Not science? …ummm psychology would disagree. I feel often in these discussions again people get lost in the probability / math piece and not the psychological effects of Luck in game.
What most users are stuck on is the variable-ratio loop, meaning if i do X over Y time, then my reaction should be Z. Time is a factor in this as much as the potential variable of drops itself.
In gambling, the hook is simple - if punter invests X into the experience and the odds are Y, Y can retain itself as big/small in terms of odds, as long as Z time delivers dopamine.
The key piece to all this math discussion - Z … your time. Too much frequency over Z time equals bad behaviors/outcomes.
In psychology this called Resistance to extinction. If there is no stimulus and the example cited by @PeterDee is valid. As the only stimulus Peter has to go on is by visual observation and without validation from the game, well the resistance extinction by way of variable-ratio or any of the other 3 schedules in turn produces - doubt, skepticism, pseudo tests etc etc.
All still housed in a scientific outcome, but just not in mathematics - psychology instead 
What the argument for probability schedule dumps and so on made by others is to enhance the outcome through post-continuous reinforcement - “trust me, its working” however that’s not in-game and its not after each time you open a chest etc.
You can say have a 1/56 chance of open a chest to get N named item. Cool, odds are 1/56 right? however if my frequency over time is 1000 and additon time between each frequency increases to over 5 days.
Well… probability aside, It may as well be 1/1000000000 as time is critical as much as the luck roll itself.
Like a slot machine that you keep feeding, but don’t get the little fails/successes to encourage more cycles of betting (no continuous reinforcement).
Doubt continues to be the meal of the day no matter how much we haggle the toss on math, data mining and where things live or don’t live in the game.
