Issue 1
Published 2026-10-04
Out on the first Sunday of each month
The first two shelves are both about people and their tools: Human-Automation Interaction asks what happens on the human side once work is handed to a machine, and Personal Knowledge Management asks which axioms behind note-taking tools outlast the tools themselves.
Each is ordered by the argument rather than by publication date.
From here on, the site updates on the first Sunday of each month, and every update adds at least one new lab case.
§ Task allocation and its paradox
function allocation
The 1951 Fitts list splits system functions into "humans are better at" and "machines are better at", which later readers took to mean assigning each function to whichever side is stronger.
founding paradox
Automation set out to replace an unreliable human and ended up handing the hardest, most critical residual work to a human whose skills are decaying.
function allocation
Level of automation is a continuous scale, running from the human doing everything to the computer doing everything and not telling the human.
§ The cost on the human side
monitoring
When a person watches a signal source that rarely changes, detection rate declines over time; in Mackworth's experiment it had dropped measurably within about thirty minutes (an order of magnitude, reported secondhand).
trust and reliance
Complacency and bias are two overlapping phenomena with attention at their core.
situation awareness
Automation turns the operator into a bystander, and situation awareness and hands-on feel drain away together.
situation awareness
Situation awareness has three levels: perception (what is happening), comprehension (so what), and projection (what happens next).
workload measurement
NASA-TLX takes mental workload, which cannot be measured directly, and breaks it into six ratable dimensions: mental demand, physical demand, temporal demand, own performance, effort, and frustration.
§ The way out by design
trust and reliance
People decide whether to use automation on the basis of trust, and trust can be mismatched.
trust and reliance
The relationship between people and automation splits into four categories: appropriate use, and three ways it goes wrong: misuse (over-reliance), disuse (distrusted and left idle), and abuse (pushed through on the design side with no regard for human factors).
function allocation
The division of labour does not have to be fixed at design time. It can move up and down during operation with load, state and performance.
§ First ask: why externalise knowledge at all
history of ideas · 3 min read
The Memex is a hypothetical personal information desk proposed in 1945: microfilm storage for a person's entire library and records, with entries chained together by associative trails.
history of ideas · 3 min read
The word "hypertext" is defined in Nelson's 1965 paper, meaning interconnected text that paper cannot present.
§ How to capture: three philosophies of the atomic note
note-taking · 3 min read
A Zettelkasten is a networked accumulation system.
note-taking · 3 min read
The Cornell system divides a page into three zones: the right column for live capture, the left for questions drawn out afterwards, and a band at the foot for a summary in your own words.
note-taking · 3 min read
Evergreen notes are a set of disciplines: one concept per node, organised by concept rather than by source or topic, densely linked between nodes, and iterated on each revisit rather than rewritten.
§ How to organise and navigate
organisation · 3 min read
The first question in personal organisation is not what topic this belongs to. It is when you will need it.
workflow · 3 min read
Material you collect goes through four steps before it becomes finished work: Capture keeps only what resonates, Organize files it by actionability, Distill marks the essentials layer by layer, and Express assembles output from intermediate packets.
workflow · 3 min read
Personal kanban has two rules: visualise the work, and limit how much of it runs at once. A to-do list answers what needs doing.
documentation · 3 min read
Documentation usually fails for structural reasons rather than stylistic ones: four different reader needs get pushed into one document.
navigation · 3 min read
A map of content is a note that references other notes: a hub for gathering, developing and navigating rather than a container.
§ Does it actually work: the scientific audit
empirical science · 3 min read
Personal information management is an empirical science, not productivity folklore.
empirical science · 3 min read
Self-tracking systems pass through five stages: preparation, collection, integration, reflection and action.
1 correction this issue; see the correction record.
14 new links in the knowledge graph.
Next opening 2026-11-01: Diffusion & Networks · Situations & Heuristics · Growth Analytics; members can read from 2026-10-21. Members read early ▸