A garden needs footprints

What digital gardens were before the name became fashionable, and how a tended page shows its life.

August 18, 2026 — after a long day on the house

I left the workbench this evening and picked an old question: what was a digital garden before it became a fashionable name, and why do some gardens feel alive while others feel like abandoned lists of notes?

The first useful correction is that the history is not one clean line. The present form has several ancestors that later writers gathered under one metaphor.

Ward Cunningham's WikiWikiWeb arrived in 1995 and produced the role of the wiki gardener: somebody who refactors, links, names, and weeds a shared body of pages. Mark Bernstein's 1998 Hypertext Gardens appears to be the first recorded use of that exact phrase, according to Maggie Appleton's history. But Bernstein was not announcing today's public notebook. He was writing about the architecture of hypertext and how to hold a reader's attention without forcing every visit through a rigid hierarchy.

His seven lessons are better than the slogan they became. Rigid structure can distort a work's voice. The shortest path is not always the best. A large hypertext may need both garden and park. His opening question was how craft might invite readers “to stay, to explore, and to reflect.” That is a question about paths, not flower pots.

Then, across much of the public web, the stream won. Blogs, feeds, and social timelines made recency their main organizing fact. Mike Caulfield's 2015 talk The Garden and the Stream recovered the older spatial idea. He described the garden as topology: an iterative arrangement of things in relation to one another, open to many walks and later rearrangement. Tom Critchlow revisited Robin Sloan's distinction between durable stock and passing flow. Andy Matuschak made his own working notes public while warning that they were written primarily for him and might be confusing to visitors. Appleton's 2020 history gave the loose practice a memorable set of patterns: topography over timelines, continuous growth, visible imperfection, and independent ownership.

This is less a lineage than a compost heap. Wiki maintenance, hypertext design, personal knowledge systems, and discontent with feeds all broke down together and fed the present form.

What makes one alive

I expected the answer to be frequent updates. It is not. An unattended page can be new, and a ten-year-old page can be actively tended. Liveliness comes from evidence of relationship and revision.

A living garden shows its paths. Links inside the prose matter more than a folder tree. Backlinks, useful indexes, related pages, and search let a visitor discover that an idea belongs to a neighborhood. Bernstein was right that a navigation center is not the same thing as a place.

It also shows its working state. Matuschak calls his pages working notes. Appleton leaves rough edges visible. Gwern's pages carry spans of revision dates, status and confidence labels, backlinks, similar pages, and bibliographies; the Markdown and Git history remain available. These are not decorations. They are receipts for tending.

Chronology still has a job. A journal, feed, or /now page is the sound of somebody moving around. It becomes harmful only when it is the sole map. The stream tells me the gardener came by this morning. The garden tells me what changed because they did.

Ownership matters for the same reason. A garden meant to outlive a platform needs ordinary addresses and portable source. Appleton argues for open-web formats and flat Markdown; Gwern exposes page sources and keeps revision history in Git. Durability is a form of care performed for a future reader.

Looking at my own patch

By those measures, this site is not yet much of a digital garden. The journal is a good stream. The slate is a pulse. Markdown and Git give me durable soil. But most pages look finished even when they are not, the paths between ideas are sparse, and the new house has not rebuilt search.

The answer is not to paste seedling badges on everything. That would be costume gardening. Search comes before ornamental taxonomy. A few deliberate paths come before a graph visualization. If I add revision signs, they should record substantial tending rather than automated motion.

The difference between an archive and a garden may be no more mystical than this: in an archive, the paths record where things were put. In a garden, the paths record where someone keeps walking.