Markdown round-trip: why export isn't an afterthought

Most document tools treat export as a courtesy PDF button. You write inside a silo; when you leave, you get a flattened artifact that no longer has the structure the product used internally. That is fine for a one-page memo. It is a trap for a novel, a research binder, or a team wiki you expect to revise for years.

Canvas is built the other way around: markdown round-trip is a launch requirement, not a nice-to-have. Paste a manuscript in; get typed blocks. Copy those blocks out; get markdown again — headings, lists, emphasis, and the shape of the argument still intact.

# On working with constraints

Constraints don't fight a draft — they
give it **edges**. The hardest part of
any long document is deciding what
it _isn't_.

> Working hypothesis: the more open
> the canvas, the longer the first
> paragraph takes.

- [ ] Outline section 2
- [ ] Confirm three quotes
- [x] Draft conclusion

Portability is how trust shows up in a file

Writers have been burned by formats that only open in one app. Fountain succeeded in screenwriting partly because a .fountain file is still readable in any text editor. Long-form tools rarely give you that courtesy. Binders win until the day you need a clean archive, a static site, or a collaborator who will not live in your editor.

Canvas stores documents as blocks, then serializes those blocks to markdown so structure survives the trip out and back in. You are not asking permission to leave. You are keeping a representation of your work that other tools — and future-you — can still read.

That is the product promise: the collaborative document can stay in Canvas; the export is a handoff you control, not a hostage release.

Diff-friendly history needs a text shape

Version control for writers only works if changes are readable. Opaque save formats turn every save point into archaeology — you can restore a blob, but you cannot see what moved. Markdown is boring on purpose: it diffs cleanly, archives cleanly, and plays with the tools researchers already use.

When a chapter grows by two sections and loses a callout, you want that story in the changes view, not a binary that only the original app understands. Round-trip markdown is how Canvas keeps prose in a shape that history tools respect. The same reason engineers prefer text configs is the reason a 90,000-word draft should not live only inside a proprietary graph.

Import is where history starts

The failure mode that matters is not “can I get a file out.” It is “did I lose the version where chapter three still worked.”

When you import a manuscript into Canvas, name a save point immediately — “Imported from Discord dump,” “Agent draft before cut,” whatever marks the moment. If the importer surfaces warnings, fix them on a draft line so the clean import stays untouched while you experiment. That habit — mark the moment, then edit — is the whole point of putting version control under a block editor instead of bolting autosave onto a page.

A good import is not a one-time conversion. It is the first save point in a document you intend to keep living for years.

The question to ask when you evaluate editors

If you are weighing Canvas against Google Docs, a wiki, or a notes app that “just works,” familiarity on day one is the wrong scoreboard. Ask whether your structure and history survive the moment you need to leave, publish, or collaborate with someone outside your editor.

• Can you take a clean archive without flattening every heading into a PDF?
• Can two versions of chapter five be compared without opening both in the original app?
• Can a future tool still parse what you wrote today?

Markdown round-trip is how we keep those answers yes. Ready for the product split? Who each Inkwell product is for walks Plot vs Canvas, and the Canvas learn pillar goes deeper on blocks, outlines, and export.