Expanding on foundational technology for distributed data for decentralized apps, we must recognize the separation of concerns between an app's data versus the user's data.
If someone offers an AI image generator as a service, the app's own data includes the model weights. This is distinct from the end user's preferences, prompts, and generated images. We would want to enable the end user to own his data, so that he can control the access (privacy, sharing) and he can take his data to other services. For data to be as censorship-resistant as identity (SSI), we introduce the concept of self-sovereign data (SSD). The user's data can be scattered and replicated throughout the Internet, served by diverse storage services without any one platform having control over availability. Think of this being enabled by a contemporary bittorrents protocol, because we want to preserve a p2p approach for censorship-resistance.
Through this separation of app data and end user data (SSD), we would design every app to honor this bring your own data (BYOD) approach. This becomes the norm for the future decentralized Internet. Every app must support bring your own identity and bring your own data, so that end users can never be locked in (and rug-pulled) by the service provider.