I like how the bees are said to figure out the answer faster than a supercomputer, when the problem involves "several" flowers. Unless "several" is 15+ flowers, a supercomputer is going to have an optimal solution within milliseconds, before the first bee has even reached the first flower. Now, if the bees had 1000 flowers to visit, I would be more interested.
Wow, how many different paths is that? Somewhere around 256 at a guess? So a computer could work out the optimal solution in microseconds purely by testing all possible combinations. Though I'm guessing the bees don't just sit there at the start working through all possible routes they're going to take that day, it's surely something that they work out as they go along. Maybe that's where they gain some efficiency advantage.
Though four flowers is hardly likely to be a real world scenario for bees. Surely bees would encounter tens of thousands or more flowers in a field, so it is there performance here that ultimately counts. Though perhaps four flowers is better for determining how they make their actual decisions?
The bees are probably faster if you include the time it takes to code a program for the supercomputer to run to actually solve the problem.
I'm saying 'probably faster' because that would depend on the algorithm and the speed of the coder or coders. I still bet a couple of talented well coordinated coders with a simplistic heuristic algorithm would still beat the bees in a realistic simulation (IE having to fly a distance to find said flowers similar to real nesting conditions, not through a hole or something 3' from the artificial flowers).
