Users own their application data independent of and *outside of* the application. A Blockstack application is only considered decentralized if its users control where their data is hosted and it can prove that they created the data.
Blockstack applications meet this principle if they use the Gaia storage system. Users may sign and/or encrypt their data in Gaia end-to-end. All files in Gaia are addressed by a user identifier, an application’s hostname, and a filename. Through Gaia, users can prove data ownership and restrict access.
In the future, users will be able to choose on an app-by-app basis which Gaia hub serves their application data.
Users are the sole administer of their own independent and unique identifiers. Within an application, users must be distinguishable by unique identifiers. The DApp cannot mask or take away a user’s identifier, and a user must be able to bind their identifier to the data they create.
Blockstack DApps should expect that each user can own one or more IDs. In turn, these IDs are owned by a private key under the user’s control. The IDs are acquired through the Blockstack naming system. First time users that log into the Blockstack application get a free `id.blockstack` in the Blockstack namespace.
Blockstack IDs are recorded on the Stacks blockchain; this means Blockstack cannot hide IDs. Blockstack IDs each have a public key assigned to them via the blockchain records that encode their operational history. This public key allows users to bind data to their Blockstack IDs through cryptographic signatures (signatures that they use when authenticating on a DApp).
Identities and *data* are application independent. An application cannot be considered a DApp unless it allows users to interact with their identities and data such that the user can later do so via a different DApp.
For example, a user that creates data in client `X` must be able access that data from a different client, `Y`, provided the client allows compatible mechanisms. Ultimately, the user has the freedom to write their own client that interacts with their own data.
Blockstack’s APIs and SDKs make it easy to build applications that adhere to this principle. Existing Blockstack applications have this property today simply because they don’t have any irreplaceable server-side logic.
In the future, Blockstack applications must continue to meet the first two principles but need not meet this one. For example, an application could encrypt data in-transit between the application’s client and the user’s chosen Gaia hub provider. In this case, unless the app divulges the encryption key to the user, the user would not have free choice of clients; the user could only use clients with which the app’s servers choose to interact. This use case could apply for users interacting with enterprise or employer-owned data.
You’ll notice the Blockstack principles avoid adherence to a particular network topology or architecture. Many DApps have defining characteristics that are implementation-specific rather than expressions of overall DApp design goals. These aspects are mentioned here as specific non-goals of Blockstack applications.
Another word for “smart contract” is “replicated state machine.” Some DApps need each peer to execute the same sequence of operations in order to fulfill their business needs (in which case a smart contract would be appropriate), but many do not.
Fundamentally, DApps should serve users by preserving user autonomy. Developers should not profit from abusive features or neglectful designs. Because Blockstack applications allow users to own their identity and data and gives them free choice of clients, any user can simply stop or avoid using bad DApps with near-zero switching cost.
