That's a consequence of memoisation. It scales better than the original code though.

Still, the original point (node is cooperatively multitasked) was clear. Anything beyond that and I just want to reach for better Fibonacci algorithms.

And this is why, any cache without an eviction policy and/or size limitation is bad (and memoization _is_ a cache). I shudder every time I see a "magical" memoization function whose interface is just

   memo function-to-memoize