The UX of Local-First

Overview

Eileen Wagner — UX researcher at Superbloom, founder of the Decent Patterns library, and builder of local-first prototypes like Upwelling (collaborative editor at Ink & Switch) and Full Screen (collaborative whiteboard on Y.JS) — comes to Local-First Conf with a clear message: stop celebrating the promises and start fixing the UX gaps that will kill adoption before it starts.

Her framing is sharp. The tech community has spent years on the why of local-first. Now we need the how — specifically, four UX challenges that require collective solutions because inconsistent patterns across apps will confuse users out of the ecosystem entirely.

Key Arguments

1. You Can't Demo Offline in a Browser

Nothing screams "internet" louder than a browser window. Wagner's team learned this building Full Screen — wrapping the app in Tauri for native downloads was the transitional fix, but it introduced new problems (managed devices that block app installs). The ideal? A new container paradigm that signals "this is a safe offline space" without the baggage of browser chrome.

File extensions turned out to be surprisingly powerful. A custom .fullscreen extension created instant mental associations: this file belongs to this app, it lives on my machine. Downloads and toast notifications ("keep your work — save an editable copy") tested extremely well for communicating ownership.

2. Ownership Is a Responsibility, Not Just a Feature

The cloud-first ecosystem actively fights local-first UX. Default save directories point to cloud storage. Google Docs requires manual opt-in for offline. Podcast apps treat downloads as exceptions. Every platform convention tells users: local is the afterthought.

Wagner pushes harder than most on the uncomfortable side of ownership. In a user study on Shamir secret sharing for account recovery, participants happily designated peers to hold their shards — but refused when asked to hold someone else's. The actual quote: "Nah, I don't want that on my phone." More control means more responsibility for backups, security, and long-term preservation. The standard privacy pitch glosses over this tradeoff.

Her practical advice: local-first does not imply local-only. Cloud backups are smart. A big peer is smart. End-to-end encryption is table stakes.

3. Sync States Are a UX Nightmare

Dropbox has six sync icons and nobody knows what they mean. Local-first makes it worse by adding offline editing, version divergence, and the question of which version is the "agreed upon" one — not just the latest by timestamp.

Wagner identifies two strategies with no middle ground:

Explicit — push/pull indicators, freshness timestamps, conflict alerts interrupting your workflow to force review. Appropriate for financial reporting, legal documents, anything where wrong versions have consequences.

Non-intrusive — background sync, last-write-wins, automatic version history. Appropriate for collaborative whiteboards, casual tools, anything where approximate convergence is fine.

The trap: mixing strategies. Pick one. Context determines the right choice, but the icons themselves are already in conflict — a filled dot means "online peer" in Slack but "unsynced changes" in Figma. The community needs to agree on these conventions now, before they calcify.

4. Version Control Needs a Non-Git Vocabulary

CRDTs don't eliminate the need for manual intervention in serious collaboration. AI won't save you either — it just pushes review up one level (now you're reviewing the AI's merge decisions). Change management for local-first means answering: how do you track changes without visual clutter? How do you group them meaningfully? How do you propose, accept, reject, and amend?

Wagner's Upwelling prototype used a "stack and layers" metaphor — the document is a version stack, and you draft on layers that merge down. Patchwork (from Ink & Switch) uses visual branches instead. Both explore the same question: what does Git feel like when the user has never heard of Git?

This is where Wagner gets genuinely excited. It's rare for UX designers to get greenfield conceptual work — most of the time you're polishing existing patterns. Version control for non-developers is wide open.

mindmap
  root((UX of Local-First))
    Offline Capability
      Browsers scream internet
      Native apps as transitional fix
      Need new container paradigm
      File extensions signal locality
    Data Ownership
      Cloud-first is the default mental model
      Files communicate possession
      Download toasts build trust
      Ownership means responsibility
      Local-first ≠ local-only
    Sync States
      Explicit strategy
        Push/pull indicators
        Freshness timestamps
        Version conflict alerts
      Non-intrusive strategy
        Background sync
        Last-write-wins
        Auto version history
      Context determines approach
    Version Control for Users
      Change tracking without clutter
      Creative privacy spaces
      Layers and stacks metaphor
      Patchwork branches model
      Git concepts adapted for non-devs

Bonus: "Local-First" Is Developer Speak

Wagner calls it out directly — "local-first" is not a user-facing term. Neither is "offline-first" or "cloud-optional." Her suggestion: "offline available" for capability description, "user-owned" for value signaling. Stop naming what you're not building (cloud) and start naming what you are.

Notable Quotes

"There's nothing that screams more internet than a web browser." — Eileen Wagner

"With a lot more control, you have a lot more responsibility." — Eileen Wagner

"You want to do either one or the other and nothing in between." — Eileen Wagner, on explicit vs. non-intrusive sync UX strategies

Practical Takeaways

• Use file extensions and download mechanics to communicate local ownership — they tested better than any UI indicator
• Cloud backup isn't a compromise of local-first principles, it's basic user safety
• Choose explicit or non-intrusive sync UX, never hybrid — the context of your app determines which
• Agree on shared icon conventions (filled dot = synced? online? unread?) before patterns diverge across apps
• Frame user-facing messaging as "offline available" or "user-owned," not "local-first"

Connections

• local-first-software-pragmatism-vs-idealism — Same conference, same community tension. Wiggins frames the idealist/pragmatist divide; Wagner frames the developer/user divide. Both argue the movement needs to solve non-technical problems to succeed
• malleable-software — Wagner worked on Upwelling at Ink & Switch, the same lab behind Patchwork and the malleable software research agenda. Her version control UX explorations are direct implementations of that vision
• the-big-questions-of-local-first — The panel discussion grapples with the same sync strategy tensions Wagner identifies. Aaron's concentric circles model maps to her explicit vs. non-intrusive spectrum
• why-local-first-apps-havent-become-popular — Wagner's talk is essentially the UX answer to why adoption lags. Bambini focuses on the sync infrastructure challenge; Wagner shows the user-facing side of the same problem
• ux-and-dx-with-sync-engines — Assmann covers the developer experience of sync engines; Wagner covers what users actually see and struggle with on the other side of those abstractions