Identifiers and identity (part 3)

In this third and final part, I will use the data model described in part 1 to explain why the term “synonym” in the resolution spec is misused.

Part 1 of this series explains that our identity model contains a means of associating an identifier with an entity, where identity is the set of characteristics that distinguish one entity from another. In XRI land, the entity is the XRI authority, where the single characteristic that distinguishes one authority from another is the CanonicalID. The identifier is the XRI that is passed to an XRI resolver as the means for associating that identifier with its authority.

If you examine the diagram attached to part 1, you will notice that the concept of synonymity appears only on the left side of the diagram (in identifier land.) This is because the concept exists only within the framework—the identity model—of the XRI client application. For example, if a client gets back the same CanonicalID when resolving =steven.churchill and @ootao*steven then the client is free to use an identity model that considers these identifiers as synonyms for a single identity. (And thus, for example, the client is free to store the CanonicalID as the PK for its user account.) But many client applications might consider this notion absurd and instead have different notions of identity and synonymity—or none at all.

The Resolution Specification makes the conceptual error of using the term “synonym” in regard to the right side of the diagram (in identity land). This is in an attempt to categorize a set of the authority’s data elements and call them “synonyms”. The spec incorrectly refers to CanonicalIDs, LocalIDs, Refs, BackRefs, MapToIDs, and the like, as synonyms. The terms use in this context is arbitrary and obfuscating.

Like beauty, synonymity is in the eye of the beholder. Only the client application can determine the synonymity model, and the client application only applies this notion to identifiers.

(Okay, there is one exception to this rule. It is okay to think of a “local identifiers” such as *steve and *steven” as being “local synonyms” within the same registry namespace. This notion occurs on the right side of the diagram.)

Conclusion

In three parts I have attempted to present the XRI authority model and dispel some of the confusion surrounding identity, authority, synonymity, and so forth. The irony is that despite the ongoing confusion in the TC and in the spec, the essence of the described identity model still stands as the underlying foundation of XRI resolution. This makes XRI resolution an elegant and exceptionally strong framework for solving many real-world problems for years to come.

Will the concepts I’ve presented here be ignored by the XRI TC (as they have continually over the last year)? Yes: I have little doubt about that. I believe, nonetheless, that it makes absolute sense for me to articulate these ideas at least once on the public list.

‘Nuff said. I will now disappear back into the woodwork.

~ Steve

xri message