You did not read the “small print”. Rossum did not say that there will be 1 million “ones” in succession which would be very unlikely. However, with one million tosses, there it would very likely happen to see 30 or 40 “ones” in a row. If one would toss a die 10sup[/sup] times, it is mathematically certain to have 1 million “ones” in a row.
Of course this is all inapplicable, just like your “bicylcle part” analogy. The chemical processes are not random, and when a very simple molecular chain just happens to form randomly - and that is not unlikely at all - then the next step will “build upon it”.
Using that pretty bad bicycle analogy, let’s suppose that small quakes randomly shuffle the parts all the time, and somehow a cogwheel would connect to the chain. From that moment onward they would act as one piece! And then some more time later another chance event would connect two other parts… and so on. The point is that one does not always start from “square one”. The small parts which are stable and molecules are very stable, will act as higher level building blocks. Of course the bicycle analogy is still incorrect. Molecules are unlike the pieces of a bicycle.
As for empirical experiments, they are conducted all the time. Scientists try to recreate the alleged environment of the “primordial soup” and in just a few **months **(!) they get very promising “not-quite-life” but very complicated results. And we are talking about **months **here, not billions of years. On top of that, what is “life” is a matter of defintion. I suggest Dr Asimov’s wonderful collection of essays “The Tragedy of the Moon”, where he talks quite extesively about “what is life, and what is inanimate”.