Discussion: Handles

[wesbiggs]

Should the DSNP specification define rules and semantics of handles? Currently, handles are left as a DSNP system-specific concern, see for example Frequency's handle system. Non-DSNP systems have different approaches to handles, such as DNS-rooted handles in AT Protocol.

What is the minimal level of specification within DSNP that would bring handles into the protocol?

• Should handles be required to be universally unique addresses?
• Can a universal namespace for handles be mandated without an assumption of there being a single consensus system?
• Can the specification support a plurality of implementation approaches to handles?
• Where do handles come into play in the spec? (Activity Content Note, Profile?)

[wesbiggs]

Community discussion 2025-04-03:

• Review of Frequency implementation
• Possibility of allowing non-suffix handles for trademark holders? Similar challenges to what ICANN faces with domain names
• Domain-based approaches change dynamics from non-transferable to transferable; subdomain-based approaches require an intermediary controller (the domain owner)
• Usability/safety concerns like homoglyph protection
• Interaction with local authorities on legal trademark ownership etc.
• Possibility of some discovery via DNS (e.g. cocacola.com can assert their handle through TXT records or .well-known files)

Should handles be in the spec?

• Handles are themselves a form of speech/communication

• Inclusivity of Unicode languages is a key principle, this could be a guideline even without specifying handle generation syntax

• Important to not view through a US-centric lens

• Perhaps handle systems are optional: "If you have a handle system, then ..." e.g. (1) must be unique, (2) must allow known language characters...

• May have an optional operation "get handle for User Id"