tl;dr: "๐๐ฃ๐ฉ๐ง๐๐๐๐" combining tabs, bookmarks, and notes into a local-first, end-to-end encrypted, non-hierarchical knowledge base that can be viewed as a web of linked nodes! From the world-wide-web to your own personal web (or collaborative organizational web).
Sidenote: Mozilla's track record sponsoring next-generation technologies is awesome, like the Servo browser engine or the perceptually-driven, patent-proof, Daala video codec based on lapped transforms (not used in AV1 or AV2, sadly). Could new labs/products incorporate censorship-resistant future-internet protocol Named Data Networking (intro video), or peer-to-peer (P2P) version-controlled file synchronization based on git-annex/DataLadโฆor both?Introduction: Motivation & Vision
Accessing the world's knowledge is only the first step; tracking and organizing it is even harder. IMHO the bleeding edge Firefox once hadโintroducing the innovations other web browsers have all adoptedโcould be regained by iterating, albeit in a radical way, on the current concepts of tabs, bookmarks (and associated tags), and notes.
Competition i.e. Field-Proven & Market-Validated
Knowledge base applications like Appflowy, Lark, Confluence, and Notion have taken over, diverting attention & love away from traditional collaborative apps like 1st generation wiki software or Google Docs. There is even a decentralized one built on content-centric networking, Ownly. Plus, over the past few years, more and more of these platforms have launched with or implemented a semantic network or graph-view of notes/pages (SiYuanโonly one with end-to-end encryption for sync, Cotoami, Obsidian, Logseq, Anytype, Supernotes, and manyยน others) as originally pioneered by Roam Research (whitepaper).
Users and organizations have outgrown cloud-based filesystem imitations with rigid hierarchical definitions of types of "files" and their structure/organization. People expect fluid and semantic data with intuitive options to edit, store, access, and view information.
Differentiation: Unique Value Proposition (UVP)
Clearly, Firefox wouldn't have a first-mover's advantage, but all existing solutions are built on top of the web, which separates them from the browsing experience and users have to explore in their browser of choice to research and find information before returning to whatever app to reference and annotate what they have found.
Mozilla has the special position of providing the portal to information itself: the web browser, complete with concepts that already map to categorizing & linking data: tab groups, bookmarks, calendar/events, file sharing/publishing, notes (discontinued), real-time communication, even social bookmarking ร la Pocket, andโif you use Sidebery or Tree-Style Tabโnested tab relations (which tab was this one opened from?).
Plus, there would be niche-dominance for not just doing one or two of the best-practices spearheaded by existing apps, but combining the best of all worlds. In particular, a local-first (whitepaper) approach, so far exclusively implemented by Typecell, making features like edit/revision history, instantaneous responsiveness via lazy-sync and end-to-end encryption (e2ee) possible using a federated graph database synchronization protocol to host these collaborative documents (Matrix CRDT repo).
MOZSSโEnterprise: Monetization
The general concept is already commercially validated and nearly industry-standard. Along with charging for other products as an enterprise services bundle to provide additional functionality not really useful to individual users (collaborative translations, shared password/credential management, etc.), additional features built on this could be paywalled such as:
- Security features: Extra access controls, 2FA/MFA enforcement, auditing abilities
- User management: SSO Authentication, automatically sync intraweb spaces & permissions based on LDAP, AD, and SCIM policy definitions
- Raise limit on file storage, number of users collaborating on one of these infraweb spaces, etc.
Implementation
Visually, each "node" can show a preview of the content not unlike typical "tiles" or "cards" used across the web, for example on Firefox's default start page showing Pocket articles (the thumbnail would be a screenshot of the tab, and the text would be user-generated), but instead of a grid they would be laid out in a zoomable 2D or 3D graph or "web" of pages.
References to other notes/tabs are represented by lines linking to other nodes on the map or graph. A sidebar can serve as a quick-access panel with a hierarchical list of important starting points and filterable by "tab groups" or arbitrary labels. Each node represents either:
- A "tab" (a URL which has been opened by the user) which can show each device it is open on, and whether for that device it is "active" or "inactive" ร la Firefox Mobile.
- A standalone note: Rich document with arbitrary data. Tabs are each attached to a note so that users can annotate tabs with their thoughts and references to other things.
- An external reference (a URI that either has only been linked to from user-generated annotations in their notes/tabs).
Power users (many in the free/libre/open source sofware community) will expect to be able to edit with shortcuts and a standard markup language without clicking around a WYSIWYM/WYSIWYG rich text editorโwhich should still be defaultโand LaTeX (ร la FidusWriter) is well-equipped, but ideally would use something simpler and de facto standard-compatible like MyST (Markdown extended to have feature parity with AsciiDoc/reStructuredText for technical, scientific communication and publication).
Phases
- Backbone: With the address bar in Firefox auto-completing to suggest tabs open on any device alongside bookmarks, these can be refactored to be unified under the hood, separated only by some metadata
- Synchronization/CRDT: Possibly rewrite syncserver to preserve modification history on top of the Matrix protocol using peritext for rich formatting and possibly Recfiles as the underlying database to store arbitrary data including aforementioned richly formatted hypertext.
- Web view: Toggle to a graph-based view of all tabs and notes allowing users to annotate them with their own thoughts & ideas. Each "node" would show a URL-thumbnail (if representing a tab) and (either way) a preview of the user-generated text.
- Arbitrary data: Support various blocks of data (cards, tasks, tables, calendars, forms, spreadsheets, canvas/boards, flowcharts, databases etc.)
- Collaboration & publishing: This is needed for enterprise, of course, but it would be nice to offer to families for free.
Challenges
- File storage: Users may expect to upload larger files, which puts added strain on sync server. Maybe this can be avoided with a p2p sync tool like listed at the top of this post
- Initial effort: There's a massive backend-lift needed to "do it right" from the beginning, otherwise the later possibilities are severely limited if not impossible. Perhaps this could be a collaboration with the Matrix Foundation, or others to share the cost.
Conclusion
There is a regular progression of proprietary โfree/libre/open source development whereby a new concept is implemented repeatedly in usually-but-not-always proprietary services and condenses into an open standard/protocol. If you look at various past examples:
- SMTP from the walled gardens of AOL, CompuServe, GEnie, MCI Mail, IBM PROFS, etc.
- Matrix from the silos of Slack, Zulip, Discord, Twist, Teams, Mattermost, etc.
- Fediverse experiments like Nostr (ร la Ditto), ActivityPub (ร la Mastodon), or AT Protocol (ร la Bluesky) from the many many [micro]blogging and social media hubs Xanga, LiveJournal, MySpace, Twitter/X, Tumblr, Instagram, Reddit, Facebook, the list goes on.
When the expected features, functionality, and best practices become hegemonic, an open standard emerges to formalize and democratize the adoption of said technology. It seems like we're approaching that point here, but who leads and wins the de facto way forward is yet to be seen. I hope it is Mozilla, but I'll be happy when anyone checks all the boxes laid out above: collaborative, local-first, end-to-end encrypted (backed by a decentralized graph database synchronization protocol and requisite CRDT), nonhierarchical, knowledge base that undercuts the expensive centralized hosting of all the current providers!
P.S. I have plenty of fire names for this other than the generic "IntraWeb"โplease steal this idea and let me help name it!
Footnotes
- The many other platforms supporting semantic network or knowledge-graphโin addition to those already mentioned aboveโinclude, but are not limited to, Capacities, Zenkit's Hypernotes, Tana, Foambubble, Constella, Amplenote, Relanote, Nuclino, MindMeister, Remnote, and probably others I've missed
9 Comments
