Glass Half Empty

A blog about design, culture, and weird thoughts.

Semilattice: Problem Complex

Published in
Senior Project


There is an exponential increase of knowledge in our world. The existing personal knowledge management systems are insufficient to assist people in processing information.

Problems with existing tools for personal knowledge management

Hierarchy > Association

Most tools store information in unique paths in a tree structure, just as how we used to store information on paper and put them in folders, cabinets, and boxes, using the unique path as its classification.

The tree structure organization used by most tools encourages collection, not connection, of ideas.

  • Human brain organizes ideas in associations. One item is always related to multiple “topics,” and categorizing it in one place makes it impossible to be used in other places. It is difficult to reuse and cross-reference ideas, and it encourages hoarding information. And hoarding is not thinking.
  • Designers from Roam Research, a project that holds very similar values, have articulated the issue in their white paper:
Many files are divorced from context; cast into a drawer, rather than methodically fitted into a broader framework of knowledge. Knowledge trees can create pseudo-relationships between files nested within a given hierarchy, but these are not explicit, and can only describe a vertical ‘parent and child’ taxonomy. Some tools, such as web pages and wikis, also allow for orthogonal linking between related files, but this takes place in an ad hoc fashion, and again, there is no ability to explicitly define relationships.

Documentation > Experimentation

In most tools, thoughts are externalized through text and images organized in linear reading order is most suitable for documentation of knowledge.

However, the production of knowledge is an evolving process of discovery, experimentation, and iteration, in which multiple mentalities of thinking are used.

  • A relevant framework from American psychologist Jerome Bruner introduced three mentalities of thinking: visual (seeing, representations, shapes, analogy), interactive (doing, performing, physically active), symbolic (logic, reasoning, language, notation).
  • This explained why digital designers like using post-it notes to organize brainstorming materials so much, because the physicality of it allows us to declutter thoughts effectively.
  • Another example is how writing software doesn’t really support people to structure their writings better, especially if in large volume. → Related Twitter conversation

We constantly associate new information encountered with existing knowledge in our mind.

An organic PKM system shouldn't encourage people to simply document well-thought materials. A step beyond documentation, and even association between existing knowledge, is to support the externalization and experimentation of information.

  • One particular perspective: most tools separate themselves from casual exploratory activities and information sources. The tools are connected through basic copy & paste functions and clipper extensions.
  • The disconnection makes association of new information with existing knowledge difficult, discouraging evaluation and iteration.

Design Goals and Pivotal Inspirations

The main vision is to design a personal workspace for learning, organizing, and iterating knowledge.

1. Building associative connections at ease between items and new information to be discovered

2. Building a medium with intuitive inputs at various levels of abstraction of thought (text + sketching)


I've been thinking about the best way to conduct concept testing with potential audience, since it feels difficult to gain insight without making the tool/medium first. Making the prototype as a "highlight" of what occupies my head is the only thing I could do. → Thread of three short prototype videos on Twitter

Moving Forward


  • Open multiple windows with Split View
  • Access bookmarks list
  • Search across the database
  • Use lasso tool to move materials between cards

Possible use cases

I assume the tool will be most helpful at online research across different information sources.

  • Brainstorming design ideas
  • Writing outline for a paper

Semilattice: Documenting the Progress

Problem, Intent, and Concepts
Published in
Senior Project


Christopher Alexander, Architectural Intelligence

1. We’re living in a world with increasingly complex problems sitting within a growing ecosystem of of social, cultural, or informational pressures.

Some examples: economic equality -> AI-powered crime prediction solutions -> Palantir/; climate crisis -> traffic congestion -> ride-sharing services; political extremism -> deterioration of public social space -> Facebook

2. Many of these “wicked problems” are entangled with digital technology, and it made me reflect on its desired role in our world.

Vannevar Bush, As We May Think; Douglas Engelbart, Improve our Ability to Improve

3. I was fascinated with the promise of computing to extend humans’ ability to process and use knowledge.

Alan Kay, User Interface: A Personal View; model is inspired by Hugh Dubberly’s SECI model and took analogy of researchers’ synthesis process

4. I want to examine how computers today help us “collect information, create knowledge, manipulate and share it,” because tools we create shape ourselves. I made a WIP model to represent the process of knowledge creation.

5. I found web browsers one of the tools in today’s computers that are most insufficient at “processing and using knowledge.” The main critique is: the archaic IA metaphors (pages, tabs, bookmarks, history (of links)) prevent people from comprehending information when the web has expanded itself massively.

6. The relationship between “comprehension,” “collection,” and “interpretation” still needs to be untangled here. To correlate with the model, people use web browsers to collect information. One could argue that the silo-ed consumption-ready experience in web browsers is sufficient for “previewing” information before they’re collected elsewhere. But my assumption is that providing affordances to let people interpret web-based content could help us comprehend better.

7. From here, I aimed to find ways to help people comprehend digital information better in web browsers.


  • Project Xanadu: the parallel comparison of documents shows the power of visual thinking. (I learned about Jerome Bruner’s mentalities of thinking from Bret Victor’s work, including……)
  • DynamicLand: if information is tangible and ready for manipulation and experimentation to its fullest.
  • LiquidText and MarginNote: excellent examples of active reading. When I look at them, I can’t help but think if the similar affordances could be provided to web browsing.
  • WebMemex: a project with a very similar intent, seems discontinued unfortunately. It mentioned fighting back the ephemerality of digital information by making personal archiving more accessible.
  • Muse: an excellent example of creating affordances for people to develop ideas in a digital scrapbook-like environment.


Design new metaphors for web browsers that encourage thinking and creating new ideas.


Right now I’m designing the fundamental elements in the paradigm. I made a very scrappy proof of concept to demonstrate the following ideas:

  • Creating excerpts and multi-media cards to support active reading
  • Visuospatial connection between pages and cards
  • Automatic grouping and manual clustering pages and cards to visualize relationships and support the thinking process

Here is a use case from my own experience as a student: researching the future of transportation and looking for design ideas.

To start, I might want to search for a relevant keyword. It is the beginning of my exploration.

Clicking on a link opens a new page connected visually to show the trace of discovery. I might stumble upon something not so interesting and go back to the search results.

Clicking on another link took me to another article, which turned out to be really insightful. A paragraph is very well-written and reminds me of a new research direction. I can take it out as an excerpt and write a note related to the paragraph.

I want to learn more about the context of this page: who wrote this? Following the hyperlink led me to the main page of Greenfield Labs, where I found another quote inspiring for a different design direction. As I expand my area of exploration, my previous traces are automatically collapsed, showing my thoughts and highlights: what I think is more important than what I see.

I started a couple of new searches related to that thought by creating new nodes from the note card I created earlier. The first two aren’t really interesting, so I put them away. Clicking on the “history” button would reveal my traces. The last search led me to an interesting article by the Washington Post. The car design of this image is interesting, and it said it’s by IDEO. There’s no hyperlink here, how can I find more information about this project?

I highlighted this image and wrote down my thoughts about space usage of vehicles. The connected search led me to IDEO’s original project. As I look back at where I am compared to where I started, my trace of thought is represented as the visual traces of exploration in the space. What’s unimportant is left out, what’s important is visible in a glance.

As I continue my research, I can organize pages and cards into subgroups to reflect hierarchy of topics. I can manually connect between pages, cards, and groups to reflect my logical reasoning. I can break the original hierarchy of pages and cards and reorganize them to reflect my "understanding."

See also

On Thinking, as a Way to Build the Future

Semilattice, pt. 1

Semilattice, pt. 2

Feedback Highlights

  • The design direction as a knowledge-creation-system assumes people are inherently incentivized to seek truth, instead of falling into certain beliefs. A more promising direction might be supporting groups who are directly incentivized to make better decisions. 

Next steps

  • Reframing problem moments: supporting objective-based exploratory research in web browsing
  • Build a workshop-like low-fi prototype to test the core concept