"...Providing you have enough spindles you should, in theory, never lose data due to a disk failure..."
The problem is that in practice, your data is slowly corrupted, without you ever knowing it. Slowly, your data rots away, due to bit rot. So you are wrong.
Have you heard about ECC RAM? Why do you use it? Well, because your bits in ram flips randomly, they "rot". Suddenly, a few bits flip on random. Therefore you need ECC memory sticks to detect the flipped bits, so the computer can correct the faulty bits.
Exactly the same problem occurs on disks. The bits flip on random, but there are no ECC functionality to detect nor correct the faulty bits on disk. This is called silent corruption and a big problem when you store large amounts of data, over 10TB, say. Having raid and enough spindles does not protect against bit rot, as hardware raid, ntfs, ext, jfs, xfs, etc are not designed to look for these kind of errors.
There are lot of research on this Silent corruption issue in this link. (initial research shows that zfs solves all these problems):