I found C++ slightly easier to work in when using 0-based indexing, but then I was using n-dim arrays that I was hand indexing.
Pretty much every other programming language I’ve encountered and worked in as a mathematician is 1-based (R, MatLab, Octave, Fortran, Maple, and Maxima) has provided tools for working smoothly with n-dim arrays, and with that in place I find 1-based much more natural and less to think about. The first item is 1, the 2nd 2, the n-th n… in my mind the 0th thing in a collection is when you have an empty set so the thing doesn’t actually exist.
I guess if you’re counting counts then you might need the number of times 0 happened, but then you might want the number of times a thing happened in the years 1967-2021, then you’re out of luck either way (except that Julia handily gives you a solution to that).