The flow of goods from A to B is one-way, and B is genuine.
Then the problem is transformed into the probability distribution of taking out three defective products from A.
According to the classical probability
P{X=0}=C33*C30/C63= 1/20
p { X = 1 } = C32 * c 3 1/C63 = 9/20
P{X=2}=C3 1*C32/C63=9/20
P{X=3}=C30*C33/C63= 1/20
Hypergeometric distribution belongs to sampling type.
And the conclusion is consistent with the classical probability algorithm, so it can be mutually verified to be correct.
As for the lottery model, it seems that it cannot be used to explain this drawing problem.
Because this model represents the probability that P appears "somewhere" in the whole permutation.
Not the probability of "overall" distribution.
LZ is a good idea, but the essence of lottery model is "conditional probability"
LZ, a single calculation method, is wrong in itself.
The "probability of defective part I" is indeed "1/2"
Essentially, it is superimposed after discussing all the possibilities of the first i- 1 time.
That is, on the premise that I is I-A, the first i- 1 time can be regarded as a probability.
But the hypergeometric distribution of sampling model has no such "premise"
This explanation is not correct, but I hope it will help you understand.
I suggest reading Probability Theory and Mathematical Statistics, Zhejiang University Higher Education Press, 4th edition.
Classical probability at the front of the book and appendix at the back