I don't think that feasable from a security standpoint. Ideally, software would be open-source and authored under the supervision of a handful of trusted parties.

The metaphorical "chain of trust" is strongest when it's composed of a few strong links, and when its cracks are visible.

What you're suggesting is that we should use a model with many small links and invisible cracks.