User blog comment:BonesMcGee/Brainstorm: Potential savings in using the hurry hatch trick/@comment-5116140-20120911130813/@comment-5445410-20120911152805

That is true for a flipping coin. This is because we understand the fundamentals behind the process however, we have absolutely no insight into how Tinyco's algorithm works. There maybe be a strong dependency between successes and failures. This we cannot check unless you have a very large dataset.

Moreover, the The geometric distribution is the only memoryless discrete probability distribution but there are continuous distributions which are also memoryless like the exponential distribution but perhaps this is unfair because the exponential distribution is it's continuous analoge.

How do we know an egg is replaced under the same conditions as soon as it is drawn ? Perhaps the first draw has different dynamics than the second draw. Perhaps the rate has time dependent diminishing returns.

I still think we should not dismiss dependencey amoung drawings here. We have no evidence to discredit it. I know independency makes computations easy but this should never be a reason to accept it.