As hinted in many of the Issues, the underlying Data Structures for the World Graph are suboptimal for the randomization step.
Ideally we'd like to reduce the World Graph to another structure where it makes sense, like swapping the location scan to a linear list scan instead of a recursive search. Or swapping the Tree Structure of a the Requirements to something flatter.
Another way to phrase it is that we're trying to see if by precomputing some values, we can get a larger gain in performance than we lose to compute that.
As hinted in many of the Issues, the underlying Data Structures for the World Graph are suboptimal for the randomization step.
Ideally we'd like to reduce the World Graph to another structure where it makes sense, like swapping the location scan to a linear list scan instead of a recursive search. Or swapping the Tree Structure of a the Requirements to something flatter.
Another way to phrase it is that we're trying to see if by precomputing some values, we can get a larger gain in performance than we lose to compute that.