digital gardens are graveyards without a gardener

by Ray Svitla

around 2020, a beautiful idea spread through the internet: what if your personal website wasn’t a blog — polished, chronological, performative — but a garden? a place where ideas grew over time, connected to each other, imperfect and evolving. a digital garden.

Maggie Appleton wrote the definitive history. she catalogued dozens of gardens — personal wikis, linked note collections, public Zettelkastens — and articulated the philosophy behind them: “tools for thought are cultural practices, not computational objects.”

it was genuinely lovely. and almost all of those gardens are dead now.

what happened

go check. pick any list of digital gardens from 2020 or 2021. visit the links. count how many were updated in the last year.

the number is grim.

digital gardens died for a specific, mundane reason: gardening is work. real gardens require daily attention — watering, pruning, weeding, transplanting. digital gardens require the same: revisiting old notes, updating connections, removing dead links, expanding seedlings into full ideas.

nobody has time for this. not consistently. not at the scale a garden requires to stay alive. life intervenes. priorities shift. the garden gets one more “I’ll get back to it next week” and then two years pass and the garden is a graveyard with a beautiful CSS theme.

Andy Matuschak — probably the most committed public note-taker alive — described the problem implicitly in his evergreen notes: “better note-taking misses the point; what matters is better thinking.” but better thinking is a continuous practice, and continuous practices compete against everything else in your life. the garden loses.

the maintenance problem

the digital garden failure is a specific instance of a general problem: systems that require continuous human maintenance don’t survive contact with human life.

blogs fail for the same reason. personal wikis fail for the same reason. CRM systems that require manual entry fail for the same reason. any tool that depends on the user consistently doing a boring task will eventually be abandoned when something more interesting happens.

Zettelkasten has the same issue at scale. Luhmann maintained 90,000 cards over a career because it was his job — his research method was his livelihood. for everyone else, the maintenance overhead eventually exceeds the perceived value.

the second brain movement inherited this problem wholesale. millions of Obsidian vaults opened in enthusiasm and abandoned in guilt.

the gardener thesis

here’s where AI changes something genuinely. AI can be the gardener.

not a replacement for human thinking — a maintenance layer. the boring work that kills gardens: checking links, suggesting connections, flagging stale content, proposing updates, pruning dead branches. AI can do this. continuously. without getting bored. without competing priorities.

an AI gardener could:

• notice that a note from two years ago connects to something you wrote yesterday
• flag a seedling idea that you mentioned three times in journals but never developed
• suggest pruning a section that contradicts your current thinking
• identify clusters of ideas that want to become essays
• keep the infrastructure alive while you do the thinking

this is prosthetic gardening. the human provides the seeds — the ideas, the observations, the half-formed thoughts. the AI does the weeding and watering. the garden stays alive because the maintenance is automated while the creation remains human.

the missing ingredient

but here’s the thing: a gardener without taste destroys gardens.

an AI that prunes, connects, and maintains your digital garden based on… what? what stays? what goes? what connections matter? what seedlings are worth developing? these are not objective questions. they’re questions about you — your values, your interests, your current direction, your evolving sense of what matters.

a content-aware AI can see that two notes are semantically similar. only an identity-aware AI can see that one of them represents a value you’re moving toward and the other represents a value you’re moving away from. the first should be cultivated. the second should be composted.

this is the identity problem again. AI can be the gardener, but a gardener needs to know what the garden is for. and the garden is for you — which means the gardener needs to know who you are.

a digital garden without a gardener is a graveyard. a gardener without identity awareness is a landscaper — making the garden look nice without knowing what should grow.

Appleton’s insight, extended

Maggie Appleton was right that tools for thought are cultural practices. she was right that the digital garden is a better metaphor than the blog. she was right that ideas should grow, connect, and evolve rather than being published as finished objects.

what she didn’t connect — because the technology wasn’t there yet — is that cultural practices need practitioners, and practitioners need to know who they are.

a garden is an expression of the gardener’s values. Japanese gardens express different values than English cottage gardens. permaculture gardens express different values than ornamental ones. the garden’s form follows the gardener’s identity.

digital gardens never had this connection. they were collections of notes — interesting, linked, sometimes beautiful — but not rooted in a formal identity that could guide their growth. the gardener’s taste was implicit, never formalized. and when the gardener left (because life), the implicit taste left too, and there was nothing to guide the garden’s maintenance.

self.md is the formalization of the gardener’s taste. it tells the AI gardener: here’s what matters. here’s what I’m cultivating. here’s what I’m composting. here’s how my values are shifting, which means here’s how the garden should shift too.

the new garden

imagine a digital garden where:

• you add ideas whenever they occur to you — no formatting pressure, no organizational anxiety
• the AI gardener connects them, maintains them, prunes dead branches
• your self.md tells the gardener which ideas align with your current direction
• the routing layer surfaces relevant garden content when you need it, based on your identity state
• the garden evolves with you because the gardener knows who you are and how you’re changing

this garden doesn’t die when you’re busy for two months. the gardener keeps maintaining. the connections keep growing. and when you return, the garden has evolved in directions that make sense because they were guided by your identity, not by arbitrary algorithms.

is this utopian? maybe slightly. but the components exist. AI can maintain. identity protocols can guide. the missing piece was always the connection between the garden and the gardener’s self-knowledge.

the graveyard question

the digital garden movement failed not because the idea was wrong. the idea was beautiful. it failed because it demanded something unsustainable — continuous human maintenance at scale — and offered something too abstract — “better thinking over time” — as reward.

AI solves the sustainability problem. identity solves the direction problem. together, they might make the digital garden actually work.

or they might create something new that we don’t have a name for yet. not a blog, not a wiki, not a garden. a living knowledge system that grows with you because it knows who you are.

what would your garden grow if it knew who the gardener was?

→ why your second brain is dead — the broader storage failure → every PKM framework is wrong — gardens in context with other approaches → identity as protocol — the gardener’s self-knowledge → cognitive prosthetic vs crutch — AI gardening as prosthetic

Ray Svitla stay evolving