually invented ammo.

Anyhow best of luck stick with it, and don't worry. It happens. Kinda more annoying as the invention tries get more expensive though.
Shadow Preldent
Deadspace Exploration Conglomerates
#5 - 2014-08-22 16:05:57 UTC
Elena Thiesant wrote:
If you do some searching, you'll see this comes up again and again and again. Random is random. Specifically, each invention attempt has that 46% chance of success. The attempts are independent, so failing one does not make it more likely to succeed the next.

But when we start aggregating samples we can say something about the probability of a particular result happening.


Quote:

Yesterday I completed 10 ammo inventions (47% chance of success) and got two successes. Last week, I completed 10 ammo inventions (47% chance of success) and got nine successes. This is normal

If you told me "I ran two sets of 10x invention jobs at 50%. The first set yielded a result outside the range 3-7. The second set yielded results outside the range 2-8."

I would say "The probability of that happening is less than .235%. That's not normal, that's out of the ordinary"



Quote:

45% chance of success does not mean 9 successes out of every 20. It means each one of the 20 has an independent 45% chance of success. They may all fail. They may all succeed (though the odds of either are really low). It's a bell curve, over time and with large numbers of invention attempts (thousands), your success rate will tend towards the average. That doesn't mean that every batch has to meet the average exactly. It won't.


Lets look at the OPs data. 9 out of 40 (assuming 50% probability). The probability of the result being outside the range 10-30 are less than .068%. Ok, either he was the unluckiest inventor in Eve that day (its got to happen to someone and they are much more likely to come and post than someone who invented 20/40) or the RNG is not as random as you think it is. Its output can have an average value of .5 without it being random at all.

If you look at the guy who claimed 20/100. Then either you don't believe his numbers (my opinion), you believe that CCP does not use a RNG for invention, or you don't believe in math.

Buy hey, they are API calls for invention results. Anyone who's done lots of invention jobs care any sharing the result data?

-SP
Falin Whalen
GoonWaffe
Goonswarm Federation
#6 - 2014-08-22 16:53:06 UTC
RELEVENT

Get educated scrub.

"it's only because of their stupidity that they're able to be so sure of themselves." The Trial - Franz Kafka 

Qoi
Exert Force
#7 - 2014-08-22 17:50:19 UTC
I'd love to see someone trustworthy do some invention with a samplesize >200 and share the results. There have been some indicators (the 20/100 story if true is a pretty strong one), but nothing conclusive. And of course data reported on the forums after the fact will always be heavily biased. Blink

http://eve-industry.org

Ginger Barbarella
#8 - 2014-08-22 17:54:06 UTC
Anecdotally my post-Crius runs without using meta items to boost the probability has been 40-45%. Using meta items to boost probability my average across non-ships is 55-65%. I've only been record the results post-Crius off and on.

And yeah, I really wish people would use the Search function; they'd see this this comes up every single FRIGGIN' time there's a patch, whether it's relevent to Industry or not.

"Blow it all on Quafe and strippers." --- Sorlac

Shadow Preldent
Deadspace Exploration Conglomerates
#9 - 2014-08-22 20:59:42 UTC
Falin Whalen wrote:
RELEVENT

Get educated scrub.



What exactly does independence have to do with this discussion.

Does anyone think that CCP uses a RNG??? At best its a PRNG so future values can be predicted from previous values which means they aren't very independent.

Even so a good PRNG would not yield the results reported.

If you're going to participate in a useless dork discussion for dorks at least try and understand the material.

-SP
Falin Whalen
GoonWaffe
Goonswarm Federation
#10 - 2014-08-22 21:48:02 UTC
Shadow Preldent wrote:
Falin Whalen wrote:
RELEVENT

Get educated scrub.



What exactly does independence have to do with this discussion.

Does anyone think that CCP uses a RNG??? At best its a PRNG so future values can be predicted from previous values which means they aren't very independent.

Even so a good PRNG would not yield the results reported.

If you're going to participate in a useless dork discussion for dorks at least try and understand the material.

-SP

Each individual attempt at inventing is independant of the others. This means that you can't put X amount of jobs in and expect Y to be succesful, that is just stupid. I've had runs where I put in eleven invention jobs at 46% and had ten come out succesful and others where only two were succesful. Sample size on each was to low to make any conclusion, but overall the many thousasnds of independant trials of invention chance i've done, it has worked out to about where it states invention chance is.

If you can't grasp basic probability theory I suppose you will just have to be one of those people who whine on the forums about how you were supposed to get Y succeful runs on your X attempts at inventing, "CCP must have changed something."

"it's only because of their stupidity that they're able to be so sure of themselves." The Trial - Franz Kafka 

Rayzilla Zaraki
Yin Jian Enterprises
#11 - 2014-08-22 23:19:52 UTC
For my first 50 shots at T2 I had the opposite results - 37 wins, 13 losses.

For the next 150 or so it evened back out to about 45%.

Gate campers are just Carebears with anger issues.

Shadow Preldent
Deadspace Exploration Conglomerates
#12 - 2014-08-22 23:24:55 UTC
Falin Whalen wrote:

Each individual attempt at inventing is independant of the others. This means that you can't put X amount of jobs in and expect Y to be succesful, that is just stupid. I've had runs where I put in eleven invention jobs at 46% and had ten come out succesful and others where only two were succesful. Sample size on each was to low to make any conclusion, but overall the many thousasnds of independant trials of invention chance i've done, it has worked out to about where it states invention chance is.

If you can't grasp basic probability theory I suppose you will just have to be one of those people who whine on the forums about how you were supposed to get Y succeful runs on your X attempts at inventing, "CCP must have changed something."



1) Each individual attempt at inventing is not independent of the others. It is generated by a computer. Computers do not make random things (unless CCP seeds their generator with random data which I highly doubt). You can predict the next outcome based on the previous outcome. That of course does not matter for this discussion, but let's at least be factually correct.

2)
Quote:
This means that you can't put X amount of jobs in and expect Y to be succesful, that is just stupid.

But you can say what is the Probability of Y out of X jobs being successful. This is what probabilistic system analysis is. And if you understood my posts, you would see that is what I was saying.

3)
Quote:
I've had runs where I put in eleven invention jobs at 46% and had ten come out succesful

And math will tell us what is the probability of that happening. In this case it is approx .252%. If this results happens (on average) more often than once every 396 batches then we can so that the system is not "fair" (as in unbiased or random) even if your average invention rate works out to 46%.
We can also say that if someone experiences a result what has a probability of occurring once every 1.9 billion batches that it is unlikely that the system is actually "fair" given the short amount of time that Eve has been in existence. Or that he was lying (still the most likely in my opinion)

4)
Quote:
If you can't grasp basic probability theory I suppose you will just have to be one of those people who whine on the forums about how you were supposed to get Y succeful runs on your X attempts at inventing, "CCP must have changed something."

When in this post did I either
a) Claim you you should get a particular result. I stated the probability of a particular result given a "fair" system and suggest that either people were lying about their results (most likely) or the system was not "fair" and that we would need a lot more data to prove that.
b) whine about anything unless doing math is whining is which case, yes, i do whine all the time
c) say anything about CCP changing their algorithm.


Also, spell check is your friend. It's embedded in browsers these days. There really is no excuse.
adriaans
Ankaa.
Nair Al-Zaurak
#13 - 2014-08-22 23:49:05 UTC
Track it over several thousand attempts and it is very close to what it should be.

For instance, for the past week I've had 14/100 on one character but 78/100 on another. Random is random, your sample sizes are WAY to small. Minimum for any sample size of value should at least be 1k+ attempts.

----True oldschool solo pvp'er---- My latest vid: Insanity IV

Falin Whalen
GoonWaffe
Goonswarm Federation
#14 - 2014-08-23 01:05:00 UTC  |  Edited by: Falin Whalen
Shadow Preldent wrote:
1) Each individual attempt at inventing is not independent of the others. It is generated by a computer. Computers do not make random things (unless CCP seeds their generator with random data which I highly doubt). You can predict the next outcome based on the previous outcome. That of course does not matter for this discussion, but let's at least be factually correct.

I'm going to stop you right here. This is a clasic example of the Gambler's fallacy. Each time you put in an invention job means you flip the coin on wheather it is succesful or not when you deliver it. YOU CAN'T BASE THE NEXT OUTCOME FROM THE PREVIOUS OUTCOMES! Each is an independant trial of the invention chance.

You can toss a coin and have 20 heads come up in a row and the next toss have tails, you can also toss a coin and have heads come up 20 times and have heads come up the next, the odds are both 1 in 2,097,152. Therefore, it is equally likely to flip 21 heads as it is to flip 20 heads and then 1 tail when flipping a fair coin 21 times. Furthermore, these two probabilities are equally as likely as any other 21-flip combinations that can be obtained (there are 2,097,152 total); all 21-flip combinations will have probabilities equal to 1 in 2,097,152. From these observations, there is no reason to assume at any point that a change of luck is warranted based on prior trials (flips), because every outcome observed will always have been as likely as the other outcomes that were not observed for that particular trial, given a fair coin. Therefore, the result of each trial comes down to the base probability of the fair coin: 1⁄2.

You being able to "predict" the outcome, based on previous outcomes, is a flat out fantasy your rational mind tries to put on random events.

EDIT:

Gambler's fallacy arises out of a belief in a "law of small numbers", or the erroneous belief that small samples must be representative of the larger population. According to the fallacy, "streaks" must eventually even out in order to be representative. Amos Tversky and Daniel Kahneman first proposed that the gambler's fallacy is a cognitive bias produced by a psychological heuristic called the representativeness heuristic, which states that people evaluate the probability of a certain event by assessing how similar it is to events they have experienced before, and how similar the events surrounding those two processes are. According to this view, "after observing a long run of red on the roulette wheel, for example, most people erroneously believe that black will result in a more representative sequence than the occurrence of an additional red", so people expect that a short run of random outcomes should share properties of a longer run, specifically in that deviations from average should balance out. When people are asked to make up a random-looking sequence of coin tosses, they tend to make sequences where the proportion of heads to tails stays closer to 0.5 in any short segment than would be predicted by chance (insensitivity to sample size); Kahneman and Tversky interpret this to mean that people believe short sequences of random events should be representative of longer ones. The representativeness heuristic is also cited behind the related phenomenon of the clustering illusion, according to which people see streaks of random events as being non-random when such streaks are actually much more likely to occur in small samples than people expect.

"it's only because of their stupidity that they're able to be so sure of themselves." The Trial - Franz Kafka 

Komi Toran
Perkone
Caldari State
#15 - 2014-08-23 05:56:16 UTC
Falin Whalen wrote:
I'm going to stop you right here. This is a clasic example of the Gambler's fallacy.

No. Not really. He's accurately describing the mechanism behind a PRNG. The rules are different than real-world distributions because the randomness is only simulated. In many PRNGs, the sequence will eventually repeat, which means that it is possible to determine your future results based on past outcomes. In fact, if you watch speed runs of some older console games, you will see that the runners have gotten quite good at manipulating the basic PRNGs in those games to create predictable outcomes.

Now, you aren't going to be able to pull that off in Eve, but that's a practical matter due to the game's complexity, not a technical impossibility.
Shadow Preldent
Deadspace Exploration Conglomerates
#16 - 2014-08-23 07:11:30 UTC
Komi Toran wrote:
Falin Whalen wrote:
I'm going to stop you right here. This is a clasic example of the Gambler's fallacy.

No. Not really. He's accurately describing the mechanism behind a PRNG. The rules are different than real-world distributions because the randomness is only simulated. In many PRNGs, the sequence will eventually repeat, which means that it is possible to determine your future results based on past outcomes. In fact, if you watch speed runs of some older console games, you will see that the runners have gotten quite good at manipulating the basic PRNGs in those games to create predictable outcomes.

Now, you aren't going to be able to pull that off in Eve, but that's a practical matter due to the game's complexity, not a technical impossibility.


No point in arguing with him.

If he doesn't understand that flipping a fair coin is not the same thing as a computer's rand function, and he doesn't understand how you use Bayes' theorem to calculate the probability of a particular outcome there's nothing we can do.

He just quotes "random is random", without understanding what that means statistically.

Then again, I tried to explain math to someone on the internet so who's really the bigger idiot?
Steve Ronuken
Fuzzwork Enterprises
Vote Steve Ronuken for CSM
#17 - 2014-08-23 15:02:13 UTC
Of course, most PRNGs are seeded with data which may as well be random.

A common one is using data derived from the incoming data on an network interface. Sure, if you can record it, you could recreate it, but with it coming from so many different sources, any other way of recreating it is pretty much doomed to failure.

tl;dr:
Yes, you can predict it, if you know enough about the initial stated. No, you won't know enough about the initial state.


People who whinge about PRNGs not being truly random, for a purpose such as invention in Eve Online, are, hmm. I'm having trouble coming up with a phrase that the ISD wouldn't have problems with.

Woo! CSM XI!

Fuzzwork Enterprises

Twitter: @fuzzysteve on Twitter

Falin Whalen
GoonWaffe
Goonswarm Federation
#18 - 2014-08-23 16:24:50 UTC  |  Edited by: Falin Whalen
Listen, I understand the difference between a coin flip and a computers rand function, and you can calculate Bayesian probability curves to your hearts content. It still doesn't handwave away that each individual invention attempt is an independant trial, and only after all events are set can you place it on your Bayesian curve and go "Oh, that was neat, a .252% probability for that one."

Statistics is the study of the collection, organization, analysis, interpretation and presentation of data, you don't have data untill all events are resolved. Yes you can run analysis on all previous invention runs, and get your Bayesian probability curve statistics, Bayesian STATISTICS will "predict" a certain amount of succesful invention runs within 1 or 2 standard deviations, just like Bayesian STATISTICS will "predict" the amount of heads or tails for a given number of coin flips within 1 or 2 standard deviations. Going, "I ran X, amount of invention runs, I should have Y succesful runs +- 1 SD." is a prime example of the gamblers fallacy. You can't know the outcome, but with Bayesian statistics you have [a pretty good guess, and that is all it is, a pretty good guess.

EDIT: Casino owners would love to have you Bayesian guys come over and try your hand at roulette.

"it's only because of their stupidity that they're able to be so sure of themselves." The Trial - Franz Kafka 

Komi Toran
Perkone
Caldari State
#19 - 2014-08-23 23:54:36 UTC
This statement...
Falin Whalen wrote:
Listen, I understand the difference between a coin flip and a computers rand function

...is disproved by this statement:
Falin Whalen wrote:
EDIT: Casino owners would love to have you Bayesian guys come over and try your hand at roulette.

Falin Whalen
GoonWaffe
Goonswarm Federation
#20 - 2014-08-24 04:47:21 UTC  |  Edited by: Falin Whalen
So what you are saying is, that you and Steve here can, with 100% accuracy, predict the outcome of each individual invention job based solely on the previous pattern of invention job successes/failures?

If so, then have you reported this bug/exploit to CCP?

If not, then you are deluding yourselves with apophenia.

"it's only because of their stupidity that they're able to be so sure of themselves." The Trial - Franz Kafka 

123Next page