I fixed b because you fixed k and I was wanting to see if the equations for b and k-1 were the same in the limit, i.e. if the kth lowest hash is exactly like a target for b. Also, when someone runs getnetworkhashps, they’re selecting a b number of blocks. Difficulty adjustment also uses a fixed number of blocks.
My simulation doesn’t specify a timespan, so I guess I could be wrong I’m wrong in saying it’s for a fixed amount of time and thereby doesn’t say or reveal anything about hashrate. But if I said it’s for a certain timespan before I ran it, I don’t see what I would need to change.