I was wondering if some would trade me 2 accuracy tier 2 enhancers for 2 accuracy tier 1s ? I keep looking on the AU but the groups are mostly too large like 10+ and I am to cheap to spend that much :).
Got it done.
What is your hypothesis
That when more than one enhancer is installed they get knock off fast than just one. Sure enough I put on a second enhancer in the second slot (had tier 1 and 2 filled with one enhancer each) and lost it on the second shot and it was the one I just got (tier 2). Also lost another of the new ones before 500 shots, the third one lasted ok but started loosing the others faster than before (tier 1 slot). With one enhancer in the tier one slot they lasted much more like advertised.
Good that you could independently confirm this 
You will need one or two hundred enhancers to show anything statistically significant
Based on my data though, each slot is indepdendent of each other.
1 Like
This is my third or fourth test (not counting the times I noticed the enhancers seemed to go fast after I added one in a new slot) and I can not afford more testing . In the past few months of testing it is always the same. As a matter of fact I had one enhancer go on the first shot just after I added one fresh from the AU. You also get the prize for being the one who says your tests are completely opposite of my mine:)
1 Like
I know money is of course an object but this is incredibly hard to test without being able to sink a few hundred enhancers. You’re going to need to hunt a significant amount under each condition (a set amount) to be able to really say definitively if it matters. When you have more enhancers on, AN enhancer (any enhancer in the set) will break seemingly far more often because each slot can potentially break every shot.
Unfortunately, enhancers cost an arm and a leg.
1 Like
According to MA enhancers on UL guns should last 1600 shots. Mine don’t. I doubt very much if I would ever use, let alone test, hundreds of enhancers. First, anything tested were randomness is involved is iffy at best and statically unsound. Second and most important, this is just an observation on my part and no matter what I find if I tested 1000’s of enhancers it would not change anything:). So really who cares ? I was just answering Immortal and don’t really care about the discussion about truth of it and as you can see someone has already disputed my findings. As always whenever anyone makes a statement about testing anything there are always someone that finds the opposite and that is due to the randomness and that is because it is the nature of the game. The only sure thing is most will lose money if they play the game, just some will lose more than others and that is the cost of having fun IMO. Oh and BTW and not to be too rude I do know what it take to test even if I choose not to go that far, so I really don’t need instructions how all the magical stuff works :scratch2:.
Sure; I’m not fan of enhancers either, due to costs and randomness.
Who knows with MA’s new eco ArMatrix range, they might go full circle. Better eco 70%+ new guns, that don’t take enhancers.
It wouldn’t surprise me, and I bet they’d be quite popular.
Rick
N~36 is the minimum for statistical significance.
This has been tested thousands of times already (by me and soc mates over millions of shots, years ago). You just confirmed what was already known. i.e. the status quo remains
dont doubt yourself…when you think something is not right trust me something is not right…your mind only picked it up because it sensed something was not normal or right…or you can go with science…stop being paranoid…stop using your mind…believe in what your told by us and accept it…why do you even have that lump in your head…dont use it…and with MAs impeccable record and fairness…you have nothing to worry about.
Are you crafting them or buying them at auction?
If you are crafting them or buying them from a trusted source, you can expect to the maximum usage given MA’s parameters, whatever they are.
If you are buying them from auction, some people have the reputation of using their enhancers X numbers of times and then selling them on auction. That could affect your experience.
So if you are testing a theory, I would use newly crafted enhancers, not “used” ones from auction if that’s even possible. I haven’t tested.
CO
never thought about that, but it does make you think twice before you pick them up from auction :scratch2:
My understanding is that they work on a per uses basis. As in, at each use, there’s a 1 out of 2000 or whatever the break rate of the enhancer breaking that is calculated at that time.
Otherwise, they would just give the enhancers a tt value right?
Statistical Significance
Because I’m procrastinating on real work…
So, one thing you could do determine how significant your data will be depending on the number of enhancers you use up is to run what is called a monte carlo simulation (there might be a easier analytical way to to do this but I know this method and I am not as good with my theory :P)
https://en.wikipedia.org/wiki/Monte_Carlo_method
Anyways, here we assume chance to break an enhancer per use is simply 1/uses_enh_expected.
We can determine how many total uses we will expect to get out of the number of enhancers using a random number generator, then average over the number of enhancers. If we do this many times, we get a distribution that we can generate a standard deviation from, which tells us, if we ran our tests x times, that ~68% of the time, our number would fall within plus or minus the standard deviation.
Below are plots of my simulations (200 iterations in each)
Expected Break Rate: 1600 uses per enh, number of enhancers = 10, 50, 100
http://i.imgur.com/Ho1mYy1.jpg
http://i.imgur.com/BqCzoVt.jpg
http://i.imgur.com/T7euRkd.jpg
Expected Break Rate: 3200 uses per enh, number of enhancers = 10, 50, 100
http://i.imgur.com/ZqEJ0us.jpg
http://i.imgur.com/3UHNmjo.jpg
http://i.imgur.com/n0SICb5.jpg
As you can see, even with 100 enhancers, your stdev is just under 10% of the expected break rate.
I believe also, you can extrapolate the N you would need to achieve a smaller stdev by taking the sqrt of the ratio of Ns (sqrt(N0/N), and multiplying by the stdev of the N you know. For example, if you want a stdev of 3% of break rate, you would need 3% = 10%*sqrt(100/N), and this would actually require an N of ~1000. N = 200 should give a stdev of ~6-7%.
Edit: Corrected scaling and bin sizes to be more uniform and clearer
Zho
1 Like
:mad: Staph it!
So if we were to run with just 100 enhancers, it should still fall somewhere around what that chart shows for 100 enhancers? Though it would be a real pita to figure out how many uses an enhancer lasts for if you have them filled in for every tier…
though if you lump them all into one total and then multiply each use by # of slots?
Way too much to read but it looks nice. You are making one mistake and all of this is for not. Randomness, the enhancer can break at anytime because it is random. You just can not get valid data for random events. Probabilities but nothing solid.
Also I would if the thought about getting used enhancers is right. Just thinking out loud here but I would think it would take a lot of code and some REAL programming to keep track of the 1000s or 10s of thousands of enhancers rather than the player. For all we know when a player gets a new to him even if it has been used, enhancer it may just reset to the player so it is like new. BTW how would you someone know that one is selling used enhancers ??? If I were I would not tell anyone and it not like you really can see what they look like.
Updated images for more clarity
If the original poster’s theory is to be tested, they would have to run one experiment with only Tier 1 enhancers run for at least N=100 (200 to be safe), and determine the average break rate.
They would then have to run the same, but with Tier 1 and Tier 2 enhancers, and run for at least 100-200 enhancers in EACH slot, and determine average break rate, ideally for both slots.
They can then attribute an uncertainty to their averages, which we can say is ~10% for N=100 enhancers. If the break rate averages are significantly different (much larger than the stdev) than you have something interesting, otherwise the results would not show that there is a different break rate for using multiple enhancers.
CozMoDan, this DOES take the randomness into account. I basically increase use counter, and determine, in a random manner using a RNG, whether the enhancer will break on that use. I see how many total uses I will get until all enhancers that I am testing break. I get an average enhancer break rate for that trial. I repeat, in this case, 200 times, and I get a distribution.
So the above is NOT deterministic but is probabilistic.
Also, I updated the plots to make them nicer…
https://www.planetcalypsoforum.com/forums/showthread.php?299542-I-would-like-to-work-on-a-theory-but-need-a-favor&p=3601819&viewfull=1#post3601819