Though I enjoy and avidly read Mr. Miller's columns and blog, I find myself in disagreement with this and will offer a contrary argument.
Consider that each additional supplier requires management, which means additional resources and cost. If the parts need to be interchangeable, as we can assume if the alternative is a single part from a single supplier, then the opportunities for problems with the interface grow with each supplier (e.g. poor fit of linkages, or communications problems between software modules). This idea of using multiple suppliers to increase fitness to the competitive landscape focuses on a limited subset of that landscape, such as occasional supply chain disruption or cost of recall. Technical performance, cost and reliability are also part of that landscape.
In terms of the Toyota brake problem, there's no guarantee that the other suppliers wouldn't have problems, too (the same? different?). We might expect problems from each of them, each problem affecting fewer customers but the same total number of customers affected. If Toyota did not also increase their supplier management resources to handle the extra suppliers, increasing overhead, then we could even expect more problems. Some of these problems would overlap in time, requiring additional engineering resources to diagnose and correct. Indeed, GM, Ford and Chrysler may provide a good set of test cases for the reduced quality and added costs of multi-sourcing.