Please upgrade your browser for the best possible experience.

Chrome Firefox Internet Explorer
×

new problem with Reverse Engineering?

STAR WARS: The Old Republic > English > Crew Skills
new problem with Reverse Engineering?

Khevar's Avatar


Khevar
06.19.2012 , 10:47 AM | #11
Quote: Originally Posted by wneubert View Post
Most things seem to be working correctly but there are a few clear and repeatable exceptions. The ones I have noticed are:

Artificer:Wieldable tree - which includes the focus mentioned above
Armstech:Wieldable tree - most of my experience is with Vibroknife
Cybertech:Earpiece tree
Emphasis added by me.

This is actually just a bad streak on your part. I have an artificer, armstech and cybertech. I've recently RE'ed to purple a level 48 focus, a level 50 vibroknife and a level 47 earpiece. The progression was typical, with no extensive bad streaks.

Any collection of stats that "prove" the 20%/10% is wrong is not taking into account the many other crafters who are not contributing their stats to the discussion that would balance everything out.

Honestly, anyone who gets a good streak (I once got 3 10% purples IN-A-ROW) seldom come to the forum to say "hey, the 10% is broken I'm getting RE's too fast"

Sothicus's Avatar


Sothicus
06.20.2012 , 04:10 PM | #12
Quote: Originally Posted by JeffKretz View Post
Emphasis added by me.

Any collection of stats that "prove" the 20%/10% is wrong is not taking into account the many other crafters who are not contributing their stats to the discussion that would balance everything out.
I'm sorry, but this is not correct. You do not need to take into account any other crafter. For a given statistical test, you only need one person's attempts in order to determine if things are working as intended statistically. Recently a Las Vegas roulette wheel hit 19 7 times in a row, then 15, then 19 again. You can be sure that roulette wheel has been taken down and examined. You aren't going to be able to convince anyone that they should take into account all other wheels as those eight 19's in nine spins would disappear in the wash.

Outliers are the bane of any statistic and most people want to sweep them under the rug, usually by grouping them into larger sets and showing that they are a small and insignificant number. But this is wrong as the 20/10% are intended for the individual, not the larger set of all people.

It is clear that there may be some issues with RE'ing as some testing by some people has indicated. However, those issues may be as simple as BW not explaining the math behind the crafting.

I should be even more clear on this for my case:

As an artificer, I find that the 2nd and 3rd blue of a given green wieldable item (the foci, generators and shields) seem to be on the 10% RE line instead of 20% although I have yet to learn a 2nd version of any of my wieldable items. Since I have limited resources, I have only been able to do about 20-25 of several different wieldable items. If the tool tip is indeed wrong and it is 10% once one is learned, then the math works out much more correctly.

Khevar's Avatar


Khevar
06.21.2012 , 10:15 AM | #13
Quote: Originally Posted by Sothicus View Post
As an artificer, I find that the 2nd and 3rd blue of a given green wieldable item (the foci, generators and shields) seem to be on the 10% RE line instead of 20% although I have yet to learn a 2nd version of any of my wieldable items. Since I have limited resources, I have only been able to do about 20-25 of several different wieldable items. If the tool tip is indeed wrong and it is 10% once one is learned, then the math works out much more correctly.
This is an interesting point.

You're saying that if you RE the first green to blue, you've experienced the 20%, but trying again for another prefix you've experienced a 10%?

I don't have enough actual collected statistics to make any statement one way or the other. I do recall occasionally taking a long time to get a second or third blue. On the other hand I also recall RE'ing 5 greens and getting all three blues in the same set of 5.

You may be right, but I don't think so. I have 5 of the 6 crafting professions to 400. I've unlocked many purple schematics for alts, companions and guildies. When I was actually tracking statistics, I had:

691 greens RE'ed - 132 successful
316 blues REed - 35 successful

These were close enough to the 20%/10% that I stopped counting.

Sothicus's Avatar


Sothicus
06.21.2012 , 11:55 AM | #14
That's correct. I have to admit that due to my limited resources I can only go off of the 60+ RE's I've attempted so far with no success for a second blue of a different trait. But at 20%, that is very unlikely... about 1 in 100,000 chance in fact. So there is something else underneath the math I think... if at 10%, I'm at 1 in 1000 which is not great either but I have a little less heartburn over.

Even so, if most of my greens pop a blue within the expected 1 to 10 tries (i.e. a reasonable curve) and everyone's second blues and even puprle's seem to come up on extreme ends... I think there is a problem with the success curve (again it all boils down to math which BW won't tell us).

Khevar's Avatar


Khevar
06.21.2012 , 04:15 PM | #15
Anectodal story incoming. It may be boring so feel free to ignore.

On the subject of RNG and statistics (and gambling -- the roulette example above reminded me).

My brother took me to Vegas for my 40th and taught me how to play craps. I spent all weekend at the tables and had a good time.

Afterwards, (being a computer programmer by trade), I wrote a series of simulations as to dice strategy, using a betting system known as "Martingale" (you increase your bet each time you lose to cover the cost of previous losses). My first model was for trying to hit hard 4 or hard 10 (2+2 or 5+5) on a single roll. I "rolled the dice" 10,000,000 times in my program. The longer I "rolled", the more likely I was to have an insane loosing streak.

1. Hitting hard 4 or hard 10 in a single roll is a 5.5% chance. So you should be able to roll the dice 18 times and on average, you would get one or the other.

2. After 10,000,000 rolls, the "average" was 1 in 18. However, the program recorded a record loosing streak of 345 consecutive rolls without hitting hard 4 or hard 10.

3. Tried again with 10,000 rolls, the average was still 1 in 18. Worst losing streak was 156 consecutive rolls without hitting either number.

You can take away from this what you will. I still believe that there aren't any hidden maths and that the 20% and 10% tooltips are in fact, correct. That is mainly due to my RE history that I actually recorded (see above).

YMMV.

Darzil's Avatar


Darzil
06.22.2012 , 11:06 AM | #16
Quote: Originally Posted by Sothicus View Post
That's correct. I have to admit that due to my limited resources I can only go off of the 60+ RE's I've attempted so far with no success for a second blue of a different trait. But at 20%, that is very unlikely... about 1 in 100,000 chance in fact. So there is something else underneath the math I think... if at 10%, I'm at 1 in 1000 which is not great either but I have a little less heartburn over.
Unfortunately the only way to remove the selection bias from the equation is to ignore everything that has happened up til now, and then start counting and recording the result. You must also post that result here regardless of what it shows.

There are several 100 throusand players, so it is likely that several experience 1 in 100,000 chances with a specific RE attempt. If they are doing more than one RE attempt, it becomes far more likely. It is likely that a player experiencing this will post on this forum about it, so we'll see it here.

Vember's Avatar


Vember
06.22.2012 , 02:08 PM | #17
Quote: Originally Posted by punktual View Post
20% does not guarantee 1 in 5 crit.
Every single time you RE the game rolls a 5 sided dice so to speak but this gets rolled every time and your success on one roll does not change your chances on the next roll.

Random is random
True, but the odds of failing as often as we do is very low...too low, it would seem, for it to be happening. For instance, on my latest reroll:

Might Armor 6, 60ish skill i think? Took me 17 tries to get a blue
Might Armor 7, 90 skill...22 tries and I still haven't RE'd a blue.

Maybe it is just bad luck, but it happens every time I start REing to get a blue...I never get one short of 10 - 15 tries.

uniz's Avatar


uniz
06.23.2012 , 12:10 PM | #18
the random number generator that mmos use are very" streaky". i have noticed and seen it with several of them.

one day you will have a true rng spread. the next time you will have a streak of either good or bad luck. i can go with no re upgrades in 25 tries and the next day i can get 7 out of 10.

however even if they are streaky in the long run they are random. the law of large numbers will prove this.

when you are in a bad streak just stop.

teolawki's Avatar


teolawki
06.23.2012 , 02:04 PM | #19
Ok. Where to begin.

I've been on since Early Release and now have 6 alts, 1 for each crafting skill. Needless to say, I've spent many hours crafting. In fact most of my ingame time, easily 25+ hours a week on average, has been spent crafting. So I think it's safe to say I have a pretty good feel for the quirks of reverse engineering.

When I began crafting with my first character, I noticed that there were occassions where the reverse engineering system seemed to be, for lack of a better term, "dead". By that, I mean that there were times I would burn through 20, 30, 40 or more items before a reverse engineering discovery. (Keep in mind, the discovery rate then was 10%.) What was odd is that these "dead" streaks appeared to be so much out of the ordinary. Everyone who's done much crafting knows to expect some long stretches without a discovery, but these extraordinarily long streaks didn't seem to fit that.

Since that time, I've added my alternate characters and done a great deal of crafting with all 6 crafting skills. As a result I now have a pretty good feel for all of them. From my experience, I don't think there is anything wrong with reverse engineering in respect to a particular crafting skill or a particular item. Reverse engineering behavior seems the same to me across the board. This is not to say there isn't a possible problem under certain conditions. In fact, I think there may well be.

I remember reverse engineering problems being a hot topic among crafters during release and for a while afterward. Then discussion slowly tapered off and was sporadic for a while. This coincided with game updates that included changes to the crafting system, in particular the reverse engineering discovery rate, and a decrease in server population. However, with the recent character moves and much higher server populations, this topic is once again being widely discussed among crafters.

As I said, I experienced what I call "dead" streaks at game launch and wasn't really sure what to make of it. Over time, I experienced this less and less although it never totally went away. Since character moves have been implemented and server populations (server loads) have increased dramatically, the "dead" streaks are back. And the higher the current server population, the more often I experience this. To me this observation doesn't seem to fit with the "you're just having a streak of bad luck" argument. And it doesn't appear that I am the only one to notice a behavior change in reverse engineering.

I have come up with what for me seems to be a workaround for this. When I hit what I see as a "dead" streak, I quit the game client and then log back into the game after 5-10 minutes. This gets me back to what my experience tells me is normal reverse engineering behavior. I would be curious to hear about other gamers as to whether this makes a difference for them with reverse engineering.

To me it seems that there are times when the reverse engineering roll is not being performed or the roll variable is not updating correctly. It also seems that this is more likely and more comon under heavier server load. Don't misunderstand. I'm not saying this is what's happening. I'm just describing my impression of what I see. And, although I don't believe this is the case, I also cannot absolutely rule out the possibility that this is simply a perception issue.

What I would suggest is uncovering the reverse engineering rolls in the UI. I don't think it would be too difficult and it would certainly help in determining if there is an underlying problem in the live server environment. And if this is all a simple matter of misperception on the user's part, it would dispel that rather quickly.

I apologize for the rather long post. Regards to all and I welcome all constructive discussion on this topic.

Blackardin's Avatar


Blackardin
06.23.2012 , 03:01 PM | #20
Quote:
2. After 10,000,000 rolls, the "average" was 1 in 18. However, the program recorded a record loosing streak of 345 consecutive rolls without hitting hard 4 or hard 10.

3. Tried again with 10,000 rolls, the average was still 1 in 18. Worst losing streak was 156 consecutive rolls without hitting either number.
You've just demonstrated the flaw in the system. Odds mean nothing except odds. They are just a chance, that is all. If they meant anything people would not be losing their shirts in Vegas, and it would be called Caesar's Shack, not Caesar's Palace.

As proven, a streak of 345 and 156 can and will happen, and that is unacceptable. I'm currently in my second round of REing a blue barrel for a second alt. I'm up to 24 on this round. I believe I hit about 30 the last time I tried and quit. Both times I've blown through mats scavenged over the course of the month with a cost estimate in the 50k range.

On a spread-sheet with no variables included, this may seem balanced over time, but in real world practice these are unacceptable results for a level 17 item with a life expectancy of a few days and a worth of around 3k.
May the Schwartz be with you....