I wonder if this lack of significant link is because number_of_bugs = time_spent_programming / time_fixing_bugs So choice of language does affect the number of

I wonder if this lack of significant link is because number_of_bugs = time_spent_programming / time_fixing_bugs So choice of language does affect the number of potential bugs. But this ratio of effort a programmer gives to bugs vs features is almost completely independent. Actual bugs making it past the programmer depend on how much effort the programmer cares to expend So a safer language doesn't reduce bugs. It's like farmers with back pain from driving tractors. Giving them comfier rides doesn't reduce their pain; it just means they drive faster upto the same pain limit If this is true, then bugs can be reduced by making bugs more painful for the programmer But also a safer language will let the programmer drive at a higher speed, releasing a greater percentage of safe code (even if the total number of bugs is the same)