Jump to content

new problem with Reverse Engineering?


Bolanos

Recommended Posts

Ok so I hope this post doesn't get overlooked but here is my problem with the current RE system...

 

If there is a 1 in 1000 chance to RE something... TELL ME BIOWARE! Don't tell me it is 20%. If it really is 20% and I go up to 30 tries with no success that means:

 

(1/5)^30 (1 out of 5 to the power of 30 tries)

1^30/5^30 = 1/931322574615478515625 = 0.00000000000000000001073741824

 

OR 1 in a VERY big number. This is the odds of not landing a 1 out of 5 (20%) in 30 consecutive rolls. Which brings me to my next point, perhaps we should be looking at the odds assuming the random generator was built with 4/5 chance of NOT getting a success each roll:

 

(4/5)^30 (4 out of 5 to the power of 30 tries)

(4/5)^30 = 1152921504606846976 / 931322574615478515625 = 0.001238

 

Ok now I realllllly suck at math and likely someone will point out problems with the above logic, but ballparking this I come up with (AT BEST) a 1.2 in 1000 chance of going 30 RE tries in a row without success.

 

If you are doing 1000 RE runs of 30, or about 30,000 items a day, this would be acceptable. But with most of us only RE'ing around 100 items or less a day we KNOW that something is broken in the RNG or RE system.

 

-----------------------

FIXES

-----------------------

 

I know what you are thinking, "ZOMG this guy is going to actually offer solutions rather than just complain? CRAZY TALK" but here is what I think could clear this all up:

 

1. Dev post explaining the RE system in better detail (what are the physical odds of RE various levels of gear, qualities of gear and stacks of items.)

2. Broken portion of RE system fixed (if it is broken)

3. A "failsafe" added to the system where each RE attempt counts up a counter. After 10 without success the 11th will always reward you. Making RE'ing at WORST a 10% chance.

 

Any chance a developer will read this?

Link to comment
Share on other sites

Ok now I realllllly suck at math and likely someone will point out problems with the above logic, but ballparking this I come up with (AT BEST) a 1.2 in 1000 chance of going 30 RE tries in a row without success.

 

I haven't checked the math, but assuming it's correct: This means that, in a population of 1000 crafters, one of them is going to draw the short straw. Now multiply that by the population of crafters in TOR. That's a lot of failure strings.

 

It's certainly possible that there's something wonky going on. However, all we have is a few anecdotes from people who've gotten screwed by the RNG at some point. The people who are actually tracking multiple extended crafting runs over a period of time have, so far, found it to be "working as intended." Yes, there are long strings without successes; this is expected behavior in a system with millions of operations. You don't see lottery winners complaining that there's something wrong with the system because the chance of winning is so miniscule that it simply couldn't have occurred naturally.

 

I personally think the crux of the issue is less whether the system is WAI, but more that the way it functions penalizes unlucky crafters on the far end of the curve to an unacceptable degree.

Link to comment
Share on other sites

Ok now I realllllly suck at math and likely someone will point out problems with the above logic, but ballparking this I come up with (AT BEST) a 1.2 in 1000 chance of going 30 RE tries in a row without success.

 

Yes you do, and yes I will. The second method is the correct way to calculate probability. The first method simulate failing 30 times with a 80% chance to succeed.

 

I don't know how many people play this game, but let's say 100.000 and we're probably being conservative. Let's also, conservatively, assume that they do 70 sets of 30 REs each, which is about equivalent to leveling 1 skill to 400. With a 0.124% chance for a streak of 30 to fail, that would mean that there should be 8.666 streaks of 30 failed REs in that sample, spread out trough the 100.000 people.

Suddenly it doesn't seem THAT unlikely for it to happen to you, right? To be exact, that chance is 8.6% for 70 steaks.

Link to comment
Share on other sites

Not going to try and argue with anyones maths but one thing I have noticed is that if I RE a stack of 5 Biochem Items (Green or Blue) I seem to always get a new recipe, which would back up the 20% chance, of course with other skills the items don't stack so I cannot directly compare but I know I have nowhere near the same level of success with my Artifice or synthweaving toons.
Link to comment
Share on other sites

My experience is that the 20% RE chance is about right... on all professions and all trees.

 

Being a bit of a geek, I put together a spreadsheet for my crafting, with all the schematics I could find, and every time I RE'd into a higher level schematic, I added that to the tree (so where applicable each green item shows the 3 blues I have researched, and each blue shows the 4/5 purples I have researched). For fun, I also recorded the number of RE attempts it took to reearch each schematic.

 

What I did not do is record the crafting level I was at when attempting the RE, so I am not sure if that plays a part, but from my anecdotal observations, REing greyed out schematics seems to be more likely to succeed.

 

On the schematics that RE into one single higher level item, my average is a little over 5.6 RE attempts per item. I have a bunch of outliers, where the RE happened on the first attempt, and several where the RE success took 10 or more attempts (22 is the record failure streak so far), but taking every value into account I come up with the 5.6ish number. If I exclude the top and bottom 10%, then it is about 5.5 attempts. Plotting the results on a graph shows an almost perfect bell curve (cut off with nothing below 1 attempt, obviously, so it has a long tail on one side).

On the green schematics that RE into one of several higher level items, I am looking at just over 5.4 attempts to RE the first blue, 9.8 attempts to RE the second blue, and 16.1 attempts to RE the third blue, again with high and low outliers that are included in the average calculations (including 3 RE attempts, 3 blue schematics). This is pretty close to the 20% RE chance that is expected, allowing for the fact that a successful RE (i.e. the 1 in 5 statistic) then can pick up any of the higher level schematics, even ones you have already picked up.

 

My sample data size is for just over 3000 schematics across the different professions, and around 21000 RE attempts.

Link to comment
Share on other sites

Guest
This topic is now closed to further replies.
×
×
  • Create New...