I think this is an interesting point. There is certainly a conflict between agility, and reliability. The reason we use APIs, and data contracts, is to increase the reliability. But arguably that can reduce agility.

So the question is, which do we want most?

I believe it's reliability we want most in both cases, but even more so with data. We expect data to be correct, and often assume it is when we use it for reporting, to drive product features, to train ML models, etc.

In each of those cases, wrong data is more harmful than no data, and can lead to poor decisions being made, or products that users lose trust in (which could also affect your brand).

Anyway, nice article and agree with most of it! 😀