They let me spend the day using the computers to work with the existing models. I tried to get some sort of grasp on what was going on, but I couldn't. All I was doing was feeding the computer numbers, then crossing my fingers and hoping it produced the output I was supposed to get. I found myself gently massaging the numbers, celebrating any movement toward the right answer, then trying to replicate what I'd done.
One of the problems is that the algorithms have an element of randomness. At least, that's what I've inferred from entering exactly the same numbers and getting different outputs. Unless there are other factors outside of my control that influence the result. I asked the instructor and he sighed, and said I clearly wasn't getting it, and maybe it would be better if I went back to work and tried to view the models in context. I asked whether someone more senior, with a better understanding, was available for me to talk to, and the instructor told me that anyone who achieves a good grasp of how the models work is moved out of the group onto other projects. I thought of the secure elevator, and wondered again what happens beneath us.