Quote:
Originally Posted by Alannah-Shidreth
I hope you don't feel that way about it, I truly appreciate it that you took the time to look.

As far as picking at the statistics of what is coming out of the system, you said it yourself that you aren't a programmer. It's no fault of yours that the RNG seems to be misbehaving some of the time.

I also take exception with the fact that the dev guys would actually have you click open 10k packs to test it. Not only is this a huge waste of time, it doesn't provide a huge sample size at all - ~100 recipes on an already low margin of 1% is far less helpful than it could be. You really should have been given better tools to test stuff like this - like a test harness that you can click a button and force execution of the routines any arbitrary number of times without actually creating a test character and opening bags in your packs to look. As a software developer, I tend to write test harnesses like this all the time.

As far as the statistics are concerned, the two sets of statistics compared to one another are pretty wildly different. I should have come up with standard deviation values from the numbers instead of the distances of the ends from the mean. Standard deviation is the appropriate way to look at the spread of numbers. In this case your first test had a standard deviation of 4.12, the second test was 6.69. The first was a little higher than I'd expect but probably reasonable over such a small sample size, the second was quite a bit higher.

Ideally, a lower standard deviation is good - a standard deviation of zero would mean a perfect balance (not going to happen where random numbers are involved) when you're comparing equal percentages of chance. In this case because the expected percentages are all supposed to be clustered very closely to the mean (each of the numbers in the first test should be very close to 25% and in the second test they should cluster around 16.7%). The larger the sample size the more the standard deviation should approach zero.

In your second test you can just as easily look at those numbers as being 18%, 9%, 12%, 19%, 23% and 27% since the total number = 108 (close enough to 100 for argument's sake). The numbers all look to be quite a bit off from 16.7% with just 9% being very low and 27% being very high. If this were equal they should all cluster around 16.7 in standard deviation, but they don't.

All of this is a moot point anyway if the dev team is too busy to cram large quantities of numbers to test the RNG and see if it performs like it should.

Thank you again for all your work on this. I truly appreciate it that you took the time to test it at all.


Hmm need to clarify that nobody forced me to do anything. I am a dev and took the time to use the tools at my disposal to execute packs like you players did. The sample size may be small to you but it is rather large considering my parameters stay the same and yours constantly change. None of you here have the ability to click 10000 kits within a short amount of time. Our RNG has been looked at several times and tweaked here and there throughout the course of the project. It has certainly shown a disdain for some people and a liking for others, but as you can see from my sampling it is random and as stated before the code for loot tables is all the same, crafting bags use the same code as everything else with a loot table. If there was a code issue with our loot tables it would affect the entire game.


Weiterlesen...