Getting the data design right for system operations is important. But what happens is that we design systems using our assumptions, assumptions we've never examined or challenged, assumptions which someday will be undermined by changes in the technology or the culture.
An instance of this in the New Yorker: writer, a house husband, unknowingly describes the problem He and his wife enrolled their child in kindergarten, filling out forms. His wife works outside the house; he works inside the house. But it turns out the school uses an app to make robocalls to a parent concerning school matters, apparently a lot of robocalls. His wife got the calls, he didn't, creating a mismatch of information, which led apparently to some tension in the marriage. When they challenged the school, turns out they could only contact one parent and someone had assumed the wife should be called.
In the good old days the number of calls from the school would have been rationed by the amount of time a human, likely the school secretary, had to make the call. These days the cost of making calls has been reduced to zero, meaning a big increase in the number made. Where the secretary could have dealt with the writer's situation, the robocaller can't, at least not with the existing data design. Since the calls don't cost, it would be easy enough to call both parents, if they desired. But that would require a new design.