Designing for expansive writing
November 5, 2024
Category:
Speculative
The cursor blinks; your thoughts zig zag. A quote glimmers in one tab, an unfinished PDF lurks in another, and a screenshot hovers on the desktop waiting to matter. Yet the page in front of you insists on straight lines—as if mind and text shared a single rail. Great writing tools shouldn't be primarily about headings, fonts, or bullet gymnastics; they’re about interactions that let thinking loop, branch, and re‑join without losing its pulse. What if the page itself could bend as easily as the idea?
When a new writing project starts, we begin with a blank canvas and a scattering of references. we highlight a statistic, clip a chart, paste a link—then watch each fragment drift into separate corners: one in a citation manager, another in a footnote, a third in a folder named _misc‑research_, in a true patchwork . Tangent questions pop up on cue, but pinning them to the sentence that sparked them feels awkward, so you collect sticky notes and neon comments like breadcrumbs you’ll never retrace.
Versions multiply. Context evaporates. _draft‑1_, _draft‑1‑final_, _final‑final‑real_. The document splinters into comically named doubles, obscuring the version history and disrupting the continuity of the work. Without integrated version control, assessing what changed—and why—becomes a scavenger hunt. The relationships between edits, the rationale behind rewrites, the narrative evolution of the document—all dissolve into isolated files with no semantic linkage. But what if a writing tool could preserve this continuity by mapping the document’s lineage, capturing forks, re-merges, and tangents as naturally as the mind produces them?
Writing is less a straight path than a winding trail of half‑sketched diagrams, sudden epiphanies, and questions blurted out at midnight. A draft sprouts from pencil scrawls, thickens with layered ideas, absorbs a friend’s margin note, then veers again when an offhand remark turns into a thesis. The workspace that hosts this stop‑and‑start pilgrimage must be porous enough to catch every trace yet supple enough to let the next detour unfold. If we design only for glitzy features and forget this organic rhythm, we end up with tools that celebrate technology more than thought.
As computing pioneer Douglas Engelbart warned, “The better we can define the tools we need, the better our tools will serve us as extensions of our human capabilities.” His reminder still rings true: before racing to integrate the latest tech, we must first imagine the quiet, humane interactions that will let thinking leave its whole breadcrumb trail behind.
The record, if it is to be useful to science, must be continuously extended, it must be stored, and above all it must be consulted. It is this consultation that is now hampered because we have lost the ability to link, to trace, to follow in the wake of thought.
Vannevar Bush, As We May Think (1945)
Tools for Thought - the old desire for augmented tools
The ambition to create tools that amplify human intellect, whether through writing systems, symbolic logic, or physical artefacts, has been a longstanding interest, framing a substantial field for exploration today in strategies for cognitive augmentation through interface design.
It can be tempting to approach innovation in software development by keeping it up to the latest popular technologies, and the implementation of large language models in nearly every text editor is perhaps another proof of that. However, this fixation might be stifling a deeper and more fundamental exploration of what would actually support the divergent way we process and synthesise information.
While AI integration certainly expands the toolset, it is still an external element to the immediacy and serendipity of human thinking. In order to support users in building structured understanding about new topics it is fundamental that the user critically engages with the new domain of knowledge, and thus, designing interactions that aid thought structuring can be far more valuable than AI integrations alone.
It might seem that I’m making an argument against the least controversial premise of AI, but that’s not the case. What I question is the idea that AI can replace the deeply personal and intimate processes we rely on to deconstruct information—processes uniquely tied to our individual ways of thinking and essential for building genuine understanding. I would argue that, as long as our tools continue to overlook fundamental, supportive interactions that foster an immediate and intuitive connection between information and the user, AI-assisted chats risk becoming little more than entertaining ornaments.
To set a compelling path forward, we can look to the foundational work of early computing pioneers whose vision was to create interfaces that align with the natural ways people think—fluidly, intuitively, and non-linearly. Howard Rheingold’s Tools for Thought (1985) captures this original ambition, focusing not on automating thought but on empowering it. Figures like Douglas Engelbart, Vannevar Bush, and J.C.R. Licklider, as discussed by Rheingold, envisioned computers as tools to augment human intellect, fostering exploratory, divergent, and collaborative modes of thinking.
To view a writing app as merely a digital sheet for depositing text reflects a limited, Cartesian perspective that separates the mind from the tools it uses, overlooking how closely integrated cognitive processes and digital interfaces are. Modern thinkers, however, encourage us to move beyond this dualism, emphasising that we operate within a dynamic integration of mind, body, and environment.
Research keeps showing that the things we pick up—pens, phones, even the chair we’re sitting in—slip quietly into our sense of self. Maurice Merleau‑Ponty liked to illustrate this with the _blind man’s cane_: after a few minutes, the tip is no longer a stick you hold but the point where the world begins to “touch back.” Across the Rhine, Martin Heidegger staged a similar scene in his Black Forest workshop, pointing to a carpenter’s hammer and noting how it disappears from awareness the instant it is “ready‑at‑hand.” Designing a writing tool, then, is less a technical project than a matter of choreographing these disappearances—building interfaces that vanish into the very act of thinking.
The history of human creativity is, at its heart, the story of turning resistance into hospitality. We plant seeds in land that once refused us food, tame animals that once fled or fought us, and hammer raw ore into utensils that fit the curve of a hand. _Every domesticated plant and animal, every tool, every utensil, every appliance, every work of art,_ philosopher John Dewey reminds us, _represents the conversion of environments once hostile or indifferent into conditions that actively favour human activity._ In other words, design at its best doesn’t merely decorate the world—it recruits the world as an accomplice to our thinking, our making, and our living.
Design Goals
In this speculative design proposal, I aim to address some key limitations of existing writing software by exploring framework-driven tools that directly support the pain points described above. Specifically, this proposal focuses on developing contextual margin annotations that allow users to capture spontaneous thoughts as they arise and seamlessly navigate between the main writing block and reference materials. Additionally, it proposes an integrated text version management at the level of the word, sentence, and paragraph within a single document, enabling users to review and compare different drafts while preserving context and narrative continuity.
Integrated Version Control for Iterative Writing
Writing is an inherently iterative process, with each project representing a sequence of considered drafts and refinements. This design supports a more fluid iteration process by enabling version management at multiple levels—from individual words to entire paragraphs—without requiring users to leave the active document. By providing an integrated version control system within the file, writers can view and assess each iteration within the broader context of their project, along with annotations and references. This approach minimises the need for duplicate files and brings all variations together in one dynamic workspace, streamlining the incremental development of ideas and text.
Word-level versions enable the management of small changes, making edits visually clear without requiring additional clutter or fragmented files and notes.
When a version history is available for a paragraph, a menu icon appears at the bottom of the text block. Opening the menu allows users to switch between versions or, by pressing Command and selecting a version, open it in a side window for easy comparison.
Alternatively, the version history icon can be toggled to open a side card view displaying available versions, allowing for an easy transition between them.
At the bottom left of each paragraph block, users can create a new version, which opens in a floating window within the writing area, keeping the current version visible for easy reference.
Support contextual tangent thinking
Effective writing software should support the spontaneity and complexity of thought, allowing users to capture and organise ideas in context as they arise. Tangents encompass all thoughts or resources related to specific parts of the text—be it a word, sentence, or section—providing a dynamic way to engage with ideas without disrupting the main narrative flow. This approach preserves continuity while enabling users to explore connections between their main writing material and supporting ideas in a non-linear way.
To achieve this, the design proposes three complementary strategies: margin notes, in-line comments, and trails.
Margin Notes
Margin notes act as a flexible space for quick, spontaneous thoughts directly connected to specific parts of the text, much like traditional marginalia in books. These notes sit alongside the main text, making them easily accessible without intruding on the primary narrative flow. Designed for capturing open-ended ideas, relevant links, or reflections, margin notes align with their reference point in the text, preserving context and continuity enabling users to revisit ideas without losing their place in the main document, supporting a non-linear exploration of ideas.
Annotations on both highlighted and non-highlighted text are displayed similarly, but comments linked to highlighted segments are color-bound to maintain a clear association with the referenced text.
The writing window can be expanded or compacted by toggling the side panel, which displays all text interactions. When the panel is closed, clicking on a highlighted section marked with a comment icon opens an in-block view of that note, keeping the annotation accessible without disrupting the main text.
In-Line Comments
In-line comments offer an embedded, editorial-style note-taking feature. These notes appear within the main text but can be shown or hidden as needed, allowing users to annotate specific words, phrases, or sections without cluttering the reading experience. Ideal for content that requires direct attention, in-line comments support on-the-fly feedback or personal editing notes, as they don’t appear in print-ready formats. This feature enables a flexible, layered approach to revision and refinement, offering an intuitive way to manage edits without disrupting the primary text.
Trails
Source trails provide a seamless way to incorporate external references directly into the writing space, allowing users to connect media, clipboard content, and highlights directly from their sources. By dragging these elements into the document, the app generates an automatic link back to the original source—using transclusion, which embeds a selection from one document within another, creating a navigable “trail” between the two. Inspired by Vannevar Bush’s Memex and Ted Nelson’s Xanadu System, this feature preserves the context of each reference, enabling users to trace foundational materials effortlessly while building a richly layered structure that supports multifaceted exploration within their writing.
Afterword
In an era where we increasingly feel destined to become operators of creation machines, we risk designing our work environments with the deterministic belief that they are meant more for operating than for creating. It is precisely at this point that our capacity for deeper, more primal encounters with truth and meaning becomes constrained, reducing our actions to mere utilitarian functions.
Reflecting on my own and others’ frustrations with the limited cognitive scope of current writing tools, I have turned my focus toward interactions that feel as natural as pen on paper—how we instinctively strike through a line, underline a thought for emphasis, jot a quick annotation in the margins, or fold the corner of a page to mark our place. In this essay, I aim to illustrate how interactions rooted in familiar behaviours, drawn from both digital and physical tools, can serve as human-centered mediums that support our natural cognitive processes in exploratory writing tasks.
The allure of implementing new technologies will always capture our attention. However, it is within the nuanced layers of an environment that meaningful, embodied experiences are truly built. In this way, design must thoughtfully and inquisitively examine what genuinely defines novelty within a given context.