[wpeterson]

The problem with performance optimization is that if you obsess with optimizing code well outside the critical path, it's often wasted effort drowned out by larger latencies.

If this code was part of a web application, the I/O bound latency of backend/persistence calls will almost always outweigh any CPU bound processing time.

So they may have made this method call 14x faster, but how much latency were they really shaving off their average request? I would be surprised if it were more than a rounding error.