As a programmer, I understand that only us really control the software behind voting system. Most folks don't have the slightest idea how it actually works, and if there's a bug (or worse, a vulnerability) the consequences might ripple nationwide. With paper, you get some failed counts here and there, but nothing major, and such errors tend to average out. And of course, people can understand how it work, and how to make it more accurate and reliable.
Once we go back to paper, a ranking system becomes quite a bit harder to implement. Much easier however is to give a yes/no answer for each candidate. Easy to count, and the end result for each candidate is how many of the voters support them. Throw in a void clause if no candidate manages to get some threshold of support (say 50%), and you even get to take "all politicians are bastard" votes into account (something that ranked choice does not have).
Alternatively, we could could give each candidate a degree of support, ranging from "unacceptable" to "ideal", and elect the candidate with the best median support. Slightly more complex, but quite a bit closer to ranked choice voting, and you keep the "all politicians are bastards" votes.