They shipped the product. Weeks later, a single misconfigured setting leaked user data.
That is the cost of not making privacy by default the baseline. Privacy by default means systems start locked down, with the safest possible configuration turned on from the first run. It is not optional sugar on top. It is the structure. Every permission, every endpoint, every datastore—secure unless intentionally opened. That choice changes everything.
When privacy is default, human error has less space to destroy trust. Settings can be relaxing, but only by explicit choice. This is where user config dependent comes into play. User config dependent systems treat the secure state as the origin point, and all changes flow from deliberate, visible actions. The bias is toward security. The bias is toward safety.
Without this model, you rely on developers remembering to flip the right bits every time. You rely on managers catching every risk in review. You rely on no one making mistakes. That is not how real systems work at scale.
The strongest architectures assume misconfigurations will happen and limit their damage. Data structures, API policies, and storage rules should enforce least privilege. Default off. Default hidden. Default encrypted. Start simple, then allow opt-ins for higher exposure only when needed.