It and others I've read on machine-checked proof showed the computer-centric one could catch quite a few problems with higher assurance of correctness. The peer review problem in science also makes me think it's important given I can't be sure the huge proofs will be adequately checked. Far as reliability on computer end, I've read on verified processors, proof assistants, compilers, and so on. Much of the risk can be knocked out but the black box that is human brains is another story.
I don't give a crap if a computer can read my proof.