Please upgrade your browser for the best possible experience.

Chrome Firefox Internet Explorer
×

Probabilities mean nothing - RE still needs fix

STAR WARS: The Old Republic > English > Crew Skills
Probabilities mean nothing - RE still needs fix

psandak's Avatar


psandak
03.27.2015 , 01:05 PM | #11
Quote: Originally Posted by AlrikFassbauer View Post
Problem is : Will I be able to endure that ? I mean REing "several thousand" items ?

At one point psychology strikes and makes me wan t to give up.
Others, meanwhile, are headstrong enough to keep on going.

Result : People are different. And thus they react differently on the same incidents.
All one really needs to do is track RE while leveling a crafting skill and will see several hundreds attempts and will get the 20% overall success rate. I know because I did it twice last year and around this time (optimizing my crew skills, I leveled Armormech on my Trooper and Armstech on my smuggler). And I've told the stories about it on this forum: On my AT I successfully REed two schematics on my first three attempts, I had numerous 20+ fail streaks one going 40+ across three different items (only got one superior schematic from the three). I still have the spreadsheet if you are interested.
The Ortzid Legacy on The Harbinger
Car'beerd (Guardian) Dalkery (Scoundrel) Blairnah (Sage) Daellia (Merc) and 11 other characters
Understanding Crew Skills / High End Schematics / Best Crew Skills / Crew Skill Money Making / Reverse Engineering Rules / SWTOR Story to date

asbalana's Avatar


asbalana
03.27.2015 , 02:40 PM | #12
Quote: Originally Posted by MSchuyler View Post
Your insistence on this is proof our science and mathematics education system is a failure. Probabilities mean EVERYTHING, your "experiments" nothwithstanding. You really need to take a statistics class and not let your emotions get in the way of your understanding. Modern science would not exist without probability. Indeed, that's what Quantum Mechanics is all about.

TL'DR: Working as intended.
Our education system must be a failure since every analysis that I've seen is based on the number of REs which are considered the number of trials. The actual number of trials is rather the number of items crafted for the RE attempt.

As an example, overnight a few days ago, I generated fifteen green armorings to RE to get a blue. I hit the blue on the first RE attempt and was left with fourteen worthless green items. In your sample and analysis, you would have one trial and one success. The correct way of looking at the situation is fifteen trials and one success. No one, especially at the highest level items which take forever to generate, is going to craft one item for one RE trial and if fail then craft another one.

Your analysis model looks at flipping a coin, noting the result, flipping the coin again and noting the result, and so on to infinity where everything just works out peachy keen. The actual model / process is filp a coin fifteen times and note if there is one success, flip the coin fifteen more times and note if there is a success, and so on to infinity where everything is peachy keen. The preceding differ in both analysis, math, and psychological impact.

So even the result of an education system that would not be considered a failure when applied incorrectly (wrong model) gives precise tight results that are incorrect and biased.

PS - I just went after a Hawkeye Implant and got it after 111 attempts. The first 110 tries were a failure (what are the odds of that?). Unfortunately model wise, I got pissed and crafted 20 of the blues overnight and then hit the Hawkeye the first thing the next morning. I was left with 19 of the blue suckers in my bag and have not sold (even though I aggressively undercut everyone on the GTN) one yet. I may just give up and RE them for scrap. How would we look at this. Did I have one success in 111 trials or one success in 130 trials? Do we add the 19 REs with 0 chance of success to our analysis?

CheaterLL's Avatar


CheaterLL
03.27.2015 , 03:00 PM | #13
Quote: Originally Posted by asbalana View Post
Our education system must be a failure since every analysis that I've seen is based on the number of REs which are considered the number of trials. The actual number of trials is rather the number of items crafted for the RE attempt.

As an example, overnight a few days ago, I generated fifteen green armorings to RE to get a blue. I hit the blue on the first RE attempt and was left with fourteen worthless green items. In your sample and analysis, you would have one trial and one success. The correct way of looking at the situation is fifteen trials and one success. No one, especially at the highest level items which take forever to generate, is going to craft one item for one RE trial and if fail then craft another one.

Your analysis model looks at flipping a coin, noting the result, flipping the coin again and noting the result, and so on to infinity where everything just works out peachy keen. The actual model / process is filp a coin fifteen times and note if there is one success, flip the coin fifteen more times and note if there is a success, and so on to infinity where everything is peachy keen. The preceding differ in both analysis, math, and psychological impact.

So even the result of an education system that would not be considered a failure when applied incorrectly (wrong model) gives precise tight results that are incorrect and biased.

PS - I just went after a Hawkeye Implant and got it after 111 attempts. The first 110 tries were a failure (what are the odds of that?). Unfortunately model wise, I got pissed and crafted 20 of the blues overnight and then hit the Hawkeye the first thing the next morning. I was left with 19 of the blue suckers in my bag and have not sold (even though I aggressively undercut everyone on the GTN) one yet. I may just give up and RE them for scrap. How would we look at this. Did I have one success in 111 trials or one success in 130 trials? Do we add the 19 REs with 0 chance of success to our analysis?
There is no "0 chance of success".
If you re something you can not learn anything from anymore you can still get the "you already know that schematic".
And the analysis has to be based on the number of res because that's how it works. RNGesus does not care how many stuff you have crafted.
CheaterLL, GM <you always wipe together>
Semi-Progression @ T3-M4
Juggernaut Tank/Sniper DPS/Sorcerer Heal

asbalana's Avatar


asbalana
03.27.2015 , 04:01 PM | #14
Quote: Originally Posted by CheaterLL View Post
There is no "0 chance of success".
If you re something you can not learn anything from anymore you can still get the "you already know that schematic".
And the analysis has to be based on the number of res because that's how it works. RNGesus does not care how many stuff you have crafted.
LOL, there indeed is a 0% chance of success. If you define success as obtaining a schematic as a result of an RE and that RE cannot give the schematic, then there is a 0% chance of success.

For example, if someone will pay me the first time I flip a coin and get a head and never thereafter then after I get the first head, I have a 50% chance of flipping a head the next time I try and a 0% chance of getting paid for it. One may well bet on the outcome of the coin tosses until they hit their first head, but would not bet on the toss after that although they have a 50% likelihood of hitting the head each subsequent trial.

One flaw in the system is that one is forced (because of crafting time and desire for a schematic that is currently needed) to engage in more tirals than necessary since one would stop after obtaining the desired schematic. If I told you that if you can place a 25 cent bet and collect a dollar if you can toss a head, you would pop down the money. If I told you that you had to flip the coin ten times and on each flip put down 25 cents and I would only pay you a dollar the first time you hit a head, you would not take the bet. The SWTOR system forces you to make more bets than would otherwise be indicated in a sequential trial system.

Never forget that logic is part of math as well as probability and bad logic makes bad probability.

The system (SWTOR) and analysis is flawed in that it does not consider trials that will occur with a 0% (yes I said it again) chance of success.

psandak's Avatar


psandak
03.27.2015 , 06:17 PM | #15
Quote: Originally Posted by asbalana View Post
One flaw in the system is that one is forced (because of crafting time and desire for a schematic that is currently needed) to engage in more tirals than necessary since one would stop after obtaining the desired schematic. If I told you that if you can place a 25 cent bet and collect a dollar if you can toss a head, you would pop down the money. If I told you that you had to flip the coin ten times and on each flip put down 25 cents and I would only pay you a dollar the first time you hit a head, you would not take the bet. The SWTOR system forces you to make more bets than would otherwise be indicated in a sequential trial system.

Never forget that logic is part of math as well as probability and bad logic makes bad probability.

The system (SWTOR) and analysis is flawed in that it does not consider trials that will occur with a 0% (yes I said it again) chance of success.
The easy solution to that "flaw" is small batch crafting. With greens, I will FREQUENTLY sit and wait, REing as they come in, because they only take 5 minutes to craft (send five or six companions to craft stagger them so that I can RE as they come in without risking another coming in too soon, and then cancel the rest when I get the schematic(s) I want). With blues, I will craft - at most - five per purple available. And if by some chance I get the schematic I want quickly...BLUES SELL WELL!

It is called knowing the system and working within it to its potential. If one crafts 15 knowing that they will likely get the schematic they want within 5, whose fault is that really?
The Ortzid Legacy on The Harbinger
Car'beerd (Guardian) Dalkery (Scoundrel) Blairnah (Sage) Daellia (Merc) and 11 other characters
Understanding Crew Skills / High End Schematics / Best Crew Skills / Crew Skill Money Making / Reverse Engineering Rules / SWTOR Story to date

Nazdika's Avatar


Nazdika
03.27.2015 , 06:38 PM | #16
Quote: Originally Posted by asbalana View Post
No one, especially at the highest level items which take forever to generate, is going to craft one item for one RE trial and if fail then craft another one.
The term for this fallacy is: consensus over-estimation. Just in case you are interested to know I do exactly that, most times: craft one, RE, craft one, RE. Except I have stacks of mats, and am really lazy. Then I might craft two, or even three *gasp* I personally do know several crafters, who craft only one piece. And others who don't.

Of course only the ones that you actually RE with a chance at success count for the probability of... a success in RE'ing Up to you to ruin mats or not. Not like you can buy tickets from a lottery that is over, and complain that you did not win. Since you clearly know whether the lottery is over, you cannot even complain that someone sold you an outdated ticket. Except to yourself

In general:
Rule of thumb in science, and market research is 2k tries show the true distribution with solid certainty. It is a curve shaped quite like the DR curves with a soft-cap. Past 2k, additional tries do not add too much to the credibility of the data anymore, statistically speaking. Lower levels as psandak wrote, several hundreds to make a decent guess about a pattern. Below that, probabilities do indeed not mean much. (Tip: Read psandak's intro to probability. Free of charge and good )

That just means, when I do 2k REs (and I am sure I am WAY and WAY beyond that in 2 years of active crafting), success rate will be close to the true rate. Tooltip says it is 20%. If the rng is set as that, I will expect the distribution to be really close to that. The fact, that statistics is strongly counter-intuitive to the human mind has been shown over and over in the respective sciences, though. I cut everyone a lot of slack here.

Avinler's Avatar


Avinler
03.27.2015 , 10:23 PM | #17
My crafting defies the assertion that everyone approaches this mechanic the same. The only time I would ever craft more than 2-3 (ideally 1) of the same item is if I can still learn 2-3 schematics via RE of that item.

Too many people are RE'ing toward a single target which biases their success rate downward because they don't count the RE successes that yield a schematic they don't want, but that's your preference not a failure of probabilities.
Avin

Founder, Retired GM of <Fortitude> on Jedi Covenant
& Andrew on SWTOR Economics

Sasaatan's Avatar


Sasaatan
03.28.2015 , 04:45 AM | #18
This is what a game is about - making decisions. In RE case you choose your batch size based on what you value more - the efficiency of not crafting anything which may be proven as unnecessary or the time-cost of working on batches. using batches of 15 is hopefully hyperbole. I personally never go above 2-3, usually just micromanage with 1.

One decision is not to RE at all and go to the market - most time saved which could be used to earn more money to by next items. If you don't like meddling with RE, just don't and go kill stuff instead. Definitely we do NOT need to have the RE subgame even more simple and straightforward as it is.

asbalana's Avatar


asbalana
03.28.2015 , 08:42 AM | #19
Quote: Originally Posted by Nazdika View Post
The term for this fallacy is: consensus over-estimation. Just in case you are interested to know I do exactly that, most times: craft one, RE, craft one, RE. Except I have stacks of mats, and am really lazy. Then I might craft two, or even three *gasp* I personally do know several crafters, who craft only one piece. And others who don't.

Of course only the ones that you actually RE with a chance at success count for the probability of... a success in RE'ing Up to you to ruin mats or not. Not like you can buy tickets from a lottery that is over, and complain that you did not win. Since you clearly know whether the lottery is over, you cannot even complain that someone sold you an outdated ticket. Except to yourself

In general:
Rule of thumb in science, and market research is 2k tries show the true distribution with solid certainty. It is a curve shaped quite like the DR curves with a soft-cap. Past 2k, additional tries do not add too much to the credibility of the data anymore, statistically speaking. Lower levels as psandak wrote, several hundreds to make a decent guess about a pattern. Below that, probabilities do indeed not mean much. (Tip: Read psandak's intro to probability. Free of charge and good )

That just means, when I do 2k REs (and I am sure I am WAY and WAY beyond that in 2 years of active crafting), success rate will be close to the true rate. Tooltip says it is 20%. If the rng is set as that, I will expect the distribution to be really close to that. The fact, that statistics is strongly counter-intuitive to the human mind has been shown over and over in the respective sciences, though. I cut everyone a lot of slack here.
Good morning.

Your post (and the two following) was an eye opener for me.

My original point was simply that the sample size is the number of items crafted to be REed and not the number of trials needed to attain success. I wrote it in response to a prior post that was insulting to another forum member by referring to a failure of the education system.

Your post also makes two comments that may be construed as less than polite. Neither addressed the point.

First, "census over-estimation" is a dismissive response. Considering that you went on to talk about 2K tries showing true but based the response on your and the input of "several" others, I would think that you are short 1,990+ opinions.

Second, your Tip to read psandak's intro to probability, was less a tip and more an insult. So let's compare and see whose is bigger. I have a doctorate in mathamatics (pure not applied) and am an actuary (FSA, MAAA, MSPA, and EA). I am self employed and work as a consultant doing mathematical, legal, and administrative analysis and studies each day. I have clients throughout the country. What are your qualifications?

I should not have said that ALL people generate more than one trial at a time. I should have qualified that statement. I do suffer from a different mental disorder; confirmation bias. My life, college, grad school, business, sports, etc., has been spent among goal oriented people. Although I believe that one should enjoy the journey and exert reasonable effort, one does what it takes to get the job done within time constraints. Everyone (no need to qualify that) in "my universe" has the same mind set.

To you and the two who posted after you, I would say that I do forget that there are those who take a different approach since it is so foreign to my way of thinking. There is nothing wrong with the doing one at a time approach, but to me it just does not compute. We will never know, but it would be interesting to see how people approach the "get a schematic" process.

I have both more credits and more mats than I will ever really need. I can probably play the game without slow down for years without earning another credit or obtaining another mat. I indeed usually generate ten to fifteen items to RE, when I go after a schematic. I indeed go after schematics that I want and need. Generating schematics by producing and REing one item at a time, to my mind, seems to be a long, tedious, and drawn out process which is likely to produce the desired result way after I may need it. Mats and credites are cheap and easy, time is dear.

I have always played dps characters. To introduce variety and complement a character being played by my wife, I started leveling a Shadow Tank. I had no schematics with defensive stats when I did. My shadow is now 47 and he and Treek are fully geared (blues, purples, and even some augments) at level. I already have the schematics at purple up to level 56 for him. If I had taken the one at a time approach, where would I be now? Still trying for level 30 stuff? As an aside, I am loving the Shadow tank and somewhere down the line will compliment him with a Sith Assassin.

To each, his/her own.

eartharioch's Avatar


eartharioch
03.28.2015 , 01:14 PM | #20
Quote: Originally Posted by asbalana View Post
LOL, there indeed is a 0% chance of success. If you define success as obtaining a schematic as a result of an RE and that RE cannot give the schematic, then there is a 0% chance of success.

For example, if someone will pay me the first time I flip a coin and get a head and never thereafter then after I get the first head, I have a 50% chance of flipping a head the next time I try and a 0% chance of getting paid for it. One may well bet on the outcome of the coin tosses until they hit their first head, but would not bet on the toss after that although they have a 50% likelihood of hitting the head each subsequent trial.

One flaw in the system is that one is forced (because of crafting time and desire for a schematic that is currently needed) to engage in more tirals than necessary since one would stop after obtaining the desired schematic. If I told you that if you can place a 25 cent bet and collect a dollar if you can toss a head, you would pop down the money. If I told you that you had to flip the coin ten times and on each flip put down 25 cents and I would only pay you a dollar the first time you hit a head, you would not take the bet. The SWTOR system forces you to make more bets than would otherwise be indicated in a sequential trial system.

Never forget that logic is part of math as well as probability and bad logic makes bad probability.

The system (SWTOR) and analysis is flawed in that it does not consider trials that will occur with a 0% (yes I said it again) chance of success.
It's not swtor's flaw, it's yours. If you want only one schematic, and you choose to craft more than one item, you are accepting the "risk" that one or more of the items you craft will be unsuitable for the purposes of reverse engineering for a schematic. You have to judge, on an item by item basis, whether the material cost, suitability for other purposes (such as sale on the GTN), time sensitivity of acquisition of the desired schematic, etc. justify the production of more units than are strictly necessary. Just like if a mob has low hp, you can choose between using an expensive attack that may do much more damage than necessary or a cheaper ability that may barely be able to kill it if you are really lucky. It's called playing the game.

The problem with your logic (which I'll call "Asbalana's Fallacy", since I don't recall seeing anybody make this mistake before) is that you are not looking at the probabilities correctly. If an item has a 20% chance to RE, it has an 80% chance to "fail". If you RE 15 items, you have a 0.8 exp 15 (~3.5%) chance to fail, or a 96.5% chance to succeed. However, that 96.5% chance includes every possible outcome (other than all failures), so succeeding on one or more attempts before RE'ing all fifteen items is not only possible, but quite likely.

In fact, rather than complain about the system, you should try to appreciate its benefits. Rather than just pay for all 15 items at once and then make one roll with a 96.5% chance, you pay for the 15 items up front but then only use as many is necessary to get the schematic, and you have the unused items to RE just for mats, sell to vendor, sell to other players, or use for you own characters, thus defraying your initial investment.

A few comments:

1) If you are trying to get a specific schematic just to make a single item for immediate use, it will probably be more cost efficient just to buy the item on the GTN.

2) If you are looking to get a specific schematic to make a single item for immediate use and future sales, your initial investment will be higher, but if you picked a profitable item, you should be able to recover that investment in a reasonable period of time (and if you factor schematic acquisition cost in your profitability calculations, the fixed initial investment cost will tend towards zero with each [profitable] item you sell.

3) If you are a completionist just trying to get every schematic, you can reduce the cost to acquire by simply making one at a time.

4) If you are just looking to make money (i.e., you don't need to make the item for immediate use), you can just decide for yourself each time you go to craft items to RE if you think that the cost of making more than one item is or is not justified by the potential reduction of time-to-market.

5) The vast majority of "extra" items can be sold for at least the cost to create regardless of whether or not you'd actually want to craft them *just* to sell.
Q: What did the five fingers say to the face?
A: It's a SLAP!
This is not my referral link