Information forest · by Tyler Angert
<- Home

What should the browser of the 2020s be?

Tue Oct 20 2020 • 5 min

What will a browser built for research, analysis, rabbit-hole exploration, messy thinking, and collaboration look like? These features are listed in the order I thought of them in, not necessarily by importance.

  1. Graph visualization and mind mapping. You should be able to easily see connections between all of the web pages you have open at once, any subset of them, and explore them / navigate through them visually. I should see connections between pages based on similar content (shared words, shared metadata), time accessed, and other people who have accessed it. Any piece of content, text, link, etc is a node that can be linked manually or automatically to anything else. Everything is indexable.

  2. Interactive history and version control. I should be able to scroll through the history of my browsing as an enormous branching graph that I can annotate, travel around, share with others, and version. I should be able to duplicate branches of searches, identify important pages or findings in a path, and combine / merge paths with each other.

  3. Predictive search paths. If I look something up on Google, I often have no idea where to go next. This browser should present entire paths of potential searches to explore. This is essentially the same thing as the interactive history point, just in the opposite time direction

  4. Super Command-F (Superf). You should be able to search across multiple pages at once, for multiple terms (this AND that), for NLP entities (people, places, things, dates), and use a human-readable regex interface (inspired by something like VerbalExpressions).

    • Example queries you could run:

      1. "Find all wikipedia links."
      2. "Find all people with last names that start with A."
      3. "Find every quote that contain a link"
      4. "Find every link in a list"
      5. "Find every sentence that contains numbers."
      6. "Find everything related to ‘mathematics’" (using something like Word2Vec).
      7. "Open all links my research partner already saw."
    • Imagine what these queries could do when applied to dozens or hundreds of pages in parallel! The fact that we are stuck to basic fuzzy search makes Chrome feel like it’s stuck in the 1700s.

  5. Collaboration. You should be able to instantly share a browsing session with somebody and invite them into the rabbit hole. How many times have you opened up a google doc to collect links that you and someone else find? Curating links shouldn’t require an external service like Pocket,, or Pinboard. It should be baked into browsing. Imagine going on a walk with a friend through the woods— you find leaves, old cans, weird bugs, and most importantly, have conversations that you wouldn’t have had alone. The same kind of shared experience should be possible during browsing.

  6. Automatic scraping and clustering. This thing should help me browse. Related to Superf, the browser should automatically scrape images, links, and relevant information (summaries, references to things I’ve already found, things it thinks I want to read) from any page I find. It should be able to visualize clusters of pages and content based on all available metadata from every piece of media: images (color, date taken, objects in the images, size, etc), text (shared vocabulary, NLP entities as said before), code, audio, video.

  7. Built in word processing. One of the most common things we do with web pages is write about them and write with content from them. Annotation is simply not enough. I should be able to write and use portions of web pages as quotes directly into my writing, be able to reference any page at my finger tips, and write in this combined 3D browsing / writing space. It’s like a collage of page bits and my own writing. How do you write something "in between" two web pages? How do you annotate multiple pages at once? I very rarely write or brainstorm in isolation and almost always have several dozen tabs open at once to reference. There should be no boundary between synthesizing thoughts and consuming information.

  8. Backlinks. Anytime you link to something, both ends of the link should have references to it. This is based on Ted Nelson’s early vision of Hypertext, is done very well in Roam Research, and is even mentioned in Tim Berners-Lee’s open design questions for the web. This means that organization of pages and ideas is bottom up / implicit. You don’t have to create folders or categories ahead of time; the relationships between things define their organization. The idea of things being in one place becomes antiquated (hierarchitis). You shouldn’t have to spend extra time deciding what folder something should live in when it’s related 10 other things.

  9. An infinitely zoomable interface (ZUI). You should be able to zoom in and out of information and take a step back when you want to see the big picture. A macroscope for the web. A hot air balloon! Infinite canvas drawing apps do this well. Another great recent example is Muse. Interactions should feel completely fluid, seamless, and not take you abruptly out of any particular context.


If there could be one product that would satisfy all of these requirements, I’m not totally confident about what it would look like. Would it have to be 3D? Does this beg for VR? Should it be super minimal and focused on search as the main interaction method? My intuition tells me that this should be 3D. My own mental model of browsing looks like some sea/forest of information that I can rapidly jump around. There are literally too many dimensions / axes of information for this to be effective in a 2D visualization. I want this thing to feel like a place. If every website is a building in a city, what do the streets look like? The sidewalks? The parks? A lot of these ideas are still useful without browsing, and more of a personal memex / intranet. Everything that doesn’t directly involve web pages (infinite zooming, auto-backlinks, 3D maps, Super Command-F, etc) all would be directly beneficial towards word processing, but I think their real power is when word processing + browsing are combined into one seamless workflow, much how content organization and writing are combined into just writing in Roam. From the browsing side, there are obvious performance bottlenecks here and it’s really not practical if it’s slower than chrome. More on this later.


  • find link to paper about visualizing browsing history · by Tyler Angert