I didn't downvote you, so I can't speak for those who did. But I replied because simply saying "20/850 is small" isn't sufficient to argue there isn't a difference between the two planes. Partly because the timescales for the two incident counts are different and so are the number of planes constructed. So normalizing the incident counts against those quantities is a first step at getting towards their significance. Though as @ggm and @ansy point out, the incident rates also vary as a function of time through the lifetime of that airplane model. So really that behavior needs to be examined as well.