Knobs conspiracy
So after watching Mobilegamer recent video and the mention of "knobs" as a hidden mechanic that can change the games difficulty and item drop rates on the fly. Can someone from CG confirm or deny the...
Forum Discussion
"EpsilonRhoMew;c-1876845" wrote:
I for one have been collecting data since the CLS event for when a single shard or gear is required and the case when zero of the target amount is collected.
Summary of 637 shards when 1 away from completing a star level:
Assumption:
Using a single sim until the condition of missing 1 shard is void.
Data is compared against 33.33% drop rate.
Data is compared against recorded shard drop rates per node.
Chi squared model is used to prove occurrences of drop rate changes under known conditions.
Result:
Dramatic decrease to 3.74% given the case before the sim. Original shard drop rates could be modeled under a Gaussian curve. Mean was 32.97+\-0.56%. Std Dev was +\-4.82%.
Being more than 5 standard deviations away from the mean when under these conditions is suspicious.
A similar result for gear is only 4 standard deviations away from the population mean. More confidence in the gear data is warranted due to a data population size of 2, 528.
Conclusion: More data needs to be recorded.
Actually pretty interested in going into this a bit @EpsilonRhoMew, could you expand a bit on how the tracking was done? e.g.
The way you present it, it sounds like 637 is the number of times, while having 1 shard needed to upgrade, you did a single sim? If so, then you got the shard to upgrade on 24 occasions? Or on 23 occasions? Because 24 occasions would be 3.77% but 23 occasions would be 3.61% (neither of which is 3.74%). I mean maybe the non whole number comes from you fitting a curve to a distribution? If so what did you use in this case? I mean 23 is probably a big enough number to approximate the (assumed) Poisson distribution with a Gaussian but maybe not?
Would be interested if you looked at the the distribution of zero shard streaks? That would give you something to fit to but the closed form expression for that is pretty complicated, Kudos if that's what you used?
But I am struggling with the choice of a chi square test when a simple binomial would do... but maybe that shows my bias.
Or are you saying that you reached a star upgrade 637 times (on farmed characters since CLS release). And had to sim 17,032 time to get those stars? Which doesn't seem to make sense? But then maybe you only upgraded a star level 24 times since 2 years ago? Although that does make more sense as you could have skipped the ones that didn't end up on 1 shard away from an upgrade?
Anyways, genuinely confused and interested, not trying to be difficult.
