First of all, it wouldn't a be a normal distribution, it would be a binomial distribution, which is discrete like a Poisson (e.g. probability of a fire).
Second of all, ICE fires might be in the thousands, but there are also millions of ICE vehicles on the road, which means the fire probability is also very low.
http://en.wikipedia.org/wiki/Poisson_limit_theorem
People are still debating about what those probabilities even are, so it makes little sense to dismiss that discussion as being futile by tacitly asserting what those probabilities are
a priori.
But most importantly, none of this is relevant to the discussion. You can still compare probabilities between two different distributions and do all the math that people have been doing. A Poisson is actually just a simple approximation of a binomial and the math is much less complex.