Im a bit lost :)

Excellent work and correct. I'd say that you are one of the first people since Armando who has the necessary understanding and background to intelligently discuss these sorts of issues.

And that is the real issue here, isn't it.

But here is where I disagree. We deal with events in life all the time that are "fair" in that everyone has the same chance, but "unfair" in that people had different results. It is simply a part of life and we all accept it... until it is a game and we want it to be "artificially fair." That is what I'd call what you are describing. You want the developers to implement systems that are solely based on resources invested (time, money, that sort of thing). While that sounds great on the surface, the problem is, BP will never do it. Games that are grind based, especially F2P grind games with a micro-transaction system in place, need randomness to encourage spending. Whether or not you agree with it, you must realize that this will never change. Sure, things with hidden or weighted odds may eventually have to be published to comply with gaming regulations, but "fair randomness" will never disappear from a game like this. If you want an "Artificially Fair" game, you are much better off playing one that is single player or P2P from the stats. Then, the developer makes their money on the front end and doesn't need the randomness to make money.

This is true... I'm pretty sure not a single one of them ran any sort of statistical analysis.

If we are making analogies, going to work and getting paid a regular/predictable amount is like farming Materi frags in the PWs with the intent to buy a unique, or running events monthly to get draken. Mini events with randomness and the jesters are more like scratch off lottery games and big lotteries. To continue that analogy, most of the more important aspects in the game have been adjusted to remove or mitigate randomness in the long run (i.e. with enough effort). In fact, I think the only truly important system that still has any randomness left is crafting, but the devs have indicated that they are looking at how they might revamp the crafting system.
Even still, Crafting behaves like a random event "with replacement" on failure, so with time and enough attempts, the odds approach 100%.... It is just expensive to approach the 100% mark.

True, the mathematical exercise was mostly academic, but you address my point above about most things having the randomness mitigated.

 

Not so much fixed as this was the whole purpose of making them random. If the devs just did some statistical analysis, they could take the randomness out and advertise the expected amount of time to get the items. I don't think this would reduce the income from these sorts of events.

Yes! I certainly do want games to be artificially fair. If games werent better than the real world, I wouldn't like games.
I don't have any real hope that BP will change this. However, I think there is good precedent from other F2P grind games that don't have as much randomness, and certainly no direct gambling. The two big ones I am thinking of is the one with exiles and the one with space ninjas (a shooter). The two big other things the space ninja game does is have a player trading system (to trade items) and weekly (or at least very frequent) dev streams. The company that develops the space ninja game is also mostly owned by a chinese company, so that isn't the main problem as has been said by others. I do like the the materi frags, cores, monthly events, glyphs, and gems. I want to spend money to support the game, but I don't feel like the devs care about me, so I wont.

 

Incorrect
The costume drop was not 100% ... you could ended up without getting a single costume in 20 or more runs.
You could ended up getting 5 costumes after 200 runs ... and all 5 being a duplicate.
You got all the costumes because they increased the drop on a big scale at the end of the event.
I was running for more than 2000 times without getting a single costume that I needed ... fortunately I am watching football so I knew when is the last chance to buy them. And I bought them on time.

 

Try again; you failed to read what I actually typed. I am very deliberate in my word choices and my words mean no more and no less than what they say.

For my hypothetical analysis, I was looking at how many instances of the costume dropping you need before you get what you are looking for. I'm well aware that the costume didn't drop every kill, so if you were trying to estimate how many total Stromball trolls you'd need to kill, you'd need to perform a secondary analysis to determine how often the drop included the costume too and use that as a multiplier with how many costume drops you'd need to determine how many total kills you'd need on average.


And you wonder why I don't take your word as proof enough that crafting is rigged? Your attention to detail at times leaves something to be desired.

 

You trying to make me stupid or you really think i am stupid?
Not just you never said anything about it ... but your BS math and statistics are just BS.

Either your dictionary is very poor or you are writing your posts in Telepathic Language that no one can read.

Yeah right

No dude ... i was not trying to estimate anything. You tried
If you tried ... You should have done it in the first place ... not me.
And you didn't

No ... i don't wonder ... because I don't care if you take my word or you don't ... as much as i don't care about your BS math stories.

 

The Gurus started fighting each other, eh? That's by no means a good omen...

I fully understand what was your analysis about. But I don't understand why did you keep saying "drops" and really avoided saying that drop amount isn't equal to kill amount. Every new player or player that had a break at the time of the event could get it the wrong way easily.
One more thing: You couldn't determine that easily how often did the drop include a costume... since that one was weighted. Also, there were sprees of drops containing a costume and sprees without a single costume... with the second ones seemingly slightly longer and happening more often. You would need an extra large data set, containing at least ~10k records, to get trustworthy results. And a set that large would require a work of more than a single person, since it's an immense amount of kills.
That's why this was the only viable strategy to me:

Moreover, I don't consider hypothetical analysis the best way to raise knowledge on things determined by PRNG... Maths may work everywhere in our world (even though without practise, the theory is nothing), but not on PRNG systems... They are way off the mathematical truth, they can't be predicted using maths. I would call them faulty algorithms that are less random than our everyday choices.

And I am wondering why don't you really try to to pay more attention to all the crafting failures you are getting... I understood really quickly the very same thing as Traki did... and I have been looking closely at them since not-that-long time. If Traki claims he has been watching that for over a year... he has for sure seen way more than I did. And regarding Traki's attention to detail... you're right.

So, in conclusion: You are both wrong. You should both try thinking thoroughly through each other's opinions...
@_Baragain_ In my opinion, your belief that you can apply maths (especially statistics) right away to every thing in the world and get it the correct way is too firm. It's not always that way.
@trakilaki You're too pesimistic... and you know very well what I am talking about. (and you are drinking too much lately)

 

So, in conclusion: You are both wrong. You should both try thinking thoroughly through each other's opinions...
@_Baragain_ In my opinion, your belief that you can apply maths (especially statistics) right away to every thing in the world and get it the correct way is too firm. It's not always that way.
@trakilaki You're too pesimistic... and you know very well what I am talking about. (and you are drinking too much lately)[/QUOTE]

I have to take serious issue with this statement. The fact that the gurus are arguing is not a bad thing. They both offer valuable insight and when it is an opinionated scenario, this is good, because additional informed opinions can help a researching player make up their mind. I don't sense any hostility in the disagreement.

My counterpoints are:
1. I don't see Baragain applying mathematics to everything in the world, just to the mathematically driven world of equipment / battle mechanics, drop rates, and crafting in his posts. Applying math to a mathematical situation is not a matter of opinion. When you are working with numbers, and applying the proper formula, the numbers do not lie, they never lie, and the result will be identical every time. No amount of wanting, feeling, or diplomacy will change this. It is a universal truth.

2.Traki being pessimistic is highly subjective. I think telling someone who has a broad basis of experience with the game, having played for a number of years, being pessimistic is justified. As much as I love this game, I will be the first to admit I have a garbage attitude about it quite often, and seeing someone else rant about the stuff that grinds my gears makes me smile, and I cool down.

Ultimately what I am saying is, let them argue. The likely result is more data (both factual and opinionated) available to the community, which we desperately need, since BP can't be bothered to produce any.

 

To draw any conclusions, you need lots and lots of data. Fortunately, I produce lots and lots of data on a daily basis with easily several hundred, if not over one thousand crafts a week. Problem is, I'd have to spend the time to record every one of those crafts in painstaking detail. Right now, a craft takes me about 10 seconds or less. Put all the items on the bench, click craft, check to see if gold lines transferred, click revert, repeat. If I were to record the necessary data to do any sort of meaningful analysis, I'd likely have to take about 1.5-2 min per craft to record all the relevant data into an excel sheet. And the larger the crafts (legendary to legendary) might take even longer since there are more stats to record. Increasing the time it takes to craft by 9-12 times would interfere with group runs, timed bonuses/buffs, and ultimatly waste a large portion of the limited time I have to play on generating this data set.

As I've stated before, that is a serious time investment to attempt to prove something that I have no reason to believe needs proving and where there is insufficient hard evidence to the contrary to convince me otherwise and motivate me to waste my valuable time.

Thank you for expressing it so succinctly. The game is a computer program, meaning that, at it's core, it is driven entirely by math. The only variable that I can never fully account for is the nature of the PRNG. Sure, we have seen some anecdotal evidence that the PRNG has short strings of improbable results (positive or negative), but these could even occur in a "fair" PRNG and we'd never be able to tell the difference unless until someone can definitively prove what causes it and if it is controllable. Until such a time, there exists no better method to objectively analyze this game than a mathematically driven approach based on a "fair" PRNG. Trakilaki can be pissed off at me all he wants and make all the subjective claims he wants, but the fact remains that my math, and those that do similar analyses, are the best and only reasonable models for understanding the game at any objective and impartial level.

This is actually related to something called "Negativity Bias." In a nutshell, people are more likely to recall negative events over positive or neutral events. It is part of human nature. I recognize the existence of this bias in my own perceptions, and while I can't control any emotions or subconscious thoughts related to negative events, I try to compensate for this bias in other ways. In real life, it may take the form of purposely recalling a happy event that occurred under similar circumstances. Or, in a video game driven by a PRNG, I realize that it is just math and eventually, with enough time and tries, eventually the numbers will be in my favor. In this way, I am not a pessimist or an optimist, but rather a realist. The only time pessimism related to this game is fully justified is when it is directed towards the developer/production team because they are humans and can make bad decisions and boneheaded choices... things that a computer program like dro_client.exe can't do.

He and I are very different. I'm logical to a fault, and he lives by his experiences. The one thing that we unquestionably have in common is that we are fans of the game and do everything we can to give back to the community. It takes different forms, but we both want to see this game and it's community improve and grow. I can still respect him, even if I don't agree with anything he has to say on a topic.

That is why I never posted that analysis during the event. It was complicated and to do a full write up at the time wouldn't have been possible with the time I had at my disposal because of real life things. That excerpt was included here as an example of how poorly people instinctually understand "random" events. An age old example is coin flipping. Is it possible to flip a coin 10 times and get heads or tails every time or a perfect H-T-H-T-H-T...? You bet it is. Just because it's unlikely to happen doesn't mean that it isn't random if it does happen. For example, have you ever noticed the game's coin flip emote has a "thumbs up" and "thumbs down" version of the animation? If a subjective person did 10 flips in a row and got 10 "thumbs up" animations, they'd say that the PRNG was not working right. While they may be right and the PRNG may be giving a string of "less than random" results, they could also be very wrong since this event could occur in real life (see linked video). Are you starting to see why I never trust human perception of "random" events when they also have a preexisting bias related to their perception of the nature of the PRNG?

See the above about heads and tails and the "anecdotal evidence" of PRNG "biased strings."

It would be an average over time. If the rate that the drop included a costume was 10%, then the 123 "drops" for the 50/50 odds of being done would go up to 1230 "Kills" before you have a 50/50 chance of having collected all the costumes. The analysis becomes even more complicated when you consider the probabilities of the large flags replacing the small flag in the drops... you can see why I didn't take the time to write it up during the event? Especially since people could buy the ones they wanted?

 

What fight?
That is how the gurus share different opinions in order to come to appropriate solution in the middle.

We are always arguing ... you can check the old post we are always having opposite opinions on some things, especially Randomness.
This argument is going on for years.

I don't know where did you get this ... but I have never had any drinking problems in my life. Just because I like to drink Rakija, Vodka or White Rum it doesn't mean I am drinking too much In fact last time I have been moderately drunk was more than a year and a half ago. I do eat every day too ... that doesn't mean I am obese ... In fact I have almost 0% fat.

Absolutely ridiculous
You are confusing PSEUDO with TRUE ... one is related to real life and and the first one is related to computer applications only.
the very meaning of the word pseudo explains it all. You can take it as "imitation or artificial" ... but the other synonyms are: bogus, false, fraudulent, deceptive, misleading ... etc.
I will quote my answer to Sebastian:

----------------------------------------------------- QUOTE -----------------------------------------------------------------------------

OK let me start ...

Quite big words ... for a human being as we know we humans are having pretty much limited mind.
Easy and straightforward to calculate ... maybe. But is it correct?
Let me give you a simple example of "easy and straightforward" calculation:
~~~ Empty bus calculation ~~~
We are having an empty bus which is driving on its own from bus station A to bus station G. Bus station A is a starting point and bus station G is a end point ... there are also bus stations B, C, D, E and F in between. The bus should start empty from the starting point ... let people get in and get out on its route ... and stop empty at the ending point.
- the bus starts empty from the station A
- 3 people get in; no one gets out - at station B
- 2 people get out; 3 people get in - at bus station C
- 3 people get out; 6 people get in - at bus station D
- 10 people get out; no one gets in - at bus station E
- no one gets out; 3 people get in - at bus station F
- The bus arrives empty at bus station G
Pretty simple and straightforward and also mathematically correct ... right?
But is that really possible in real life? No it is not.
I will give you another example where mathematics and logic are involved:
~~~ Painting a wall ~~~
One man can paint a wall in 4 days. Mathematically correct. Logically correct
2 man can paint it in 2 days. Mathematically correct. Logically correct
4 man can paint it in 1 day. Mathematically correct. Logically still correct
8 man can paint it in half a day. Mathematically correct. Logically also correct
16 man can paint it in 1/4 th a day. Mathematically correct. Logically may be correct
....
....
....
128 man can paint it in 45mins. Mathematically correct. Logically almost wrong
256 man can paint it in 22.3 mins. Mathematically correct. Logically wrong
512 man can paint it in 11 min. Mathematically correct. Logically impossible
1024 man can paint it in 6 min. Mathematically correct. Logically impossible

"Ah ... math is awesome ... we love math" (good for you ... I don't like mathematics I love women ) but let us see what mathematics really is.
Mathematics is just a tool for analyzing and understanding the real world problems ... and a lot of it is not created by discovery but by design. It deals with idealized abstractions and pure concepts that exist only in human minds. It is a perfect eternal realm of absolute truth completely separated from physical reality. All theories, theorems and formulae have been derived from axioms. And what are those axioms? Axioms are just assumptions ... and these assumptions might be or might not be accurate in real life.

Spoiler: Joke

Physics professor has been doing an experiment, and has worked out an empirical equation that seems to explain his data. He asks the math professor to look at it.
A week later, the math professor says the equation is invalid. By then, the physics professor has used his equation to predict the results of further experiments, and he is getting excellent results, so he asks the math professor to look again.
Another week goes by, and they meet once more. The math professor tells the physics professor the equation does work, "But only in the trivial case where the numbers are real and positive."

Almost everything we are dealing with on our computers is not just mathematics but rather physics. Because computers are real they are not imaginary.
Let me show you first (before getting to the main question on this topic) that your example is vague and incorrect.

For a start ... when using measurements we are using metric system ... we are not using feet, elbows or nails
Like Mikey already said

In reality all modern day experiments and researches have shown that the odds are always something like 51/49 in favor to whatever side was up when the coin is thrown into the air.
But coins are not perfect ... they are usually having one side heavier than the other one. They are also getting dirt over the time and some of them are not even perfectly symmetric. But lets assume we are having a perfect coin. In this case we must not neglect the environment.
Toss a coin in Amsterdam and toss a coin on Mount Everest.
Or ... simply toss a coin in a water pool while diving.
Will you get same results? No you will not.

Spoiler: Joke #2

A farmer has problems with his chickens: all of the sudden, they are all getting very sick. After trying all conventional means, he calls a mathematician to see if they can figure out what is wrong. The mathematician tries. He stands there and looks at the chickens for a long time without touching them or anything. Then all of the sudden he starts scribbling away in a notebook. Finally, after several gruesome calculations, he exclaims, "I've got it! But it only works for spherical chickens in a vacuum."

So lets move to the next part ... that ~9Km fall.
You are speculating again and assuming things up.
You can't start an experiment with only your assumption. You have to measure things up and get valid data.
You are trying to simplify the things up ... while in reality they are so complex and almost impossible to predict or measure.
In order to get it right ... you have to measure and calculate:
- height, mass, weight, body structure, body shape, health condition ... etc etc etc
- weather conditions and air pressure, density, humidity, wind direction and strength ... etc etc etc
- you should check the perimeter if anything might obstruct your experiment in any way
- etc ... etc ...etc
Your experiment will end up differently every time you conduct it over and over again ... it will never be the same. So how can you calculate the odds to something you can't even get to work out or predict?
It can always end up in different scenarios ...
- the person might always die
- the person might not even land
- the person might only suffer broken bone
- the person might land unhurt ... what if there is a force ... like strong wind ... that can negate the gravity. You know something like the example with he bullet shot from a train in motion. Mythbusters actually proved this to be true with a football shot from truck.
- even if there is a force that can negate the other forces and the person taking the fall land like a feather ... unhurt ... there is always possibility the person still to be dead because of heart failure before even starting the fall
- etc etc etc
So you are comparing and calculate the odds of the first example with the coin and the second example with a person falling from ~9KM height ... both examples in which you can't measure and predict things up.

So you may say "in what way all this has any relation with the topic" and your claims that "everything is easy and straightforward".
Very easy (actually very complex ). Like i said it is all physics not mathematics. Computer is real device it is not existing only in our minds. So if it is a real life device ... you have to have realistic approach to the calculation.
How do we know that the code and the algorithm has been written in such way so it can represent the Theory of Probability (which of course is a theory not a proof) in best possible faultless way?
We can't.
But even if we know the code and it is perfect (no possibility of errors) ... we can't neglect the environment. One small example:
you see everything inside computer is working only if there is a electric current running through the separate parts. Every part is made from some material and they are almost always different with different properties. When a device is working it creates a heat ... and heat is their worst enemy. You can't negate the heat ... you can lower it down to certain point and control it but you can't negate it. Many components are temperature sensitive and can have different properties with different temperature.
The point here is ... like professor Michio Kaku says "systematic errors creep into very delicate calculations"

And that is only one example ... not to mention that different computers are having different architecture (we are not talking computer architecture here) therefore they will act in different ways for a same executed application and many other things that can lead to different outcomes. In fact we have already been talking on this topic on few occasions here in the forum ... some time ago.

I challenge you to run ^ couple of automated tests and simulate the drop 10K times ... but on 10K different machines
Let me know of the outcome.
I would predict every single machine would probably have a different drop ... but like i said predictions and assumptions without real data are wrong.


Chemistry is physics without thought.
Mathematics is physics without purpose.

---------------------------------------------------------------End Quote --------------------------------------------------------------

You are aware yourself that you can never prove me wrong
The only way one of us to prove the other one wrong is by committing a felony ... that is the only way. In that case we will both get perma-banned ... so I don't think it is worth trying. Lets continue in the good old fashion

 

I didn't try to say that it was biased. I was just trying to say that such sprees may significantly alter the outcome of any statistical calculations and they happening a lot mean that it would require a much greater data set to reduce the alterations so that your outcome will be generally true.

And once again:

PRNG systems in games are hardly ever based on regular rand() functions that the programming language the game is written in has to offer. Usually, they take a few values from the game (eventually current server time) and process them with a mathematical algorithm.

Yes... as I said before quoting you, it's a mathematical algorithm. But the real problem is, what does the algorithm take into account?
This is an example of a PRNG algorithm that looks pretty random:

Code:

bool prngLinePassingDetermine(lineValue)
1. Take current server time
2. float a=[(hour)mod(7)]+[(minute)mod(7)]+(hundredth-of-second)
3. a=a/111
4. a=a*(playerXpos)*(playerYpos)
5. a=a/c
6. if a>lineValue return true else return false

c=some constant based on what position values could be. If they were ranging from 0 to 20, it could be 200 ((20^2+0^2)/2).

And that is just an example of how the algorithm could work to, at the same time, seem random and be completly stupid. I am not saying this is what the game is using, but I consider it (whatever it is) equally nonsense.

For understanding the balance between characters, or for understanding what builds are better and what are worse... true, I am doing it myself quite often (even if I am not sharing my calculations in forums like you or some other people do). But not regarding the PRNG. Unlike the "random" tools in real life (like dices), PRNG is not even trying to be somewhere near "truly random" (which, again, doesn't exist, and I'm one of those who do not believe it exists even on quantum level). Like in the example I provided above, it can be predicted only if you know the algorithm, and the algorithm in crafting is probably way more complicated than that. You can't even say that every line has an equal theoretical probablity to pass, because that would require the line value (or ID - but that is a different topic) not to be used in the algorithm... which I doubt, and anyways you have no reason to say it's the other way around.

 

That they do. In fact, I've even dealt with that before when I did my last true analysis of a "random" event in DSO, one where we know the theoretical rate. I analyzed the behavior of crit rate and determined if the PRNG could be trusted to generate a weighted random number based on a character's crit rate. The sample sizes were rather small (115-130 range), but the results were promising and not unexpected/unusual enough to justify going back and running the experiment with a larger sample. I kept track of two stats. One was the running crit rate (the total average of the sample after every hit) and a "local average" of ten before, the current hit, and the 9 after. In my first example, I also listed my longest chain of crits and non crits.

So, without actually figuring out the average rate that trolls dropped costumes in addition to a flag, I couldn't even begin to tell you if "5/9 runs" or "3 in a row" are to be expected, or if they are a true statistical anomaly.

For that matter, hash function with a low collision rate could be used as a PRNG, but hashes take longer and more processing power, so that is unlikely. Or, they could use any old large polynomial and use a good potential seed. Or they could.... well, the list of things they could do is very long and ultimately irrelevant since we have no clue what they are using. The PRNG is a black box that we can't peer inside of (unless we commit that felony that Trakilaki was referring to ), and we can only make assumptions based on the results we get when we put stuff in. My point is that, based on what I've actually analyzed, the PRNG behaves decently. I've seen many claim otherwise, but very few try to prove it.

But, until you can see inside that black box, it is "random enough" that no one has been able to prove otherwise.

Says who? If the PRNG has low collisions, like a good hashing algorithm, the using the line values or IDs wouldn't make a difference because, by the time you add in the other inputs, the outcome should be unique enough that even having two identical inputs (Item#1 and Item#2 in the case of crafting) would be irrelevant.
Hash "hello world" and then hash "hello World" and a good algorithm will return values so different that you would have no clue the inputs were related. In case you are curious, here are those two strings hashed with SHA256. Can you tell they are only one letter different from looking at the outcome? Or do the strings "appear" "random enough" even though they are generated by a completely predictable and repeatable (albeit near irreversible) process?

Code:

B94D27B9934D3E08A52E52D7DA7DABFAC484EFE37A5380EE9088F7ACE2EFCDE9

DB4067CEC62C58BF8B2F8982071E77C082DA9E00924BF3631F3B024FA54E7D7E

Code tags used so that the numbers line up and are easier to spot common values.

They both have '4' as their 3rd value, '7' as their 6th value, and each have 'D' as their 3rd from last. 4 and 7 are both one greater than their position in the hash. Does this count as a pattern, or a coincidence?
If you ran server time through something like this and used the hundredths of a second as your seed, the output would change so drastically without repetition, each hundredth of a second that you'd have a very strong random number generator.

I have no logical way to argue one way or the other, but I'd argue from another perspective:
Programmers are lazy.
It is a lot easier to number the input lines from 1-4, 1-8, 1-12, or 1-16 and have the PRNG pick 1, 2, 3, or 4 numbers from that pool without replacement. Then, they don't need to get creative in setting up their inputs based on the circumstances of calling a random number and can simply call "dso_rand()" any time they need a random number and then process that random number for what they need. Not that it would matter if their PRNG is halfway decent to begin with, as described above.

Your arguments are interesting from a programmatic standpoint, but arguing about DSO's PRNG is a dead end discussion because any argument about the PRNG being "not random" should actually read "not random enough" and needs statistical evidence that demonstrates such. I may not have huge samples, but I'm the only one who's done actual analysis and my results fit in the "random enough" category. So, until someone demonstrates objective, concrete, statistical evidence that the PRNG is not "random enough," I will continue to use the assumption of "fair random" in any analysis because no one has any viable alternative.

Oh yes... and to this day, there still isn't a winner. I'm too lazy to prove my point, and your point is intrinsically unprovable without math that you won't/can't do (can't "prove" feelings with more feelings, after all). And thus, we are at a standstill.

 

-ponders- If we make a few assumptions that 1) the underlying crafting system follows a similar pattern regardless of item type (ie. boots vs gloves vs belt won't have huge differences) 2) rarity with an increase in the number of lines is only a magnification of the basic mechanism... Then couldn't you simplify your data collection by focusing on a specific item type (eg. belt) which would completely eliminate needing other items? Then further simply by saying green-to-blue craft only? And then in order to eliminate other variables which you could explore in other experiments, maybe you just pick the specific tier you always farm in so you only throw say... t4 green belts on the workbench and record the crafts. And if you don't want to waste any blue items you make, just throw them into slot 3 and 4 of whatever blue items you were crafting anyways.

Ok, done stirring the pot. -runs-

 

We are both wrong ... your math and my observation are wrong ... but only when speaking about crafting.
You want math I will give you the "the most complex" math as a proof:
6>4
5>4
I know you can figure it out what it is and where are we wrong

 

The only thing I can think you are talking about is using good gold lines of the type that you aren't crafting for in the second or third slots (in addition to good gold line of the type you do want) into slots three and four in hopes that the stat duplication/transformation bug comes along and blesses you with a little luck. Like using this item in slot 3/4 in hopes of getting a 18.314% crit damage line or a 18.2xx% line if the stat duplication/transformation bug kicks in.


Other than that, care to share?

 

Sharing is my middle name

So here how the story goes.
I was questioning myself ... how is it possible the crafting is much more difficult now than before.
I had to think with clear head and took your math in one hand ... and took my experiments and results in the other.
I said to myself ... his math has to be accurate (which is, until we start talking PRNG vs TRNG) BUT my results are accurate and precise as well ... because I can lie to you but I can't lie to myself. So just like every math theory starts with assumption ... and then we have to prove the assumption is correct ... that is exactly how I approached.
Any alleged bug had to be taken out of consideration.
I was puzzling myself for an hour or bit longer thinking "What is the missing link that will connect these two" . Then i started rolling the dice on the workbench with some items.
Believe it or not ... I have got 2 exactly the same outcomes in 5 rolls , using green items. But OK, improved items are not a benchmark since it is quite easy to get same outcome few times in few rolls.
And then ... I have noticed something we were not paying attention to.
WTH ... it was so easy and so simple ... and it was in front of our eyes all the time.
Bloody hell ... I was neglecting my own Wiki crafting guides.
There I got the missing link ...
You are referring to your and Armando's calculations in your posts ... but if you click the link you will notice those are dating back to 2016.
That is Craft 2.0 and later on it is referring to Crap Next Generation.
And that is the mistake we were both doing.
We were talking about nonexistent crafting system ... we were talking about THIS crafting.
But the New Crafting is currently present in the game.
So what has been changed?
This

Spoiler: This

is what is different.
That is the missing link.
We are crafting unique items therefore we are neglecting there are 2 more "enchantments" than just the regular item's enchantments.
But if we were crafting legendary items like we used to ... we would have to take those two into consideration. And they are changing the outcome ... because they are acting like 2 more enchantments.
You can transfer Tier ... and you can also transfer Base Stats during the crafting process (talking about regular items not uniques).
So that was exactly what was affecting my results. I got 2 same out comes in few rolls but in fact they were not the same. The tier and the base stats were different. That means it is not a repetition.

And the effect on higher rarity would be much more devastating.
We are not paying attention to Tier and Base Stats because we are focused on unique items ... but they are having effect in the process crafting just like any other enchantment.
Well ... technically there is one more variable ... the item's level ... but since the level is getting averaged and we know the level beforehand we can nullify it.

That is why

Under the new crafting system with these new items there are 6 transferable stats on legendary items.
While with the old items there are 5. Old items are having no base stats (exception are the weapons and offhand items) while having T0 (regardless if the tier is displayed as T0-T3 the tier is T0) .

So ... there are 3 enchantments on Improved items and 6 on Legendary items under the current crafting.
And that is changing the odds drastically.

 

Yes, and no. They do affect the output item, but you are describing unique and independent variables (unless you can prove correlation between what you are talking about and the resulting transferred enchantments). Tier and base stats each operate independently from each other and from transferred enchantments, each with a 1/4 (25% chance) of being used on the final item.

That is only relevant if you are planning to use the resulting legendary. I can't think of anyone who will use a legendary over a unique. Sure, they may use the legendary for a bit while trying to get the unique they want to transfer the enchantments to or farm the cores, but since the ultimate goal is a full unique build, those two variable can be ignored for most cases.

Let me give you a practical example:


Crafting an Extraordinary Torso from 4x Magic Torsos.

There are 2x gold lines going into the mix (one on the first item, and one on the second item.) We want 2x gold lines on the Exo.

The first enchantment transferred has a 2/8 chance of being gold, and the second enchantment has a 1/7 chance of being a gold line (assuming the first one transferred). That is how we calculate the 2/56=1/28=3.57% chance of getting a two gold line exo... but lets go further.

There is a 1/4 chance that the base stats from item 1 transfer, a 1/4 that item two's base stats transfer, and so on.

There is also a 1/4 chance that any given item's tier transfers.

So, to get the tier from item 1 and the stats from item 1 has a 1/16 chance of happening. In fact, any pairing of item tier and item base stats has a 1/16 chance of happening and there are 16 possible pairings, all adding up to 100%.

All told, with your four "enchantments" in this example (Tier, Base, E1 and E2), there are 8*7*4*4=896 possible unique outcomes with 32 of those combinations having two gold enchantments and some combination of tier/base. Now, if we don't care about Tier/Base (like we usually don't), that reduces back down to 1/28 odds of getting a desirable two gold line outcome.

Growing up my parents had a saying; "A difference that makes no difference is not a difference at all." While you are right that these two variables affect the outcome of the craft, they don't affect it in any meaningful way for almost all crafting examples. Armand's chart is still accurate and relevant and this is nothing relevant if the user is planning to do pristine core crafting with the result.

Nothing in your post is relevant to modern crafting except in the incredibly unlikely chance that a user wants to use a legendary over a unique.

There are also gem slots, but those are averaged too, so they can also be ignored.

 

Incredibly unlikely is when you get same pattern:
Enchantment 1
Enchantment 2
Enchantment 3
Random enchantment

5 times in 20 rolls.
That is almost impossible.
And it doesn't ends there ... it happens every day over and over again.
It is incredibly impossible to be unable to craft an extraordinary to legendary item into a legendary item over 2 years ... knowing that you are trying every day with average of 20+ rolls.
That is why i can't craft a shield for over 2 years. I am not talking about leg to leg but exo to leg crafting ... i can't even proceed to leg to leg crafting

anyway ... i still stand behind what i said.
As for the PRNG (not related with crafting) ... I will show you how easily it can be manipulated ... but in private
And i am not talking about tampering with or changing game files or contents ... it only requires regular gameplay and a bit of knowledge what to do in particular moment. So it is not an exploit or anything ... just a regular gameplay.

Just like you, i have no time to make all those screens, taking notes, writing down patterns and analyzing rolls.
That requires an enormous amount of time ... which i don't have.
That is also one of the biggest pains when making a wiki article ... it is not that hard to make the article but the making of the content for the article requires enormous amount of time.
But i have a built in chip into my brain that collects all the data and performs all the calculations for me