Indeed I think there is a valid discussion to be had, in terms of how far can we go with reasonable sacrifices. Also the analysis finds only a correlation betwe

Indeed I think there is a valid discussion to be had, in terms of how far can we go with reasonable sacrifices. Also the analysis finds only a correlation between memory safety and expected user base. It doesn't find any correlation between safety and performance, nor does it find one between safe or unsafe internal abstractions and memory safety for users. Personally I believe, that blaming users for "holding it wrong" is not an effective way forwards. As a programming community we've tried that many times, and I've yet to see it succeed at scale.