It's not even the reduction in storage costs in this paper that is groundbreaking. They talk about a way to not only reduce storage costs, but optimize for repairs. Repairs are costly at scale and reducing resources where possible: network, cpu, disk reads, etc is ideal.