I have RE'd pretty much with good luck every other 168 schem on my Artifice without an issue, and within the 20% success rate. But then I tried to RE the Advanced Russan Relic of Focused Retribution at least 30 times, and over the course of several logins and days, and nada. It's probably closer to 45 times. I looked on the GTN and don't see any purple ones there, so I'm wondering if I have worst luck ever, or this thing is broken. Anyone successfully RE one of these?

Yup on Shadowlands and on Bergeren Colonoy with no issue.

RNG is RNG and one of the many flaws in the crafting system is that you can have long runs of bad luck. I've had runs that were less likely than winning the Power Ball a number of times. Haven't won the Power Ball yet.

Edit - I am on now and see three pages of purple on the GTN on Shadowlands and none on Bergeren. Funny how the different servers vary on what one can find on the GTN and on pricing and sales on the GTN.

Yeah I'm on Bergeron, so that's good news. Okay. Back to vendor to get that freaking zoosha solution. Thanks.

You do know you can get zoosha solution from Archaeology crew missions, right? The only problem is that they don't always show up in the list, so you might have to zone-jump a few times.

Yeah I'm on Bergeron, so that's good news. Okay. Back to vendor to get that freaking zoosha solution. Thanks.

Let me know your character's name or send me a mail in game (Smith-tsin) and I will send you a stack of the stuff. It is too expensive to buy from a vendor and makes REing painfully expensive..

As noted, you can run Arch missions to get it and when Smith is crusing Yavin 4 gathering I always run the Vanadium Flux and Zoosha missions in the background so that I will have a rich supply.

I have RE'd pretty much with good luck every other 168 schem on my Artifice without an issue, and within the 20% success rate. But then I tried to RE the Advanced Russan Relic of Focused Retribution at least 30 times, and over the course of several logins and days, and nada. It's probably closer to 45 times. I looked on the GTN and don't see any purple ones there, so I'm wondering if I have worst luck ever, or this thing is broken. Anyone successfully RE one of these?

As Asbalana points out:

RNG is RNG and one of the many flaws in the crafting system is that you can have long runs of bad luck.

I do not consider it a flaw, but that is different debate...one that I will not get into here.

I too had trouble with one of the new relics too but mine was ephemeral mending. I actually stopped trying for a while -moving on to a different relic. I eventually got it. Same thing happened with trying to get the blue Anodyne reflex stim (took 21 tries to get from green to blue), and a myriad of other items over the past three years.

I succeeded on my first attempt to RE a 180 piece, and critted the first time I made it. So every time I go for a long string of failures, I just think about that.

I do the same thing. I remember releveling armstech (for optimization) - the first two of three REs I did in that process generated schematics. Granted grade 2 blue barrels, is not all that big a deal, but later on in that process when I had a string of 41 failures (across three items), I thought about that 2 for 3 to start and said to myself "well this balances against that. Oh well."

I think a good feature for RE'ing is if you don't get a new schematic from the RE and its possible, the chance for the next one is increased to 40%, then 60%, then 80%, then to 100% chance of getting a new schematic, then after it succeds, it resets back down to 20% again.

I think its a good idea if you ask me.

I think a good feature for RE'ing is if you don't get a new schematic from the RE and its possible, the chance for the next one is increased to 40%, then 60%, then 80%, then to 100% chance of getting a new schematic, then after it succeds, it resets back down to 20% again.

 

I think its a good idea if you ask me.

You know what happens when you do that though? The rate of success drops to 1 in 2.5 (that is probably what you really want ). Obviously this is a non-starter because the devs have decided that the rate should be 1 in 5 over the VERY long haul. Which in fact it does.

In order to put this argument to rest I created a simulation using Excel. It runs through the process of reverse engineering 500 "unique" (albeit generic) items. Running many sets of data, 91% of the time success comes in 10 attempts or fewer, and 3/4 of that are in the first 5 attempts. The point is that long failures streaks are overemphasized; in the grand scheme they are insignificant blips on the radar.

That being said, an incremental increase per failure is not a bad idea, but let's assume that the overall average result has to fall very near 20%. This means that a base plus increment system cannot start at 20%. So I experimented a little. Suffice it to say that the combination of starting point and increment that I came out with is 10% and +4%. This combo basically caps the failures to 15 in a row. The question is would the playerbase rather have:

the RE system as it is now, where a simple majority of the time (~51%) RE success is in the first three attempts but there is a relatively small chance that a very long streak of failure will occur (~3.5% of 20+ failures)

OR

A RE system where the longest possible failure streak is in the teens, but a plurality (~45%) of the time it will take 3 to 5 attempts to succeed.

Now remember this is only theory. There is absolutely no indication that the developers would do this, and chances are they would not because it is a more complex system with more potential points of failure. But this was an interesting experiment nonetheless.

I think a good feature for RE'ing is if you don't get a new schematic from the RE and its possible, the chance for the next one is increased to 40%, then 60%, then 80%, then to 100% chance of getting a new schematic, then after it succeds, it resets back down to 20% again.

 

I think its a good idea if you ask me.

Nah, it's a massive buff.

On average right now you have to RE five times. They could keep that average but have it ramp up, but that would mean it would start lower than 20%, and then get higher. It obviously would never go to 100%.

WoW uses systems like this on quests. It reduces the number of "oh, three things, three kills", while eliminating the "been on two for an hour" thing.

.......... The point is that long failures streaks are overemphasized; in the grand scheme they are insignificant blips on the radar............

 

.

Simply, that statement is both opinion and dead wrong.

I (as most) have had my share of long long streaks of fails that although possible are statistically improbable.

1. Long streaks tend to be demotivational and have a psychological impact on players. One central concept in MMO design is to provide a rewarding and enjoyable experience for players. There should be a balance between challenge and reward. The only thing that I can say with certainty is that the uncertain will surely happen and unfortunately happen all to often. The RE system is not fun. A hit / successful RE often only ends up providing a sense of relief from a current blip or remembered debacle rather than a sense of accomplishment.

2. Your experiment is dead wrong in concept. If I flip a five sided coin I would expect the results that you identified. I flip, pick up the coin, flip again, pick up the coin, etc. In SWTOR RE crafting I do not craft an item, RE, craft another, RE, and then craft again. There are no sequential quntized trials occurring. I craft ten items and then RE. No one is going to sit around and wait 15 / 20 / 30 minutes for an item to be crafted to try an RE. So even if I hit on the first try, I still have to RE 9 other now worthless items to get the mats back. In some cases this has value if I am getting desired kit components or can sell the unneeded items, but still I have to run ten items to attempt to realize a one in five chance.

3. We are happy campers, you and I. We are rich (credits out the wazoo) and cargo holds full of mats. If we want to RE something, we take a ten minute work break and send out the troops. To a new player, a poor player, or one with few mats, not so much. Your blip can bankrupt them. Attempting to get equipment to progress on can stall out their game during the blip. This system works strongly against the newer and poorer players, just the ones that you want to attract to the game.

4. I want a Hawkey earpiece so I am going to RE a 20% green to start my journey. By the time I get done with Crit, Redoubt, and what not, I have to go through the RE process not five or ten times but many many unless I get lucky. Each time I end up with a eapiece with stats that I don't want, I have to start over again. Strange thing about that. Each time I restart, I have the same chance of a fail streak or as you would call it a "blip". It has always amazed me how after building about every possible craftable item that I am left with the perception that the Hawkeys always come last in the process. What are the odds of that happening? Retorical question, don't want an answer. But whatever your answer would be or your math will tell you (see 1. above), the process leaves a bad taste and sense of relief when completed and little or no feeling of accomplishment or victory.

In a nutshell, a little RNG is good and too much is terrible. SWTOR crafting and other aspects has too much.

I agree with you that long failure streaks can be very discouraging. I just wanted to point this out, though:

It has always amazed me how after building about every possible craftable item that I am left with the perception that the Hawkeys always come last in the process.

If you're REing specifically for Hawkeyes, and you stop after you crit one rather than opening every branch, naturally they're always going to be the last ones in the process. Kind of like your keys are always the last place you look -- because that's when you stop looking.

If you find yourself opening every other RE branch and Hawkeyes still appear last, then yeah -- RNG sucks sometimes. My luck with it tends to be very up or very down; sometimes I'll crit a Hawkeye right away, and other times it will take me ages. If the fail strings get long enough and frequent enough in a single crafting session, I tend to just step away and do something else for a bit.

1. You might be surprised but I agree, but it is the way it is and it ain't gonna change.

2. wrong. If you RE something that has "no research available" you can still get a "success" the system just tells you "you already have that schematic." translation to my system: those are two unique "items". Same thing with green to three blue or blue to purple. My system does not care that those three "generic items" are in fact the same "specific item." All it knows is that I want item X to generate ANY schematic when I RE it. The chances of getting a SPECIFIC schematic from green to purple is an entirely different set of calculations

3. And that is why I recommend that new players NOT take crafting skills. Crafting is a significant front end investment, and new/poor players cannot afford to make that investment. Believe me, I KNOW what you are talking about because I made the mistake when I started playing: there were times I had to not advance class abilities because I did not have enough credits. I did not get piloting III until WELL into 50 (the level cap at the time; and the skill cost 100k).

4.

Strange thing about that. Each time I restart, I have the same chance of a fail streak or as you would call it a "blip".

yes the same 3.5% chance at a 15 fail streak.

yes the same 1.1% chance at a 20 fail streak

versus the 20% chance to get it on the first try

or even the 20.97% chance you get it within the first six tries

You've made my point. You are focusing on the small chance of negative result instead of the 7 to 20 times more likely positive result.

2. wrong. If you RE something that has "no research available" you can still get a "success" the system just tells you "you already have that schematic." translation to my system: those are two unique "items". Same thing with green to three blue or blue to purple. My system does not care that those three "generic items" are in fact the same "specific item." All it knows is that I want item X to generate ANY schematic when I RE it. The chances of getting a SPECIFIC schematic from green to purple is an entirely different set of calculations

IIRC, at launch, if you learned a schematic, you had a 1/n (n=possible blues/purples from that green/blue) of getting an particular version, and if you had already learned it, you got the "you already know this schematic" message saying you just critted for no effect. Now, at least if you learn one schematic you didn't want, it doesn't count against further RE attempts.

IIRC, at launch, if you learned a schematic, you had a 1/n (n=possible blues/purples from that green/blue) of getting an particular version, and if you had already learned it, you got the "you already know this schematic" message saying you just critted for no effect. Now, at least if you learn one schematic you didn't want, it doesn't count against further RE attempts.

I distinctly remember getting the "You already know that schematic," several times when I had crafted green and blue relics on my Artificer, and armorings and mods on my CT. I got the schematic and REed the extras after failing to sell them or they were not worth selling at all. Maybe it's a bug with the new stuff.

That being said, even if I mis-remember, my simulation assumes items to RE simply exist - no crafting involved. No leftovers and no extras from crit crafting. Look at it this way: it represents REing top end items; items only available through comms, tokens, and others crafting them. Most of the time, players will acquire one at a time and RE it and then wait for the next opportunity, and when they get the schematic they stop REing that item.

I know what you will say next: that I am trying to justify what I did. And to an extent you're right. But I challenge anyone to create a better/more accurate simulation without programming skills.

I distinctly remember getting the "You already know that schematic," several times when I had crafted green and blue relics on my Artificer, and armorings and mods on my CT. I got the schematic and REed the extras after failing to sell them or they were not worth selling at all. Maybe it's a bug with the new stuff.

I meant that at launch, if you were (say) RE'ing a green with three possible blues, and you learned one, and kept RE'ing (because it wasn't the one you wanted), the next time you successfully RE'd, you'd have a 1 in 3 chance that the schematic you "learned" was the one you had already learned, so you'd get the "error message". That was changed to what is now -- if you successfully RE something, you will get a schematic you don't know *if one is available*. If you RE with no new schematics available (and I do this as well if the mats are worth more than somebody will pay for the green/blue), the system still checks to see if you would have RE'd something, but since there are no new schematics, it gives you the message.

That being said, even if I mis-remember, my simulation assumes items to RE simply exist - no crafting involved. No leftovers and no extras from crit crafting. Look at it this way: it represents REing top end items; items only available through comms, tokens, and others crafting them. Most of the time, players will acquire one at a time and RE it and then wait for the next opportunity, and when they get the schematic they stop REing that item.

 

I know what you will say next: that I am trying to justify what I did. And to an extent you're right. But I challenge anyone to create a better/more accurate simulation without programming skills.

Actually, I wasn't looking to dis your model*, I was trying to point out that things could be (and were) worse. While your model may have some flaws (I didn't look at it closely enough to judge) that affect the *quantitative* results (the actual numbers), it does a good enough job of illustrating the *qualitative* point you were trying to get across -- namely, that if the devs want a long-term 20% success rate, and players want the chance to increase [monotonically], you have to start with a value less than 20%, which means that you have a lower chance to RE in fewer tries.

Perhaps an easier way to say this would be that the harder you make it to take more than 5 tries, the harder it will be to get it in fewer than 5 tries; the only way to make it never take more than five tries is to make it never take fewer. While (in theory) I'd favor a guaranteed "RE 5x and get success guaranteed" system, I don't think most people would like the fact that it would likely be on a per-item basis (and thus introduce an enormous amount of data to be tracked) to avoid people REing four X Barrel 2s and then the Advanced X Barrel 37. And any system that didn't track on a per-item basis would be subject to such abuse, even if it wasn't a strict 4 fail 1 success pattern.

* I could, if I wanted, but you successfully disqualified me with the "no programming skills" requirement, although I'd argue that it's math skills, not programming skills, that determine that validity of a model

I meant that at launch, if you were (say) RE'ing a green with three possible blues, and you learned one, and kept RE'ing (because it wasn't the one you wanted), the next time you successfully RE'd, you'd have a 1 in 3 chance that the schematic you "learned" was the one you had already learned, so you'd get the "error message". That was changed to what is now -- if you successfully RE something, you will get a schematic you don't know *if one is available*. If you RE with no new schematics available (and I do this as well if the mats are worth more than somebody will pay for the green/blue), the system still checks to see if you would have RE'd something, but since there are no new schematics, it gives you the message.

Ah, ok, my mistake.

Actually, I wasn't looking to dis your model*, I was trying to point out that things could be (and were) worse. While your model may have some flaws (I didn't look at it closely enough to judge) that affect the *quantitative* results (the actual numbers), it does a good enough job of illustrating the *qualitative* point you were trying to get across -- namely, that if the devs want a long-term 20% success rate, and players want the chance to increase [monotonically], you have to start with a value less than 20%, which means that you have a lower chance to RE in fewer tries.

Thank you. Asbalana snipped one piece of my post and focused on that. Then I made the mistake of focusing on Asbalana's focus And then I made the false assumption that you were siding with Asbalana. :o

Perhaps an easier way to say this would be that the harder you make it to take more than 5 tries, the harder it will be to get it in fewer than 5 tries; the only way to make it never take more than five tries is to make it never take fewer. While (in theory) I'd favor a guaranteed "RE 5x and get success guaranteed" system, I don't think most people would like the fact that it would likely be on a per-item basis (and thus introduce an enormous amount of data to be tracked) to avoid people REing four X Barrel 2s and then the Advanced X Barrel 37. And any system that didn't track on a per-item basis would be subject to such abuse, even if it wasn't a strict 4 fail 1 success pattern.

Well the solution to that would be to have a series of vendors: craft five of one green and turn them into a vendor for a box that will have one of the three possible blue schematics...same for blue to purple. Or make it even simpler: craft 15 green and turn them in for a box with all three blue derived from that green. Only problem is that's a LOT new vendor items; EVERY green gear schematic learned at the trainer would have to have a blue box and then 3 purple boxes (1 for each blue).

Ah, ok, my mistake.

Thank you. Asbalana snipped one piece of my post and focused on that. Then I made the mistake of focusing on Asbalana's focus And then I made the false assumption that you were siding with Asbalana. :o

FYI, for future reference, I'm unlikely to ever side with people that say things like, "I (as most) have had my share of long long streaks of fails that although possible are statistically improbable." Just sayin'

Well the solution to that would be to have a series of vendors: craft five of one green and turn them into a vendor for a box that will have one of the three possible blue schematics...same for blue to purple. Or make it even simpler: craft 15 green and turn them in for a box with all three blue derived from that green. Only problem is that's a LOT new vendor items; EVERY green gear schematic learned at the trainer would have to have a blue box and then 3 purple boxes (1 for each blue).

An easier solution would be to have a dialog box into which you transfer 5* identical items and then click an "RE" button. They could just make it a flat 20% per item chance to RE for the gambling types (so one item gives you a 20% chance, two items gives you a 40% chance, etc.). I would probably always do 5x for a guaranteed RE on regular items because the predictability would outweigh the random time and cost to get a desired item, but the time to acquire end game (192, and hopefully soon to be 198) items would make waiting for 5x to guarantee the RE even more painful than the RNG is now.

* This would be a good time to change the few 10% RE chance implants to 20% for consistency

Statistically, if we can assume that the distribution curve is a standard normal bell curve, then we can extrapolate some expectations for the results of RE'ing an item.

Average chance : 0.2 (20%)

Standard Deviation: 0.0816

Within 1 STD, we'd expect a 68% chance that you'd see results within the following ranges:

Min: 0.1183 (11.85%)

Max: 0.2816 (28.16%)

This would translate to somewhere between 3 and 8 attempts in order for a RE to be successful.

Within 2 STD, we'd expect a 95% confidence that our results should be between the following ranges:

Min: 0.0367 (3.67%)

Max: 0.3633 (36.33%)

This would translate to somewhere between 2 and 27 tries to get a new schematic to proc for you.

Of course there could still be outliers, but needing more than 30 attempts to get a result should occur less than 5% of the time.

The calculation gets somewhat more complicated when you consider things like the Tier 11 implants where you have 3 possible blue results which in turn result in 3 possible purples (each). If your goal is to get a specific purple, then starting from the green schematic, it will take a lot more time.

The time calculation (which has also entered this conversation) is another factor which (I believe) is more significant in the player's satisfaction level when learning new schematics.

Statistically, if we can assume that the distribution curve is a standard normal bell curve, then we can extrapolate some expectations for the results of RE'ing an item.

 

Average chance : 0.2 (20%)

Standard Deviation: 0.0816

 

Within 1 STD, we'd expect a 68% chance that you'd see results within the following ranges:

Min: 0.1183 (11.85%)

Max: 0.2816 (28.16%)

This would translate to somewhere between 3 and 8 attempts in order for a RE to be successful.

 

Within 2 STD, we'd expect a 95% confidence that our results should be between the following ranges:

Min: 0.0367 (3.67%)

Max: 0.3633 (36.33%)

This would translate to somewhere between 2 and 27 tries to get a new schematic to proc for you.

 

Of course there could still be outliers, but needing more than 30 attempts to get a result should occur less than 5% of the time.

 

The calculation gets somewhat more complicated when you consider things like the Tier 11 implants where you have 3 possible blue results which in turn result in 3 possible purples (each). If your goal is to get a specific purple, then starting from the green schematic, it will take a lot more time.

 

The time calculation (which has also entered this conversation) is another factor which (I believe) is more significant in the player's satisfaction level when learning new schematics.

Your distribution assumption is incorrect. The actual distribution is a negative exponential curve.

Your distribution assumption is incorrect. The actual distribution is a negative exponential curve.

We know this for sure? Or is this based on sampling data?

........

 

Thank you. Asbalana snipped one piece of my post and focused on that. Then I made the mistake of focusing on Asbalana's focus And then I made the false assumption that you were siding with Asbalana. :o

 

.

Oops did it again. Since the snippet was a conclusion and springboard for the rest of your statement and based on a false model, I thought it was the right point of response.

Simply, a little RNG is good, but all RNG is not. It is demotivational, contrary to quality game design, and a cheap and lazy way to go. One can talk statistics until the cows come home (I have a Phd in Mathematics although I confess pure math not applied) but in the end the human experience is what matters and 100% RNG is not quality experience.

We know this for sure? Or is this based on sampling data?

Based on my sampling but backed up by the probability of continued failure.

That probability of failure starts high: 80%. Follow the track from 1 failure to 30 and you get a negative exponential curve:

80.00%

64.00%

51.20%

40.96%

32.77%

26.21%

20.97%

16.78%

13.42%

10.74%

8.59%

6.87%

5.50%

4.40%

3.52%

2.81%

2.25%

1.80%

1.44%

1.15%

0.92%

0.74%

0.59%

0.47%

0.38%

0.30%

0.24%

0.19%

0.15%

0.12%