Maybe they are not interested in optimization for the sake of optimization? For my use case it has been plenty fast anyway (not that it's a particularly intensive use case).