Some interesting conclusions had came up after the Korean Labs made the tests with Icarus Weapons. I took advantage of the PTS situation and compiled some data from experiment with S84 Mage / Fighter Weapons.
Enchanting 271 mage weapons and 218 fighter weapons seems that the 40% success for mage and 70% success for fighter observed from Korean Labs tests still holds true.
With 271 mage weapons I got 1 +9 and with 218 fighter weapons I got one +16.
Some new numbers regarding the chances for ++ OE weapons taking into account 40% and 70% are that with about 200-250 fighter weapons you get easilly at +16 whereas the number for +16 mage weapons is about 150.000 weapons needed:)
Never got into really calculating how much difference 40% and 70% would make but it really makes a great difference even at lower OE levels.
Get a fighter weapon, flip a coin (success rate for +5 about 50%) and expect a +5, roll a dice (success rate for +8 about 16%) and expect a +8.
On the other hand, get a mage weapon, flip a coin and be prepared for crystals as the most probable outcome( +4 chance lower than 50%), roll a dice and expect a +5 weapon.
Get 15 – 20 fighter weapons and expect to get a +10. To do the same with a mage weapon you would need at least 500 weapons.
On the other hand with 500 fighter weapons you could get 2-3 at +16.
Get 60 – 70 fighter weapons and expect +13 / +14 / +15 and a chance for the red weapon. With 60-70 mage weapons the best you can hope for is a +8.
Full article with actual data from testing, tables graphs and conclusions is posted at
It's statistics and has nothing to do with random factor. Rock can hit your head at random but once we talk about flipping a coin(or enchanting for that matter) it not random anymore.
So we got the world conspiracy theory again. The small problem is that there isn't any reason for this conspiracy. Because NCSoft could simply put 40% for mage weapons and 66,66% for non-mage weapons. Live with it or don't. Or they could put the 66,66% chance of success as an average chance with mage weapons much lower chance... which will mean the non-mage weapons must have greater chance of success then 66,66%... which doesn't correspond to the the findings of the author. End of course the author should know 200+ trials are not enough for a real statisctics.
BTW - a lot of time ago i tried to make a majestic set +6. For the gloves (which were 70kk each) I needed only 2 trials! But with the boots (14kk each) i made 49 fails.
It's statistics and has nothing to do with random factor. Rock can hit your head at random but once we talk about flipping a coin(or enchanting for that matter) it not random anymore.
I wonder why then, when you work in statistics, you use random variables XD
PS: and in any case, I'd call that probabilities...
You could repeat this test 500 more times, the percentage of success will always vary. Each time being different based on pure random luck or being unlucky a factor that would require a lot more data then you have access too.
So many factors being left out as well, testing blunts vs. swords. Variance in success of different grades of weapons A/S/S80Dyn/S80Ica/lowS84/highS84. Also the time between attempts, the sequence of attempts straight or stagered, all attempts/success on a server vs. a single players attempts/success.
The only thing a player can take away from the test you posted is that weapons blow up on attemps. And they should probaly consider the cost they may incur. how much better will you be with +16 weapon? Consider a 150 ele weapon +0 does more damage than 0 ele +16. Maybe the investment you should spend your time and money on is a 300 ele weapon or 120+ Armor pieces which would definately have much more benefit.
You can't give someone a static number or % of sucess with OE.
out of 10 mage weapons, and 10 melee weapons, blunt and sword...
10 mage weapons blow up before reaching +6
5 melee weapons reach +7, 1 +11...
ok, those numbers are fudged, because i never recorded the enchant succes rate...
but i find +7 is the barrier on melee weapons where 3 is the barrier of mage weapons.
basically your chance at +7 is close to 70%, while mage weapons your chance at +4 is 30%.
out of 24 mage weapons, one reached +6.
and that was a good rate...
blew up at +7 though.
although this is 84 on pts.
as for lower grade mage weapons, i'm not sure if it applies.
as i have gotten +5 on first try before with A grade...
I used to make a ton OEing dark screamers and changing them to homunkulus's at mammon to sell. Was heavily into enchanting back then and mage weapons just never work, blew up at least 10+ SOMs and max I had was +5, Homunks I wouldn't touch with a barge pole unless they were melee weapons.
Same on PTS, I've had a few +10 - 14 melee weapons, crafted a ton of icarus hands and picked up a bunch of sacrediums and only had one to +7, which was problably really lucky. Look at the gear people have, loads of glowing red gear, but mage weapons are still rare to see higher than +7.
You could repeat this test 500 more times, the percentage of success will always vary. Each time being different based on pure random luck or being unlucky a factor that would require a lot more data then you have access too.
So many factors being left out as well, testing blunts vs. swords. Variance in success of different grades of weapons A/S/S80Dyn/S80Ica/lowS84/highS84.
Sure the % success rate will vary, but it will be between some upper / lower bounds. If we were talking about 1% or 2% then you would need a lot more data than you could have access to. But when flipping a coin, 20-30 tests are enough to start discovering the 50% chance. In case of weapons we were trying to figure out a 67% chance. At least for the lower enchants the sample is more than enough to discover that the chance is not 67% but 40%.
According to my tests and Korean Labs tests there seems to be a ~40% success rate for mage weapons and ~70% for fighter. Also about the weapon grade, it's clear we reffer to S84. The same test from Korean had to do with Icarus. But when common wisdom, meets with 40% for Icarus, meets with 40% for S84, then .. there seems to be a pattern ..
About the blunt / swords issue. Carrying out the tests I used a mix of Sacrediums (blunt) and Archangel Swords (about the same number). They seemed to have identical behavior and that's why I made single tables / plots. As far as the fighter weapons were concerned, there were all types of them, the mix was made from what I picked up, so, there is a "random" non mage weapons mix, that's why I thought naming it fighter weapons seemed accurate. The objective was not to test against a specific weapon but the overall % success rates between 2 classes of S84 weapons
Also the time between attempts, the sequence of attempts straight or stagered, all attempts/success on a server vs. a single players attempts/success.
The time between attempts, the sequence of attempts, etc should not play any role at all (at least in theory with where OE attempts should be independant experiments)
The only thing a player can take away from the test you posted is that weapons blow up on attemps. And they should probaly consider the cost they may incur. how much better will you be with +16 weapon? Consider a 150 ele weapon +0 does more damage than 0 ele +16. Maybe the investment you should spend your time and money on is a 300 ele weapon or 120+ Armor pieces which would definately have much more benefit.
You can't give someone a static number or % of sucess with OE.
Yup, the main point is that it does not worth it. And yes, you can give a number, at least a good approximation one. 40% success rate is much more accurate than the 67% advertised.
Goes back to the old OE homunkulus thing.
I used to make a ton OEing dark screamers and changing them to homunkulus's at mammon to sell.
Yeap,anyone who was trying to make ++++homus knew that he had to oe daggers.
Exactly, I mentioned that, I also remember those good old times
Look at the gear people have, loads of glowing red gear, but mage weapons are still rare to see higher than +7.
Your observations are the same with mine and seem to confirm a +7 -> +8 break point, though, as I mentioned, the sample of weapons used to test +7 -> +8 could not be considered big. On the other hand, 11 weapons testing +7 -> +8 are not a small number either. Anyway, it could be bad luck, it could be a smal sample of weapons, or it could also be indeed a break point. IMHO it does not matter anyway, because you already need an insane number of weapons (25 best case scenario) to reach +7 or in other words, a +7 straight up using only one weapon has a 4% (best case) chance.
You could repeat this test 500 more times, the percentage of success will always vary.
Yes you are right. Data will vary. However what you fail to see it that variance from expected success rate would be only 1-3% off.
We have 2/3 or 67% expected rate of success and what results shows us that actual success rate is 40%. All variables you mentioned would end with results being between 64-73% success rate. Bad luck don't make 30% off.
I was wondering how long before someone would post something to validate the L2 urban myth that mage weapons are more difficult to enchant than fighter weapons. Seems we get one every year.
I'm not here to poke holes in anyone's findings, however off base I may believe them to be. I have my own enchanting history to fall back on, and that tells me that I can enchant a mage weapon to +16 and +17, because I have on multiple occasions on Live. Anecdotal? Perhaps. But try telling Jesus that he can't walk on water when he knows that he can.
I am here, however, to rail against the psychological effect of said findings. It's sad when you're on the PTS -- the PTS!!! -- and people are whining about how difficult it is to get a +16 Archangel or +16 Sacredium. AYFKM?!?
Of course it's difficult, but don't assume that because you see +16 Jade Claws, +16 Lava/Pysael Swords, +16 Eternal Core Swords, +16 Dual ECS, +16 Dual Mamba... +16 whatever-whatever... running around that enchanting those weapons wasn't difficult, too.
The difference is that when a Dual Skull Edge Dagger breaks trying only +4 through +8 forty or so times in a row (as was the case for me), no one complains about it. They move on, kill more elpies, buy more enchants, and keep trying for the bliss of a red weapon. But let a mage weapon pop at +4 and the "L2 Chicken Littles" come out of the woodwork.
If I was one of them, I'd be whining in Giran or compiling my "statistical findings" this very moment to prove that Skull Edge Dagger Duals cannot possibly follow NCSoft's STATED position that ALL weapon enchants have a 2 in 3 chance of success. I'd try to convince others that my failures are indicative of their ceiling, too. I would form an authority consensus with like-minded people who also had a difficult time enchanting Skull Edge Dagger duals to high levels...because there really aren't that many +16 Skull duals on the PTS, and that must mean we're right. Right?
Wrong.
**** the statistics. Treat enchants like the prettiest girl (or boy, for those whom this applies) that you want to ask out. She may have any number of reasons why she may accept or decline your offer, but at the end of the day, all you really will hear is "yes" or "no." Do as you will, but never listen to anyone who tries to discourage you, simply because she rejected them.
yes, it is all psychological.
unfortunately unlike you, most people consider numbers, which means that they will always fail.
for someone who the exception that proves the rule...
you sure do talk big.
good for you, we're all glad you cah achieve +16 mage weapons.
and in fact, we'd all appreciate it if you could get ours to +16, thanks in advance.
good for you, we're all glad you cah achieve +16 mage weapons. and in fact, we'd all appreciate it if you could get ours to +16, thanks in advance.
Sure, give me all of your assets and the enchants, step back and don't say a thing when I succeed or fail. But you could save us both some time, this needless back and forth & pointless speculation, and lots of aimless analysis and do it yourself...and live with the results.
All weapons have a 2 in 3 chance of a successful overenchant. Period.
I was wondering how long before someone would post something to validate the L2 urban myth that mage weapons are more difficult to enchant than fighter weapons. Seems we get one every year.
I'm not here to poke holes in anyone's findings, however off base I may believe them to be. I have my own enchanting history to fall back on, and that tells me that I can enchant a mage weapon to +16 and +17, because I have on multiple occasions on Live. Anecdotal? Perhaps. But try telling Jesus that he can't walk on water when he knows that he can.
I am here, however, to rail against the psychological effect of said findings. It's sad when you're on the PTS -- the PTS!!! -- and people are whining about how difficult it is to get a +16 Archangel or +16 Sacredium. AYFKM?!?
Of course it's difficult, but don't assume that because you see +16 Jade Claws, +16 Lava/Pysael Swords, +16 Eternal Core Swords, +16 Dual ECS, +16 Dual Mamba... +16 whatever-whatever... running around that enchanting those weapons wasn't difficult, too.
The difference is that when a Dual Skull Edge Dagger breaks trying only +4 through +8 forty or so times in a row (as was the case for me), no one complains about it. They move on, kill more elpies, buy more enchants, and keep trying for the bliss of a red weapon. But let a mage weapon pop at +4 and the "L2 Chicken Littles" come out of the woodwork.
If I was one of them, I'd be whining in Giran or compiling my "statistical findings" this very moment to prove that Skull Edge Dagger Duals cannot possibly follow NCSoft's STATED position that ALL weapon enchants have a 2 in 3 chance of success. I'd try to convince others that my failures are indicative of their ceiling, too. I would form an authority consensus with like-minded people who also had a difficult time enchanting Skull Edge Dagger duals to high levels...because there really aren't that many +16 Skull duals on the PTS, and that must mean we're right. Right?
Wrong.
**** the statistics. Treat enchants like the prettiest girl (or boy, for those whom this applies) that you want to ask out. She may have any number of reasons why she may accept or decline your offer, but at the end of the day, all you really will hear is "yes" or "no." Do as you will, but never listen to anyone who tries to discourage you, simply because she rejected them.
Not believing in statistics is almost the same as believing that sun is shining when it's actually storm all over.
If you say statistical proofs are just psychological imagination than we all live in imaginable world as every business which provides goods and services for us uses them in someway. If not statistics we wouldn't be here chatting.
Statistics are knowledge and knowledge is power. Having knowledge we can react according to it to to more with what we have or be more efficient.
However it's not the proof that we should stop OEin'g mage weapons. It's just a proof that they are harder to OE. In the end it's up to us whether we will take the risk or not.
Not believing in statistics is almost the same as believing that sun is shining when it's actually storm all over.
If you say statistical proofs are just psychological imagination than we all live in imaginable world as every business which provides goods and services for us uses them in someway. If not statistics we wouldn't be here chatting.
We're in the shallow end of the pool, clearly. I'll try not to splash as I run by you. I'm sure you grasp the subtlety of my post, but let's not make any assumptions.
Stats are not knowledge, hence power. Forget what you learned in Marketing or your Statistical Methodology course. Or better yet, go back and re-read some of those texts. You missed a crucial part of the lesson. Stats simply reflect the outcomes of recorded events. What we do with that data, how we interpret it, is where the trouble begins. Surely you see that.
Believing in stats? I didn't realize this was a faith-based process.
You mention the sun, fittingly, since I'm sure at one time there was a stat which "proved" that the sun always rises and will continue to trend that way. I'm sure everyone accepted that as fact and power was obtained and exercised throughout the land, and all was good in the world. Until one day astronomers observed a distant supernova for the first time, a star not unlike our own, having a really, really bad day. I imagine if that star system had sentient life, say statisticians, they'd be burned at the stake for their statistical findings...well, if their sun hadn't done the job already. (Maybe they, too, were told that there was a 2 in 3 chance of the sun going kablooey, but they ignored it because their observations deemed otherwise???)
Well, there you have it. Info via leaked sourcecode, and two independant experiments by a player on the current PTS and NCSoft employees on the Korean PTS result in data which pretty much confirms the numbers from the leaked sourcecode. I'd say that pretty much settles it, at least I personally will work with these numbers from now on to judge if enchanting something is worth the risk. Others are free to believe whatever they want, I don't care if they hang on to the obviously wrong "everything is 2/3 chance" info. Interesting by the way that the OE chance for weapons stays the same until +15, while armor OE chance goes down with every plus, so that trying to go for higher than +6 DEFINITELY is not worth the risk.
**** the statistics. Treat enchants like the prettiest girl (or boy, for those whom this applies) that you want to ask out. She may have any number of reasons why she may accept or decline your offer, but at the end of the day, all you really will hear is "yes" or "no." Do as you will, but never listen to anyone who tries to discourage you, simply because she rejected them.
The analogy is not correct. Asking someone out cannot be formalized in any mathematical way, the fact that there are only 2 possible outcomes means nothing.
Only if it was "known" that girls / boys you ask to go out "should" say yes in 67 / 100 times asked, you could make some reasonable assumptions. In that 67/100 case, if a friend of yours went to dinner with 70/100 girls asked and you went to dinner with 40/100 _maybe_ then you could make a logical assumption that something goes wrong either with you or the girls you asked for dinner :)
A better example (not involving human behaviour) would be to have 2 coins that should be identical, constructed to give you 67 % heads and 33 % Tails. For some "unknown" reason, a lot of people suspect for years (though without any evidence to prove it) that coins are not the same.
Then I flip the first coin 250 times and note down the results, flip the second coin 250 times and note down the results.
Then a random sample of people start flipping coins random number times and also get the impression that coins are not the same. Apart from the people flipping the coins, any "outside" observer can see many Heads from the 1st coin but many Tails from the 2nd. There are only 2 explanations here.
1) People mass hallucinating, statistics fail (close to paradox), coins are identical.
2) People are correct, stats are correct, coins are not identical.
Everyone can choose which explanation seems fit.
I just flipped the coins many times and noted the results. My results seem to agree with previous results and anecdotal evidence. If you ask me I take the simplest, logic explanation. The 2 coins are not identical :) Someone switched sides on the second coin :)
I can't believe people are actually saying "screw statistics" and such..... this is a computer game, enchant results are detirmined by a logical "random" algorithim, of course theres a statistical rate of success or failure! EG: each enchant a random number is generated...
If Num >= 7 then Fail
if Num < 7 then Success
After a large number of enchant attempts the succes rates will start to show clearly. You could fail 20 items in a row, you could succeed 20 in a row, that's just luck, but eventaully with enough attempts the statistical rate of success will come true.
Whilst the success rates of the enchants are interesting I am more interested in how the "random" number is generated.... as it's a well known fact that it is impossible for a computer to generate a truly random number it has to rely on performing an algorithim on a certain event (usually milliseconds in time).
Whilst the success rates of the enchants are interesting I am more interested in how the "random" number is generated.... as it's a well known fact that it is impossible for a computer to generate a truly random number it has to rely on performing an algorithim on a certain event (usually milliseconds in time).
Whilst the success rates of the enchants are interesting I am more interested in how the "random" number is generated.... as it's a well known fact that it is impossible for a computer to generate a truly random number it has to rely on performing an algorithim on a certain event (usually milliseconds in time).
Generating a pseudo-random row is the most basic command for every CPU. But it is still random because the randomness depends on exact time when you press the key... which is completely random of course.
BUT how this random number is used in the game - this already depends on the programmers. It may be a server-random, account-random or toon-random. I presume it is server random 'coz 1. Making it account-random or toon-random will load the server CPU, 2. There is no reason to make something other besides server-random.
BUT there are many other random actions in the server - actually every hit is random, so it may be toon-random after all:-)
Generating a pseudo-random row is the most basic command for every CPU. But it is still random because the randomness depends on exact time when you press the key... which is completely random of course.
BUT how this random number is used in the game - this already depends on the programmers. It may be a server-random, account-random or toon-random. I presume it is server random 'coz 1. Making it account-random or toon-random will load the server CPU, 2. There is no reason to make something other besides server-random.
BUT there are many other random actions in the server - actually every hit is random, so it may be toon-random after all:-)
Didn't know psuedo random number generation was a process of the CPU, I always just understood it as a software funtion that performed an algorithim based on a certain number, like milliseconds of cpu time as I said earlier. It's not something I've looked massively into though, I'm more experienced in easier programming languages such as C# and Java than anything low level.
It wouldn't suprise me though if the calculation was extended to take into account factors such as wit, character name, position etc. I remember someone mentioning a glitch in earlier chronicles where a certain location in game enabled people to enchant items with 100% success rate, probably just a myth though.
The fact is, if there were a truly reliable way to improve enchant success then there would be programs out there to do it, and many many more +16 weapons out there. Although there are some people about with tons and tons of +16 weapons.... one in particular has about 5-6 accounts sat in grian with sellshops full of +16 gear and +16 weapons equipped on each toon.... this doesn't mean she knows a trick though, just that she might be rich irl.
Didn't know psuedo random number generation was a process of the CPU, I always just understood it as a software funtion that performed an algorithim based on a certain number, like milliseconds of cpu time as I said earlier. It's not something I've looked massively into though, I'm more experienced in easier programming languages such as C# and Java than anything low level.
It wouldn't suprise me though if the calculation was extended to take into account factors such as wit, character name, position etc. I remember someone mentioning a glitch in earlier chronicles where a certain location in game enabled people to enchant items with 100% success rate, probably just a myth though.
The fact is, if there were a truly reliable way to improve enchant success then there would be programs out there to do it, and many many more +16 weapons out there. Although there are some people about with tons and tons of +16 weapons.... one in particular has about 5-6 accounts sat in grian with sellshops full of +16 gear and +16 weapons equipped on each toon.... this doesn't mean she knows a trick though, just that she might be rich irl.
There was a program that produced +++++ weapons and a ''service'' to make +30 skills.Nscoft
find out when a draconic bow was oed to +28 on Teon if i remember correct 2 years ago,when such enchant cant be done with game mechanics and they found a way to stop the use of this program and only recently they reset skills of those toons to 0.
Also there was an exploit you could somehow safe oe several weapons-maybe thats why they added to game the system message everytime someone oe gear-,but i have also seen that some toons seem can oe gear/skills with much less fails,while others always break on +4 whatever they try to oe.If the randomess is bound on account or toon,then its totaly unfair and should be changed.
is it calculated on server side? or on user side?
it might be interesting if certain computers were just better at enchanting...
so if mr lucky enchants were to sit at someone elses computer, they might see a normal fail rate, while that other player sits at lucky pants computer and achieves +16 within 20 enchants(mage weapon)
is it calculated on server side? or on user side?
it might be interesting if certain computers were just better at enchanting...
so if mr lucky enchants were to sit at someone elses computer, they might see a normal fail rate, while that other player sits at lucky pants computer and achieves +16 within 20 enchants(mage weapon)
Ofc it's server side lol.... In our side exploit random factor by braking into running game files would be rather easy.
And yes there was programs that produced safe enchanting but I think those programs took info from leaked C4 files. Information was analyzed, week spots detected in client-server communication and later exploited.
Didn't know psuedo random number generation was a process of the CPU, I always just understood it as a software funtion that performed an algorithim based on a certain number, like milliseconds of cpu time as I said earlier.
I don't know how it is done; I am not a programmer and studied this looong time ago so i don't actually remember anything besides the fact it always generated absolutely the same pseudo-random row. I guess you are correct and this is basic command from low level programming language. But pressing the key "stop" (or "enchant") is what gives the real randomness.
I would accept the mage weapons have lower chance for success if there was some reason NCSoft to do it and then to lie about it. Actually introducing another chance of success for mage weapons means additional programming work to be done and I simply see no reason for them to lie about this.
Also the recorded results on l2wh from the real OE shows the chance is really 2/3. The recorded results are for all kind of weapons but if the mage weapons were at 40% and non-mage at 66,66% then the recorded average success would be less then 66,66%.
Also the recorded results on l2wh from the real OE shows the chance is really 2/3. The recorded results are for all kind of weapons but if the mage weapons were at 40% and non-mage at 66,66% then the recorded average success would be less then 66,66%.
If you mean http://www.l2wh.com/tools/oes.php take a closer look to the ratio of fighter / mage weapons. The data is comprised almost exclusively from fighter weapons, there are 15 times less mage weapons. You could easily spot the difference by observing there the rates for DS and Homu. According to the same page that states that all are 68%
The problem with l2wh page is that they calculated the average 68% on a random sample of weapons and generalized that it applies to all. If you take into account that the vast majority of weapons recorded are fighter ones, that their chance is about 70% and not 66% and that the fighter weapons give higher OE levels (so many more trials and numbers to "average") it's easy to see how they missed the exception.
They are averaging having a lot of fighter weapons with 70% and 15 times less mage weapons with 40%. Averaging a high number / high rate of fighter weapons and a low number / low rate of mage weapons would just create an average for fighter weapons slightly lower than the real one and not a general OE rate for all. Actual for fighter was 70%, expected for all types was 66%, average of all gave 68%, it all fits well. The situation is a bit different ..
Lets say we have 1500 trials with fighter weapons and 100 trials with mage weapons (about l2wh ratio). Then lets suppose that 70% of fighter and 40% of mage succeed. You could calculate an average of 68%
(1500x70% + 100x40%) / 1600 = 68%
That means that if the data has 15 times more "non mage" weapons, you could come up to 2 different conclusions
1) All weapons have the same rate, 68%
2) Though mage weapons were 1/15 of the sample, their number is big enough and differences are to big to calculate an average, thus fighter weapons are 70% and mage weapons are 40%
I was always wondering why it has gone unnoticed for so long on a very popular / well informed (these times) site. Even the users of the site OE'd DS and went to mammon to change them for Homu :)