Built a force-directed graph of my research lineage in one HTML file, more serendipitously than deliberately, from scratch, no D3, no React, no build step, no dependencies.
About 70 nodes, about 100 edges. Real publications, real DOIs, real relationships: citational, methodological, generational.
This was serendipitous. Slipped past the abstraction layer, got close to the mechanism, and the next thing I knew Coulomb repulsion and Hooke springs were carrying my scholarly corpus around the screen.
The constraint that mattered was archival durability. If a browser cannot still render the artifact in 2045, then for my purposes it failed.
It runs at FCP 0.768s, LCP 0.768s, TBT 0.027s, with two requests total. Keyboard traversal, ARIA, screen reader support, battery-aware freezing on mobile.
I have written HTML for a long time, but only recently started working this close to primitives. They feel like a dusty old book, neglected, unfashionable, and still full of load-bearing ideas. It also works on a shit phone.
I am not against the frameworks. I like frameworks and NPM, I just want the artifact to outlive them...At some point it stopped feeling like a twenty-pound dictionary and started feeling like a wrench.
To preserve computational scholarship, a durable method is needed to package data, analysis, and visualization into a single, executable artifact. Current approaches, however, rely on a fragile ‘wet-nurse’ model of external dependencies: APIs, frameworks, and platforms that inevitably deprecate, churn, and sunset. This paper proposes a zero-dependency HTML architecture, a ‘digital paratype,’ that is born weaned: it runs in any standards-compliant browser, with no runtime installation or external fetches
My name is Tim Jones. I am a field botanist who always winds up as the IT guy. After rescuing too many servers and rotting projects, I wanted to build something archivally bulletproof. I welcome you to do the math on this.
I’ve spent my life documenting species that need to outlive me, sometimes with a twelve-foot alligator two feet off my nose. In botany, we value the specimen: a physical, verifiable record designed to survive centuries.
When I turned to data architecture, I found modern web tooling to be the opposite. Most systems are brittle stacks that fail when a single dependency shifts. I wanted to build what I call a Digital Paratype. In taxonomy, a paratype is a supplementary specimen supporting the primary record. This is a research paper that functions as a permanent, interactive instrument without external dependencies.
The Stack: Vanilla JS (ECMAScript) and CSS3. No frameworks, no NPM, no CDNs, no build steps. The Weight: ~40 KB for the technical methodology; ~325 KB for the live analysis. The Performance: 100/100/100/100 PageSpeed scores—standard-compliant logic.
The Features:
Instant State Toggling: Millisecond latency for complex data views.
Deep-Linkable State URLs: Share a specific analytical view without reloads.
Self-Auditing: The document resolves itself for durations, text, and metadata at load time to stay evergreen.
The Archive: A seven-year longitudinal Michigan Cannabis market study, manually transcribed for fidelity.
The Millennium Math: The limitations section surfaced a constraint that materially shaped the design: The 5MB Ceiling. By maintaining a linear growth rate of 5.3 KB per year and at the current data payload of 34 KB for seven years- I encourage anyone to do the math on this.
I’m looking for feedback on UI responsiveness and whether a zero-dependency, monolithic approach like this is the only viable path forward for archival academic publishing.
About 70 nodes, about 100 edges. Real publications, real DOIs, real relationships: citational, methodological, generational.
This was serendipitous. Slipped past the abstraction layer, got close to the mechanism, and the next thing I knew Coulomb repulsion and Hooke springs were carrying my scholarly corpus around the screen.
The constraint that mattered was archival durability. If a browser cannot still render the artifact in 2045, then for my purposes it failed.
It runs at FCP 0.768s, LCP 0.768s, TBT 0.027s, with two requests total. Keyboard traversal, ARIA, screen reader support, battery-aware freezing on mobile.
I have written HTML for a long time, but only recently started working this close to primitives. They feel like a dusty old book, neglected, unfashionable, and still full of load-bearing ideas. It also works on a shit phone.
I am not against the frameworks. I like frameworks and NPM, I just want the artifact to outlive them...At some point it stopped feeling like a twenty-pound dictionary and started feeling like a wrench.
Live: https://tjid3.org/test/kg12.74
DOI: https://doi.org/10.5281/zenodo.19337250