I think "does one need Bob" is the wrong way to look at it.
Humans are narrative animals. Storytelling is a human universal. Done right, personas are a way we can engage our narrative capacity to help us in thinking through the complexities of a domain. (Done wrong, they let our preconceptions and prejudices erase the complexity of our domain.)
I'm not personally a huge fan of personas. Instead, I'm big on primary user research, where we spend a lot of time studying actual people, their lives, and how our products fit into them. But if the group of makers is large enough, you need some way to organize the insights from that into something convenient for discussion.
At my last startup we were working on something related to purchasing decisions. For us, the satisficer vs optimizer distinction was key. We did dozens of user interviews, and user testing each week.
We ended up with two main personas, one for the satisficer and one for the optimizer. It was handy to be able to talk about Bob vs Cate in terms of how each would react to a given design choice. By having specific imaginary people made it easier for us to put ourselves in their shoes, to talk about how the product fit into their lives.
We would also talk about specific real people (e.g., user test subjects, actual users) and about theoretical constructs. But having fictional archetypes was useful to us in thinking empathetically without getting bogged down in a specific person's details.