Append-Only Journaling: A Method

There's a way of journaling that doesn't have a name yet, even though several people have independently invented it. Andrej Karpathy described it in a 2025 X post, calling it the append-and-review note (@karpathy, 2025). Tony Stubblebine framed a narrower version in a 2017 Better Humans article and called it interstitial journaling (Better Humans, 2017). Paper-journal keepers have done it for centuries without naming what made it work.

This post defines the method, names its three rules, traces where the term comes from, compares it to bullet journaling and morning pages, and lists the failure modes that kill it. The label is append-only journaling. The structure is older than the label.

Key Takeaways

• Append-only journaling is a personal note-taking method with three rules: one continuous surface, every line timestamped, never edit or reorder old entries. Review happens by scrolling, not by restructuring.
• The method generalizes Tony Stubblebine's interstitial journaling (Better Humans, 2017) and matches the practice Andrej Karpathy described in 2025 (@karpathy).
• The name borrows from append-only data structures in computer science (Wikipedia), where new data can be added but existing data is immutable. Same principle, applied to personal writing.
• It is not bullet journaling, not morning pages, not a daily diary. The differences are structural and worth knowing before you commit to it.

The Definition

Append-only journaling is a personal note-taking method with three structural rules: one continuous writing surface, every entry timestamped at the moment of writing, no editing or reordering of past entries. Review is reading, not restructuring. Everything else in the practice is downstream of those three rules.

The rules:

1. One surface. Not a folder of notes. Not a tag system. One continuous file, note, or diary that everything goes into. The point is to remove the "where does this go" decision at capture time. If the answer is always "the same place," capture cost falls to one keystroke and a thought.
2. Timestamp every entry. The timestamp is part of the entry, not metadata about the entry. It can be auto-generated (notetime, a paper journal with handwritten times) or manual (the date header in a Markdown file). What matters is that every line is anchored to when, not just to what.
3. Never edit or reorder past entries. New thoughts go at the end (or top, depending on the surface). Old entries stay where they are even when they're wrong, redundant, or embarrassing. The unevenness of the log is the feature; it preserves the actual history of what you thought when, which is the only useful kind of history a journal can carry.

Notice what's not on the list. Review cadence, length per entry, format, capture trigger, choice of tool. Those are personal preferences. The three rules are the load-bearing parts; everything else can move.

Where the Name Comes From

The term borrows from computer science. An append-only data structure is one where new data can be written but existing data cannot be modified (Wikipedia: Append-only). Journaling file systems and log-structured databases use the principle to guarantee data integrity through immutability. The personal-writing version uses it for the same reason: an entry whose context can't be retroactively rewritten is more honest than one that can.

The practice predates the name by a long way. People have kept paper journals with timestamped entries and a no-erase rule for centuries. Lab notebooks in scientific work follow the same three rules and have done so since chemistry was a profession. The technique was just never abstracted out as a method with structural rules anyone could state.

The closest contemporary articulation is Tony Stubblebine's interstitial journaling (Better Humans, 2017), which applies the same three rules but specifically to project transitions in knowledge work. Ness Labs popularized it for the broader productivity audience (Ness Labs). Append-only journaling generalizes it: same structural rules, broader application; the timestamp is part of the content whether the writing happens between meetings, during a debugging session, in the middle of the night, or once at end of day.

Andrej Karpathy described the same shape in a March 2025 X post, calling it the "append-and-review note." The structure is identical, with one variation. He reorders by manually copy-pasting old entries to the top, which is a soft violation of the third rule. The canonical method does not. For more on the Karpathy version, see the writeup of his system.

Why It Works

The method works because human episodic memory is organized along a temporal axis, not a categorical one. Successful recall is accompanied by a measurable jump back in time at the neural level (Folkerts, Rutishauser, Howard, J Neurosci 2018), and the size of a person's temporal contiguity effect predicts how well they recall (Sederberg, Miller, Howard, Kahana, Memory & Cognition 2010). For the longer version of that argument, see the principled case for time as the organizing axis.

The short version. A system organized by when matches the way the brain retrieves; a system organized by category doesn't. Append-only journaling makes the time axis the only axis, which removes everything that gets in the way of that match.

The second reason is capture cost. There are no decisions at write time besides what to write. No "where does this go," no "what tag," no "which file." The fastest possible capture wins because anything slower means losing the entry. This is also why the no-edit rule matters as much as the timestamp rule. Editing is a decision too, and it's a decision that pulls you out of the writing surface and into the management of the writing surface.

The third reason is the no-edit rule itself. An entry you can rewrite is an entry your future self will rationalize. An entry you can't rewrite is contradicted by your own next entry, which is the only feedback loop that actually corrects thinking. An honest record beats a curated one, especially for journaling whose purpose is self-knowledge rather than performance.

How It Differs from Other Journaling Methods

Append-only journaling is structurally distinct from bullet journaling, morning pages, gratitude journaling, the traditional one-entry-per-day diary, and the daily-notes-per-file pattern that Logseq and Obsidian users adopt. The differences aren't aesthetic; they're load-bearing.

Bullet journaling (Ryder Carroll, public 2013, book 2018). The Bullet Journal Method centers on an Index, Rapid Logging, and Migration. At the end of each month you rewrite unfinished tasks into a new spread, which forces reassessment (Wikipedia: Bullet journal). This is the opposite of append-only. Migration is required editing of past entries; the Index is a maintained taxonomy. Both add structure that the append-only method intentionally removes. Bullet journaling fits people who want a structured task system; append-only fits people who want a structured thinking record.

Morning pages (Julia Cameron, The Artist's Way, 1992). Three pages, longhand, stream-of-consciousness, written within 45 minutes of waking, designed to be unread (Julia Cameron via NPR, 2022). Morning pages are time-boxed, length-bounded, and explicitly throwaway. Append-only is opportunistic, length-free, and explicitly kept. The methods have different goals. Morning pages are for clearing the mind. Append-only is for keeping the record.

Gratitude journaling. Topical: each day you write three things you're grateful for. The structure is the prompt, not the chronology. Append-only is chronological by default, and the prompt is whatever you're thinking right now. Gratitude journaling fits inside append-only as one possible content (a gratitude line is just a line), but its structural commitments are different.

Traditional diary (one entry per day, prose). A diary is one continuous surface, but each "entry" is paragraphs written at the end of the day, summarizing. Append-only entries are atomic lines written at the moment, not summaries. The temporal granularity is finer; the unit of writing is one thought, not one day.

Interstitial journaling (Tony Stubblebine, 2017). The closest neighbor. Same three rules, scoped to work-context project transitions. Append-only generalizes the same structural commitments to all writing where the timestamp is part of the meaning, not just the spaces between meetings.

Daily-notes-per-file pattern (Logseq, Obsidian daily notes). One file per day, often outliner-format. The chronology is preserved across files but each day is its own document, which fragments long-running threads. Append-only uses one continuous stream, so a thread you started in March is two scrolls away from the entry you're writing now. Different storage shape, different retrieval feel.

Common Failure Modes

The method is structurally simple, which makes the failures structural too. Five recurring ones, each with a specific diagnosis:

1. Treating it as a publishing surface. The moment you start writing for a future reader (your therapist, your "second brain," your hypothetical biographer), entries get longer, more explanatory, more performative. Capture cost goes up, the entry rate falls, and within a month the practice is dead. Write for nobody. The log can be read later by you or by no one; both are fine; an audience isn't.
2. Retroactive editing. "I'll just clean up that entry from last week." Once you start, the no-edit rule is broken and the log stops being trustworthy. The fix is a strict rule: corrections happen as new entries. "March 12: I was wrong about the redis lock thing in the March 8 entry. The actual problem was..." That keeps both the original and the correction in the timeline, which is the truthful record.
3. Backdating. Adding entries with a fake earlier timestamp to "make the log complete." This produces a log that's worse than no log because it asserts things about your past state that aren't true. If you didn't write it then, write it now with a note. "Forgot to log this on Tuesday. The pricing call was..."
4. Letting capture cost creep up. Adding tags, formatting, mood scores, structured templates. Each addition is small; together they erase the only advantage the method has, which is that capture is one keystroke and a thought. The right answer to "should I add a field for X" is almost always no.
5. Skipping the review. A log you never read is just a confessional. The review doesn't have to be elaborate; it just has to happen on a recurring interval. Every two weeks is a reasonable starting cadence; see the two-week note review ritual for the practice in full. The review is reading, not restructuring; if you find yourself wanting to reorganize, that's a sign the no-edit rule is being violated.

Where It Shines

Append-only journaling fits any writing where the when is part of the meaning. It does not fit knowledge bases, wikis, evergreen notes, or anything meant to be read by other people in a non-chronological way. Use a different tool for those.

The domains where the method genuinely earns its place:

• Personal journaling and daily reflection.
• Work logs and engineering decision records.
• Meeting notes across many meetings.
• Trade logs and trading journals. See logging trades by timestamp for one worked example.
• Debugging sessions, incident records, research lab notebooks. Same structural fit; same three rules. The lab-notebook tradition was the original append-only journal.

The common thread: the question you'll later ask the log is "what was I doing or thinking then," not "everything I know about X." If the question is "everything about X," use a wiki.

Frequently Asked Questions

What's the difference between append-only journaling and a regular diary?

A regular diary is usually one entry per day, written as a paragraph at end of day. Append-only journaling is many short entries throughout the day, each carrying its own timestamp, with no editing of past entries. The temporal granularity and the no-edit rule are the structural differences.

Do you ever delete entries in an append-only journal?

Almost never. Corrections happen as new entries; embarrassing entries stay where they are. The exception is genuine accidents, like a duplicate entry or a typo caught in the moment of writing. The rule serves a purpose. It makes the log a record of what you actually thought, not a curated version.

Can you do append-only journaling on paper?

Yes. The original implementation is a paper journal with handwritten timestamps. The digital version is faster (no manual time entry, full-text search), but the method itself is medium-agnostic. What matters is the three rules, not the surface.

Is this the same as interstitial journaling?

It's the same three rules applied more broadly. Interstitial journaling (Tony Stubblebine, Better Humans, 2017) specifically frames the practice around project transitions in knowledge work. Append-only journaling generalizes it to any writing where the timestamp is part of the content.

What apps support append-only journaling?

The method works in any append-only writing surface: a single Apple Notes note, a paper journal, a Markdown file, a daily-notes plugin in Obsidian or Logseq. notetime is built around the method natively (every line auto-timestamps; the model is one continuous diary). The method is not the app.

The Short Version

Append-only journaling is a method, not a tool. Three rules: one continuous surface, every entry timestamped at the moment of writing, never edit or reorder past entries. The cognitive science backs why it works. Several independent inventions have validated the structural shape. The failure modes are predictable.

If the writing you do most often is time-anchored, the method gives you a record that matches the way you actually remember. Adopt the three rules in any tool that supports them; the noun is now defined.