The difference between Darwinian selection and life appearing by chance is probability: It is very probable that a molecule that can self-replicate better will replicate more - Darwinian selection. It is not probable that just the right environment will come to help along a not-yet-self-replicating molecule. If the outcomes prior to one point in the chain are not relevant, then the chain effectively starts there; if you need certain non-common compounds in the 'warm pond', though, it doesn't start there. And yes, there could be more than one road to life (for example, you mentioned a variant of 101 steps where the molecule goes wrong and then an environment comes that redresses it). But the number of roads to life is limited - that is pretty certain after the failures of artificial abiogenesis. Yes, but of this immense number of possibilities, the overwhelming majority are gibbeish. Nature didn't eliminate those; we did. Nature advanced blindly; we don't. 100! gives such a small chance not because of the number of possible reactions at each point, but because the 100 environments must be independent from each other (environment 1 must not create environment 2, etc) and they must come in a specific order and no other. I took these environments as independent of each other. That's why I took the number of environments as small as 100; if environment A creates B creates C, for example, I took them all as a single environment 1 in the 100 chain (due to the high probability of B, C). Let's say that there are 1.000.000 roads to life (and shame on our biologists for not discovering even one) - this increases the chance of life emerging, but not enough.