Tending
Every morning, before dawn, the beaver inspects the dam. This is not the dramatic part. Nature documentaries film construction; tending is never filmed. The beaver waddles along the dam's upstream face, finds where overnight current has loosened a stick, and repairs it. The ratio of building time to tending time, in a beaver's actual life, is heavily weighted toward tending. The dam is built once in days or weeks; it is tended every day thereafter for the rest of the animal's life. Jamie's entire practice — returning to the same landscapes across decades — is tending. Her work is the record of what maintenance looks like when performed on a cognitive-ecological scale rather than a structural one. The argument the chapter builds, and that this entry carries, is that AI frameworks, educational curricula, governance structures, and parental engagements all require tending, not building-and-walking-away — and tending is the capacity contemporary culture is least equipped to sustain.
The Privilege of Return — Contrarian ^ Opus
There is a parallel reading that begins with who gets to tend. Jamie's decades-long return to Scottish landscapes is underwritten by institutional position, publication contracts, cultural capital sufficient to make repetition artistically legible rather than economically desperate. The teacher revising her curriculum each semester may be performing maintenance work, but she is also likely adjuncting across three institutions to survive, her tending compressed into whatever hours remain after economic survival is secured.
The beaver metaphor naturalizes what is actually a distribution problem. Real organizational tending — the daily inspection, the patient revisiting — requires protected time, which requires resources, which requires power to claim them. The Berkeley frameworks propose 'protected mentoring time' as though protection were a design choice rather than a political outcome. In most organizations, tending work is what gets done by whoever lacks the status to refuse it, often women, often without the title or compensation that would make the work visible. The romantic framing of tending as 'unglamorous but essential' can obscure how its invisibility is sustained by the same systems that make construction visible — not because construction is naturally more legible, but because construction serves capital accumulation in ways that inspection cycles do not. When tending becomes a virtue rather than a structural necessity, it becomes one more form of unpaid labor the culture can sentimentalize rather than resource.
In the AI Story
The cultural asymmetry is severe. Building is legible, narratable, heroic. It ships products, generates revenue, appears in quarterly reports. Tending is none of these things. The teacher who revisits her curriculum each semester, the parent who has the hundredth conversation with a child about screens, the organization that treats its AI governance as a living structure rather than a deliverable — all perform work that is invisible to the metrics that determine what gets valued and funded.
Jamie's Cairn (2024) enacts the tending logic at book length. A cairn is built by many hands over many years, each adding a single stone. No individual contribution is dramatic. The cumulative effect is navigational — the cairn tells the next traveller where the path goes when fog obscures the landmarks. Each of Jamie's essays adds a stone to a structure her readers collectively inherit.
The Berkeley researchers proposed AI Practice frameworks — structured pauses, sequenced workflows, protected mentoring time. Sound proposals. But a proposal is a plan for a dam, not the dam itself, and certainly not the daily inspection that keeps the dam functional. The question is not whether organizations can design good frameworks but whether they will tend them.
Tending resists the AI moment's dominant affect. The tool offers constant responsiveness and rewards the dramatic intervention. Tending requires the discipline of the undramatic return. Segal's foreword describes learning this through his fourteen-second pause — the moment he realized his thinking had to catch up to his doing, and that the catching-up was itself a form of maintenance.
Origin
The concept surfaces explicitly only in Jamie's later work, particularly Cairn (2024), though the practice has been continuous since Findings. The Kathleen Jamie — On AI volume elevates it to a load-bearing term for thinking about what AI-era institutions actually require.
Key Ideas
The ratio is inverted. Construction is the overture; tending is the opera. Most of the work, across most of the life of the structure, is maintenance.
Tending produces the baseline. Long-term observation yields knowledge of change over time — the data the culture of building cannot generate.
Tending cannot be automated. The current changes daily; the inspection must too, and the inspection requires the attender's sustained relationship to the specific structure.
Tending is the capacity most at risk. The culture rewards shipping. Tending ships nothing and is therefore invisible to the systems that determine what gets valued.
Appears in the Orange Pill Cycle
Maintenance Requires Infrastructure — Arbitrator ^ Opus
The conceptual claim is correct at 100%: most of what sustains complex systems is maintenance, not construction, and AI frameworks will fail without tending. The diagnostic about cultural asymmetry is also fully right — building is legible to quarterly reports, tending is not. Where the entry and its counter-reading diverge is on what follows from this truth.
The romantic/critical split dissolves when you ask: who is positioned to tend what, and what makes that positioning possible? Jamie's practice demonstrates genuine tending, and it is enabled by institutional support that makes decades-long return economically viable. Both things are true. The teacher revising curricula performs essential maintenance, and the revision happens in the margins of survival labor. The Berkeley proposals are sound, and 'protected time' requires organizational power to protect it. The move is not to choose between celebrating tending and critiquing its distribution, but to recognize that real tending requires real infrastructure.
The synthetic frame the concept benefits from: tending is not a virtue or a practice, but a structural requirement that must be resourced like any other infrastructure. The beaver is useful not as a moral exemplar but as a biological fact — dams require inspection, inspection requires time, time requires conditions that make return possible. The AI governance question becomes: what institutional arrangements make tending materially possible, not just conceptually valued? The answer will vary by context — for some organizations it's protected time, for others it's hiring people whose job is maintenance rather than extraction, for educational systems it's funding models that reward sustained engagement rather than scalable delivery.
Further reading
- Kathleen Jamie, Cairn (Sort Of Books, 2024).
- David Edgerton, The Shock of the Old (Oxford University Press, 2006).
- Andrew Russell and Lee Vinsel, 'Hail the Maintainers' (Aeon, 2016).