That's a good question :) We'll give it a shot:
Imagine you have a set of questions that are sequenced from easy to difficult. This algorithms groups the questions into X pools. The first pool is the smallest one and contains the easiest questions. The second pool is a bit bigger and contains a bit more difficult questions. The thirds pool is larger then the second and has more difficult questions in it, etc.
Players receives the first question from pool 1, second question from pool 2, etc. untill the number of pools is exhausted. So if you set 10 pools, the maximum number of questions answered is 10. If you set the option that players can replay the quiz, the players perceive a very high learning curve. They quickly get to know the questions and answers for the first pools. But it still remains engaging because the later pools are more challenging.
So in short, this algorithm is optimal when you want your players to feel and perceive a quick learning curve, while still keeping them challenged.
Does this explanation help a bit?