I think that’s a fair question. But I’d distinguish between understanding the implementation and understanding the behavior of a system.

We’re already accountable for systems whose implementations we don’t fully understand. We rely on abstractions, contracts, tests, and observability to establish confidence in their behavior.

To me, the interesting question is how much implementation-level understanding we actually need to be meaningfully accountable. That depends on the complexity of the code, the boundaries around it, and how well we can verify its behavior.

I’m not arguing that we should blindly trust code we don’t understand. I’m arguing that reading and understanding the code isn’t the only way to establish justified confidence in it. And that we should deliberately design systems to make that possible.