Just-In-Time-Trees (JITTs): Next Step in the Evolution of Markup?

Recording multiple possible encodings for a text has been treated as a problem of syntax (CONCUR, milestones, stand-off markup, Bottom-Up Virtual Hierarchies, Layered Markup and Annotation Language, MECS/TexMECS, etc.) and as a problem of parsing (Earley's algorithm, active chart parsing, tree-adjoining grammars). While offering various advantages and shortcomings, these efforts fall short of isolating the fundamental difficulty that gives rise to the problem. The simple tree model, which XML enforces, is a symptom and not the cause of this problem. Descriptive markup divides texts into content and markup. Or as previously stated by the authors, "Markup is metadata about #PCDATA." The fundamental problem is that all prior methods treat markup as static trees of metadata about #PCDATA. If that changes to: Trees are assertions about metadata, the fundamental difficulty of representing multiple trees in a single document instance resolves into a processing issue.