...

**The loot system MUST have a memory, otherwise it could not produce the results we see - and i will for sure not rate some silly "dev notes" from MAs inept staff higher than my own perception & experience, no f'ing way.**

Dear Wizzszz,

everyone is free to believe what he likes to believe but he is also responsible for his own believing.

For the first time, MA is communicating something about loot. It can be true or false. To falsify it we would need quite a lot of data, something that is not achievable easily. Perception is for sure rather useless here. On the other hand, we have studied the system in the past and all the data we had so far suggested a random system.

For instance, the double bombing experiment did show that return rate is lower and it did correspond to the expected reduction in hit rate when assuming that a find in an area has a Poisson distribution. Furthermore, the time between globals per avatar follows an exponential distribution, something that you would expect when dealing with a random system. It is however rather unclear if ATH’s do belong to this process or if they are triggered by something else.

Nevertheless, all those systems don’t need a memory but if you observe such a system and do record data, then many do think that they are able to break it. For instance, take a fair coin. The probability to get heads is 1/2. If you had 4 heads in a row, what’s the probability that the fifth toss will also be heads?

Well it is still 1/2.

Sure, many will state that the probability of the series is 1/32 as five independent events will have the probability of 1/(2^5) and therefore it must be more probable to observe tails in the fifth toss, but a series of 4 heads followed by one tail has the same probability. Hence, both possibilities for a series with five tosses ( 5 heads or 4 heads plus tail) do have the same probability of 1/2 and the information of 4 heads in a row is useless to predict the fifth toss.

On the other hand it is true that recorded data might influence perception. Let’s assume that a global has the probability of 3/100. To get two globals in a row is a rare event and would have the probability of 9/10,000. Hence, if one has recorded a global then to record a further one is quite improbable as 9/10,000 is quite lower then 3/100 * 97/100 = 291/10,000.

Similarly, if you had a series of bad loots, say no global in 100 kills, then the probability of the series 100+1 “no globals” is still more probable then 100 “no globals” plus 1 “global” as (1-3/100)^101 is greater than (1-3/100)^100*3/100 as you could easily verify.

Nevertheless, it is however quite improbable to observe a series of 100 “no globals” as the probability of at least one global in 100 loots is larger than 95% (1-(1-3/100)^100 = 95.2%) and hence the alternatives to 100 “no globals” as 49ng-1g-50ng or 1g-99ng etc.. are in sum quite more probable.

Hence with a random system one is able to perceive what he likes to perceive.