Comment by cogman10

I'd say that this is premature optimization.

It doesn't take long for hash or tree lookups to start outperforming linear search and, for small datasets, it's not frequently the case that the search itself is a performance bottleneck.