Contribute
Open questions
Uncertainties still under investigation, and questions raised by published work. Reproduce an experiment, contribute a case, or challenge an explanation.
From published work
Backlog
Items not yet started. Each has a permanent URL and intended contribution.
What changes when AI can pursue an assignment?
A practical distinction among responding, following a workflow, and choosing steps toward a goal
How should people and agents share the work?
Collaboration patterns for exploration, execution, review, handoffs, and disagreement; attention to human attention and retained expertise
What does evidence actually establish?
Distinctions among source fidelity, truth, coverage, reasoning, and uncertainty; treatment of shared errors among reviewers
When is independent action justified?
A contextual delegation policy using demonstrated performance, consequences, authority, reversibility, review costs, and feedback
What makes knowledge work good?
Distinctions among accuracy, usefulness, originality, timeliness, consequences, and effort; who gets to judge
What is the relationship between information, knowledge, and work products?
An account of sources, interpretations, reusable understanding, and artifacts without assuming one universal definition of knowledge
How do we discover the question worth asking?
An approach to problem framing, ambiguity, reframing, and deciding what to investigate
How do we develop explanations and create alternatives?
Ways to move from observations to hypotheses, designs, and competing possibilities; room for creativity
How do we decide when evidence cannot settle the answer?
Distinctions among factual uncertainty, preferences, trade-offs, and responsibility; ways to judge when enough investigation has been done
What does an agent need to know?
A map of task context, domain context, preferences, history, permissions, and missing information; ways to obtain tacit context
How does shared knowledge remain useful as things change?
Practices for provenance, contradictions, freshness, revision, retrieval, access, and correction; an experiment testing reuse
How do we learn from outcomes that arrive late or have many causes?
Distinctions among artifact checks, experiments, forecasts, causal explanations, and decision quality
What changes when several people and agents work together?
An examination of coordination, incentives, authority, conflicting goals, local optimization, and process design
What is the smallest useful practice someone can adopt?
A repeatable starting method that fits one real assignment and avoids unnecessary overhead
When does structure help, and when does it become bureaucracy?
A comparison of effort, work quality, review burden, flexibility, and reuse across different levels of structure
What would make this an approach that others can learn and improve?
Conditions for transferring practices across contexts; ways for readers to contribute cases and challenge claims
The questions and concepts so far
Working definitions, the topic map, tensions, and gaps
What changed my mind
Earlier position, new evidence or experience, revised position, and unresolved uncertainty
One task, examined from beginning to end
A small investigation into what an agent needs—and where human judgment enters.
Compare two ways of delegating similar work
Compare an ordinary request with a lightweight explicit assignment. Record quality, total effort, review burden, failures, and any constrain…
Revisit work using preserved knowledge
Return to an earlier assignment after new evidence appears. Test whether retained sources, assumptions, and decisions help revision.
Return to a prediction or decision
Record a specific expectation, resolution conditions, assumptions, and decision rationale. Revisit when evidence arrives.