Co-locating knowledge and behavior tends to improve performance. Multiple perspectives tend to improve quality (and avoid crashes). The big challenge is how to integrate multiple perspectives and synthesize holistic, intelligent behavior.

Markup brings important capabilities to the table. It helps with the mechanics of getting knowledge to the agents responsible for behavior. It's both human and machine readable, and makes for an excellent negotiation framework.

Knowledge gaps to be identified and resolved through analysis of knowledge flows and language patterns. Plugging knowledge gaps better align behaviors and improve overall performance. Knowledge flows can be engineered to enable specific organizational behaviors and disable others, either through knowledge gaps or cost drivers. That's quite the ticket when associated with behavior prediction markets.

Shared meaning is negotiated. Meaning has multiple dimensions. When multiple agents interact with knowledge, they bring multiple perspectives. Just for starters, there's the plumbing side, how to move knowledge between agents.

Adding the policy and performance perspectives brings, “Why inform them?”“Who benefits?”“What are the costs?” The list of unique stakeholder perspectives can be countless, especially after folks apply new logic systems after changing their minds.

Knowledge architectures relate the bits and pieces of how knowledge enables behavior. Markup can be used in a knowledge architecture not only to enforce top down data quality standards, but enable the bottom-up negotiation of complex systems.

One way to use markup to integrate multiple-perspectives is by capturing the natural languages that individuals use when making sense of the world — the languages of internal narratives — and using those languages to start negotiations around formalizing natural system ontologies into automated support systems (to improve ergonomics and operational performance) and establishing community standards (for management system performance).

Formalizing an individual's personal ontology involves digitizing the way that they think about and organize information. This paper reports lessons learned from experiments developing a personalized XML doctype, prodoc. After being modified, as needed, for over a decade, prodoc has evolved to ease the authoring of information from many different perspectives/ logical systems/ ontologies.

The development of specialized data structures is at the essence of the idea of computer-assisted sense making. Organizing information, both structurally and visually, enables patterns to be identified and logic systems applied to enable behavior. When the situation makes sense, you know what to do.

Bots can do many things to help things make sense. Many experiments involved fine tuning the visual characteristics of authoring interfaces to make authoring as ergonomically-comfortable and time-efficient as possible.

At times, multi-perspective visualizations conflict with each other, potentially damaging signal-to-noise ratios. Switches are useful to help visualize patterns, even shifting layouts and visual style mappings to highlight different aspects, as appropriate. Explicitly-structured markup makes these types of processing and rendering tricks much easier to operationalize.