That’s the gambler’s fallacy, and not my proposition.
It’s not the drop rate that needs to be observed, it’s the range and distribution. A player can’t predict the whether or not the next boss kill will generate a legendary drop, nor make predictions about drops during specific runs. But you can make predictions that within some range you will have a specific number of legendary drops within a reasonable margin of error, and you can observe distribution. And that’s because the distribution is designed to be relatively even over a range - that is not random.
I remember heycarnut and his epic post. It was possibly the best post, or one of the best in any gaming forum I’ve ever read. And I agreed with some of his premises, but not all of his conclusions. As just one example,
To repeat, transforming / weighting does not mean “not random” or even “less random”, it means a bias / distribution / weighting has shaped the distribution of outcomes. You still most certainly cannot predict them from simple observation.
Yes, yes, it does. It may be that a specific outcome of one event can not be predicted, but that does not mean that the system of events is unpredictable. (which he says himself later in the post).
But that doesn’t matter here. His argument, or one of them, is that weighting and distribution don’t affect the randomness of RNG, and that’s true - the dice roll is the dice roll - random. The dice roll is random, but the drops over a range are not.
Games use weighting, selection by ranges, selection directly by value etc.:
His thesis is that a single event cannot be predicted by the player, therefore the drop rates of the RNG/loot table of Unreal games are random.
No surprises here: the distribution of outcomes mirrors what we can predict, but the actual individual outcomes (I’m not showing all 100K for obvious reasons) are most assuredly unpredictable
A distribution of true random events would have marked clustering and long periods with no event. That is not the case in BL2. When you farm for a Lyuda for 90 runs without getting a drop, get 2 drops in 300 runs, or get 27 drops in a row, then I would say that the system is random. but if you’re getting a drop about once every 10 runs and get about 10 drops every 100 runs, that’s a very even distribution.
He says about the Pete farmer who observed exactly 15 drops in 5 straight sets of 100 runs that “15-18 would be expected”. No, a system of random events would have no such expectation. Random outcomes do not have even distribution over a specific range. Borderlands 2 boss drops do. You can’t predict which specific drop will be a legendary, but the distribution of drops over a range is not random at all.
If you expect a specific distribution over a range - that’s not a random system. You will not see an ‘expected distribution’ of real dice rolls over a range, even though the six faces are not weighted - that’s random.