Reverse engineering 20% my foot

[Solates]

On a side note if they want it fixed send me a message just throwing it out there. I live close enough to austin to make it worth my day to travel over there and unjack this horrible scheme.

[Cerion]

really no comments on my math, and just one agreeing with my final statement?

 

Sure, I'll comment.

 

You point this out: x=30;1 - ((4 / 5)^30) = 0.99876206

 

Which translates to 99.876206% Now in a population of, say, 100k crafters, how many would you expect to have a bad streak? That number is 100k- 99,876 = 124. So out of a population of 100k, one could expect 124 players to have a bad streak. If there are 500k players, then 5x124 = 620 players with a bad streak.

 

Those of you having a bad streak are simply members of that population expected to have a bad streak. Sucks for you, but doesn't mean anything is broken.

[Cerion]

The OP is nothing new. Many people have started threads, posted about the same, and nothing has changed.

 

Bioware figures no one has unsubbed over it yet, why bother fixing it.

 

Fanbois point out that statistics are evil and random.

 

Crew Skill are an epic fail. Don't bother with them.

 

lulz...because people understand math, they're now labeled 'fanbois'. Hilarious.

 

High School Math teacher: Johnny, why aren't you studying your statistics?

Johnny: Because if I try to understand this stuff I will be seen as a fanboi by others on the interwebs! Peoples opinion of me on the interwebs is more important than knowledge!

[KorinHyvek]

And it does track how many you have re'd...how do I know this? Because once one procs you usually get the other two right behind it because you have finally RE'd the 50th one.

 

Correlation =/= Causation. You're right, the crafting system could be designed more intuitively, and that may or may not be something they're working on. But please, don't accuse people of being "fanboys" just because they understand statistics.

[Glower]

But I totally agree with the people who say that it should improve per failure.

 

^this

[Cerion]

Let me also be clear that even though I'm sure the statistics are right and the system is working as intended, I'm not a fan of purely RNG systems.

 

Not only will you have populations with bad streaks, but in large populations such as SWTOR, you're bound to have some poor unlucky sod who the RNG gods have determined to be the one who is always unlucky. Through no real fault of their own, they're punished by the RNG system.

 

That's why I agree that anything to mitigate streaks only improves the game.

[Zhiroc]

I don't mind an RNG. There's a common fallacy in the real world that if you work really, really hard at something, you'll eventually succeed...

 

I just wrote a quickie simulator in python. I ran 1000 dice rolls at a 20% success rate:

• 194 successes out of an actual roll count of 990, or 19.6% success (I ignore the trailing failures)
  
• median: 4 rolls to success
  
• max rolls seen for success: 27
  
• about 10% took 10 rolls or more
  
• about 5% took 15 rolls or more
  
• about 1.5% (4 tries) took 20 rolls or more
  

 

Edit: I just ran the simulation with 100,000 dice rolls:

• Median is still 4
  
• Max rolls: 42 (!!!)
  
• 90th percentile: 10 rolls
  
• 95th percentile: 14 rolls
  
• 99th percentile: 21 rolls
  

Edited May 11, 2012 by Zhiroc

[Zhiroc]

I just calculated my overall RE success since 1.2 released: 153 attempts, 33 successes for a rate of 21.7%. The longest string was 15, and 3 at 12. In other words, 1 out of 4 items took over 10 rolls to succeed. This puts "over 10" at roughly the 88th percentile, which is pretty close to what my simulation had. Edited May 11, 2012 by Zhiroc

[DeltaEkotte]

Saying that there is an XX% chance to RE something implies that it is based on RE'ing a stack of 100 items. Let me put it this way, if you have 10 items stacked in your inventory with a 20% chance to get a schem, if we follow what I said before then you will get 2 schems. But from what I have seem is this, you have 10 items stacked in your inventory, you hit RE and move the slider so that you RE all of them at once. You RE the stack in your inventory, but the game breaks the stack of 10 items down to 10 stacks of 1 item and RE's each item. So while you have 1 stack of 10 in your inventory, you RE 1 item 10 times. In short, stacks of items in your inventory is only to keep it from filling up, when you RE a stack of XX items you RE 1 item XX times.

[Semitote]

20% is 20%.

 

You could R/E it 6,000 times - you still only have a 20% chance to get the pattern each of those 6,000 times.

 

Your odds don't increase the more times you do it.

 

I R/E'd 5 things this morning on my operative... I learned the patterns on three items in a row from those five.. People tend to forget about that luck they when R/E'ing other items... They only care about the one that takes forever to learn the pattern for.

[Goretzu]

20% is 20%.

 

You could R/E it 6,000 times - you still only have a 20% chance to get the pattern each of those 6,000 times.

 

Your odds don't increase the more times you do it.

 

I R/E'd 5 things this morning on my operative... I learned the patterns on three items in a row from those five.. People tend to forget about that luck they when R/E'ing other items... They only care about the one that takes forever to learn the pattern for.

 

Yeah they do.

 

However that doesn't preclude a RNG issue (or wrong % setting issue).

[Nanos]

Here's my vote for a %1 increase per try and/or RNG streak breaker.

[Slayd_]

Have to agree it's bugged.. 20% should mean 20%.. I can understand perhaps it taking up to 7 tries to get a schematic.. but I myself have been doing some REing today and after 22 RE's I got 1, yes one schematic.. that's less than 5% chance there.. that's absolutely ridiculous.

 

And anyone saying we don't complain when we get a schematic on the first try.. I would prefer to consistently get a schematic according to the percentage indicated rather than possibly getting it the first try. And I can tell you from experience that more often than not I end up making 2-3 times the projected amount of items to get schematics.. there is a problem here that needs to be addressed.

[Zhiroc]

Have to agree it's bugged.. 20% should mean 20%.. I can understand perhaps it taking up to 7 tries to get a schematic.. but I myself have been doing some REing today and after 22 RE's I got 1, yes one schematic.. that's less than 5% chance there.. that's absolutely ridiculous.

 

And anyone saying we don't complain when we get a schematic on the first try.. I would prefer to consistently get a schematic according to the percentage indicated rather than possibly getting it the first try. And I can tell you from experience that more often than not I end up making 2-3 times the projected amount of items to get schematics.. there is a problem here that needs to be addressed.

 

Write a simulator like I did (see a post above), and you will see that long strings of failures are certainly possible, in fact, likely over time. I ran my simulator for a million dice rolls, and after doing that 10 times, each experiment had one string that required over 50 rolls to succeed, and about half of the experiments had one that took more than 60.

 

In fact, the 90th percentile is at around 10 rolls, meaning that 1 in 10 times you try to get a specific schematic, you will take 10 or more. The 95th percentile is at around 14, so 1 in 20 will take that many. And the 99th percentile is at 21, so 1 in 100 will take over 20.

 

Like you, I thought the RE chance was broken, but luckily, I keep complete records for every RE attempt I make. I was surprised to find out that my actual success rate is slightly over 20%. It goes to show that we humans suck at having a feel for probability.

[VeovisX]

20% is 20%.

 

You could R/E it 6,000 times - you still only have a 20% chance to get the pattern each of those 6,000 times.

 

Your odds don't increase the more times you do it.

 

I R/E'd 5 things this morning on my operative... I learned the patterns on three items in a row from those five.. People tend to forget about that luck they when R/E'ing other items... They only care about the one that takes forever to learn the pattern for.

You are correct the odds don't increase, but the probability that your next RE will success does. Those are 2 different things. If you keep trying you will eventually get it, unless your RNG is broken.

[Cerion]

Write a simulator like I did (see a post above), and you will see that long strings of failures are certainly possible, in fact, likely over time. I ran my simulator for a million dice rolls, and after doing that 10 times, each experiment had one string that required over 50 rolls to succeed, and about half of the experiments had one that took more than 60.

 

In fact, the 90th percentile is at around 10 rolls, meaning that 1 in 10 times you try to get a specific schematic, you will take 10 or more. The 95th percentile is at around 14, so 1 in 20 will take that many. And the 99th percentile is at 21, so 1 in 100 will take over 20.

 

Like you, I thought the RE chance was broken, but luckily, I keep complete records for every RE attempt I make. I was surprised to find out that my actual success rate is slightly over 20%. It goes to show that we humans suck at having a feel for probability.

 

Very astute of you! Nice work. This is what I've been trying to say. With a large sample size (such as your simulation parameters or the Players in this game), you're going to have a population of players that has horrid luck. People who think the RNG is broken are simply those people who are falling into this population.

 

Having said that, I think BW should implement a mitigating system to prevent long empty streaks.

[siegeshot]

Or they could just implement a little "xp bar" to each item that says okay you need to get this 'research bar' to 100%. And then if you crit reverse engineer, lets say, you gain more of a %.

 

As is last night I got a crystal recipe I wanted upgraded in 2-3 tries and the shield recipe I wanted upgraded in 1 try. But I have another crystal recipe that's over 15 tries.

 

It's better than it was, but it still needs a total overhaul.

[MorgothPl]

The XP bar would be perfect solution. Everyone would know how many items you need to RE. I was trying to get hte blue barrels lev 49 - it took 2 greens for Reflex receipe and about 9 greens for Skill receipe.

 

From my own observations on cybertech and armstech, it is worst at levels 33, 35 and 37 - it takes most green items, to get blue receipe in that level bracket.

[siegeshot]

The XP bar would be perfect solution. Everyone would know how many items you need to RE. I was trying to get hte blue barrels lev 49 - it took 2 greens for Reflex receipe and about 9 greens for Skill receipe.

 

From my own observations on cybertech and armstech, it is worst at levels 33, 35 and 37 - it takes most green items, to get blue receipe in that level bracket.

 

Thank you for the kind words. I started a thread up on the suggestion board. Hopefully I don't get trolled into oblivion. Here's the link if you're interested.

 

http://www.swtor.com/community/showthread.php?t=454473

[ekimmak]

I've managed to get the purple level schematic for a mod, by reverse engineering a mission reward by mistake.

 

I have also lost count of how many times I've reverse engineered medic packs on my sniper, without getting the reusables, and of how many medical supply missions I've run without getting the purple grade for reusables. I run these at the same time, meaning my sniper is pretty much broke, since the two companions he'd rather run with are the ones that should be crafting. (Vector is also an annoying case, as either he runs gathering missions to keep up supplies for crafting, or running diplomacy missions to get those supplies)

[ekimmak]

Why would you learn anything disassembling it that you didn't learn building it in the first place? The system as it stands makes no sense as 'reverse engineering'. I'm not even sure why they called it that. A system where you could disassemble a device built by someone else to learn how to make it would be interesting, but that's not what this system does... it just breaks stuff back into component parts, and sometimes you apparently find a prize inside.

 

This.

 

It's very frustrating to be unable to reverse engineer stuff, say, a random mod off the ground, to learn how to make them. I've picked up overlevelled purple stuff, and don't know what to do with it. I can't sell it on GTN because no one ever buys stuff off it, and there's no point to reverse engineering it, because neither do I learn how to make it, nor do I have anything I can learn from it. Highly frustrating.

 

I reckon it should be that each time you reverse engineer something, you pick up a schematic fragment of the item. Get a certain number of schematic fragments, and you can "spend" it to learn the next level of the item. It doesn't make sense that my character learns how to add crit to his gear when he's looking for defence.

[siegeshot]

I used to actually think you could randomly learn gear you found off mobs by reverse engineering it. I've probably reversed 100 items and just gave up some months ago actually getting something. Not to this day have I ever got a single item from reverse engineering the items off mobs. I never realized it was because that functionality didn't exist until 1.2.

 

So frustrating. :rak_06:

Edited May 14, 2012 by siegeshot

[Zunayson]

Math says that there's a 10% chance that when you reverse engineer something 10 times you get no recipe. It isn't accumulated. Each and every chance is a new roll. You don't seem to mind it when you get the recipe first try.

 

This. For every time you get it the first try, that's as likely as missing it four times (or five? Bad at maths) times in a row.

[Lady_Alyria]

I got 4 new schematics out of 5 REs yesterday, it was 1,2,3 and 5. I'm sucking up all your luck, sorry bro. Edited May 18, 2012 by Lady_Alyria

[Drewhat]

If Bioware were to mess with the results of the 25% of people trying to research so that research was easier it would shift a whole lot of demand down. Demand for materials would go down. A proper and less intrusive solution would involve not affecting the total resources consumed. When stupid people are continuing research paying 20-30 times the worth of something when they many never make that value back, that is what is factoring in to the cost (and demand) of materials.