Ernest PKM Personal Knowledge System Workflow (2023.25)

Published:
  Lastmod: 2024-02-03
Ernest PKM
by Ernest Chiang

Post Title Image (Photo by Ash Edmonds on Unsplash)

✳️ Introduction

(Initial version on 2023-02-16)

Since childhood, probably elementary school, I’ve been fascinated by notebooks, enjoying recording numbers, temperatures, times, the number of pigeons on the neighboring building outside the window, and how long it takes for them to fly around and return to their pigeon house, etc.

As I grew older, I’ve always regarded notebooks as tools for recording, memorizing, and quick review. However, I felt that they didn’t match my workflow, whether at work or at home. In recent years, I’ve been better at integrating them into my workflow and documenting my current Personal Knowledge Management (PKM) process. This workflow is derived from the various shares of many people who are smarter and more focused on PKM, Smart Notes, and other areas than me.

This workflow may not be suitable for everyone, but I hope that by sharing these ideas, we can spark conversations and discussions that help us, and even our next generation, reduce the time spent exploring and improve learning or work efficiency, ultimately contributing to society or humanity.

AI, Machine Learning, ChatGPT, Google Bard, Amazon Titan, Auto-GPT, and other models and artificial intelligence tools can handle some tasks for us, but they also operate following a certain process. If we can master the idea of process design, I believe we can still make a difference in society and organizations. Let’s encourage each other.

(Updated to 2023.25 version on 2023-04-19)

Originally, I thought the Ernest PKM series of articles would be updated once a year, but in just a quarter of a year (0.25 year), various AI thoughts and products have descended like meteorites, prompting a forced industry upgrade.

Since the GPT version numbers have evolved from GPT-1, GPT-2, GPT-3, GPT-3.5, to GPT-4, Ernest PKM has also undergone a minor upgrade, integrating both human brain and AI (MLOps) workflows, resulting in the Ernest PKM 2023.25 version. I chose to publish it on 4/19 Generative AI Meetup to share and interact with old friends and new acquaintances whom I haven’t seen for a long time.

✳️ TL;DR

If you just want to see the tools, then this article is only three sentences:

  • PKM = a “personal” “knowledge” “system” for yourself, be comfortable and don’t feel pressured.
  • Choose a note-taking system that suits you. Choose other tools based on your information sources and access scenarios.
  • No matter your profession, industry, knowledge system, or what tools your upstream and downstream choose, AI reads much faster than us. Don’t waste time and bravely use Readwise.

If you want to coexist with AI and build your PKM to create something beautiful together, prepare a pot of tea or a cup of coffee, and read on:

  • Refer to others’ PKM building ideas (e.g., the “IIDEE” framework currently used by Ernest PKM) and construct your knowledge system = build your “workflow” and “framework.”
  • After constructing a version, try running it for a few months, accumulate experience, and continue to adjust.

✳️ Scope

  • We often spend time dealing with fragmented, single-point tasks.
  • We spend less time assembling processes, building structures, and creating systems.

“Assembly of processes, establishment of structures, and creation of systems” begins with “decomposition.”

So I break down the “Personal Knowledge System (PKM)” into three parts: “personal,” “knowledge,” and “system”:

  • Personal: It’s for yourself, be comfortable, and don’t feel pressured.
  • Knowledge: Knowledge 1 is the recognition and identification of a topic through cognition, leading to a certain understanding, and these understandings possess the potential ability to be used for specific purposes.
  • System: With processes, tools, and the possibility of automation. Related terms include “structure” and “framework”.

This article is about “personal use”. As for corporate and organizational knowledge systems and productivity processes, I will share separately. 👉 If you are interested, you can reserve a spot here, and I will update you with any new developments or related information.

“It’s for yourself, be comfortable, and don’t feel pressured.”

Forgetful people are still forgetful (oh, don’t get me wrong, it’s me), and we can integrate context and design various triggers for ourselves to start work tasks.

  • Combine with cyclical review to form inner cycle.
  • Combine with goal orientation to form external output.

✳️ Revision History

I plan (hope) to gradually update the content of this article, and as time goes by, some tools may be adjusted.

  • Ernest PKM 2023.25 version (2023-04-19)
    • Integrating both human brain and AI (MLOps) workflows.
  • Ernest PKM 2023 version (2023-02-16)
    • In the 2021 version, there were originally 7 process steps, which have now been integrated into 5 process steps in 2023.
    • Disperse “tools” into each process step.
    • Classify the “scoring mechanism” into the “information source” and “output” steps at the beginning and end.

✳️ IIDEE Workflow Chart

The current (2023) revised Ernest PKM workflow chart and decomposition structure are as follows, with five stages that can be abbreviated as “IIDEE”:

ernest-pkm-loop-2023
  • 1️⃣ Input sources
    • 1.1 Types
    • 1.2 Tools
    • 1.3 Scoring
  • 2️⃣ Inbox
    • 2.1 Tools
  • 3️⃣ Distilling
    • 3.1 Patterns
    • 3.2 Frameworks
  • 4️⃣ Executing
    • 4.1 Extend Knowledge
    • 4.2 Build Business –> Workflows
  • 5️⃣ Expressing
    • 5.1 Types
    • 5.2 Tools
    • 5.3 Scoring

I’ve broken it down into more detail, and you can use this skeleton as a reference, then adjust the process steps to suit your own needs. It should be handy, comfortable, and sustainable for you.

I wish everyone can create your own PKM workflow. You are welcome to share your PKM workflow with me in various ways, and we can exchange ideas.

✳️ IIDEE+AI Workflow Chart

Starting from Ernest PKM 2023.25 version, in addition to sorting human thoughts, it also integrates human and computer work and some historical context in the structural breakdown. This version was presented at the 4/19 Generative AI Meetup ft. HappyDesigner. Although there was no recording at the event, those who missed it can still simulate the experience of the last 10 minutes of the meetup by changing slides every 4 seconds. You can also request a recording?!

This version is divided into four sections:

  • 1 = Investigating Things (格物), Extending knowledge (致知)
    • Knowledge Node
    • 0️⃣ Knowledge Sorting Toolkit
  • 2 = Making the intent sincere (誠意), Rectifying the mind (正心)
    • 1️⃣2️⃣3️⃣4️⃣5️⃣ IIDEE Basic Framework
  • 3 = Cultivating Oneself (修身), Regulating the family (齊家) 👉 Diverse (多元), Establishing a family (成家)
    • It could be a person or AI, so it’s diverse and establishing a family
    • 3-ernest-pkm-s-v-o
  • 4 = Managing the state (治國), Bringing peace to the world (平天下) 👉 Managing the corporate (治企業), Bringing peace to the industry (平產業)
    • Looking forward to using such a human-computer integrated workflow in corporates and industries.
    • 4-ernest-pkm-with-ai

These breakdowns have no right or wrong, as long as it’s comfortable and suits your personal needs.

I wish everyone can integrate AI workflows into their PKM workflows, increasing productivity and creating value. You are welcome to share your PKM workflow with me in various ways, and we can exchange ideas.

0️⃣ Knowledge Sorting Toolkit

1-1-ernest-pkm-knowledge-node 1-2-ernest-pkm-tupao-toolkit

(TODO: Add more details later)

0.0 Knowledge Node

0.1 Relationship

0.2 Classification

0.3 Namespaces

0.4 Workflow

0.5 Iteration

1️⃣ Input sources

During the process of interdisciplinary self-learning, you’ll encounter a variety of information sources in different formats. Depending on each person’s situation, some formats may be easier to absorb than others, and some can be designed for automated workflows. Some sources may be more structured, while others may be more fragmented. It depends on our needs and adjustments can be made periodically or irregularly.

1.1 Types

Below are some common input source types that I frequently use. You can also observe various aspects of your life and identify your commonly used sources.

graph LR; B(1.1.1 Books) --> IS(1.1 Input Source Types); A(1.1.2 Articles) --> IS; V(1.1.3 Videos) --> IS; P(1.1.4 Podcasts) --> IS; C(1.1.5 Courses and Speeches) --> IS; CT(1.1.6 Conversations and thoughts) --> IS; class IS mermaidTier1; class B,A,V,P,C,CT mermaidTier2;

1.1.1 Books

Books are generally more structured information sources, with relatively solid, verified, and organized content.

Of course, some books are collections of multiple articles, which can be filtered through the “scoring mechanism” in later sections or selected for reading specific sections.

Under what circumstances would Ernest choose to read books:
  • Most books are relatively inexpensive.
    • Don’t get me wrong, the absolute price may be unaffordable compared to each person’s current situation, but compared to the effort, time, and resources invested by the author, obtaining relatively complete knowledge content at such a price is generally quite cheap.
    • If you find it not worth the price after reading, you can use the “scoring mechanism” to mentally remove it from your list.
  • In book selection, I mostly choose classic books that have been published for a while.
    • This way, more people have already helped me filter and even analyze the cases in the books.
    • In the historical evolution after the publication of the book, which ones withstand verification, which ones don’t, whether new parameters need to be added to my “scoring mechanism” for judgment, etc.
    • If necessary, you can also include the person recommending the book in the “scoring mechanism.”
  • Combining the above two points, if obtained in e-book format, the acquisition cost is often even cheaper, and there is a chance to connect digital workflows for note-taking and organization.
Ernest PKM Workflow:
  • Safari Books (O’Reilly Media)
    • Safari Books (O’Reilly Media) can be read on computer browsers, iPads, and iPhones.
    • I usually download a few books to my iPad for offline reading. Reading situations include long-distance transportation and places where telecommunication signals are poor but need to stay for a long time.
    • Images in books can be saved to Photos or taken as screenshots.
      • For personal use. If you plan to share publicly in later workflow steps, please consult relevant copyright laws and regulations or contact the author or publisher for confirmation.
    • Safari Books (O’Reilly Media) can be imported into Readwise for consolidation.
    • For more details, please refer to Section 1.2.1.
  • Amazon Kindle
    • Kindle can be imported into Readwise for consolidation.
    • For more details, please refer to Section 1.2.1.
  • Kobo
    • Readwise has supported Kobo note import since 2022 (I had been manually developing a small program for this but paused it).
    • Convert your Kobo books to read on your Kindle. (I purchased a 2021 Kindle Paperwhite (5th Generation), which is faster than my previous Kobo reader.)
    • For more details, please refer to Section 1.2.1.

1.1.2 Articles

Articles usually refer to content that can be read in a short amount of time, such as within 15 minutes. They provide more fragmented, piecemeal, and real-time information or knowledge, and usually require subsequent processing, organization, and linking to transform into easily usable and readily available knowledge.

I currently first channel all the articles I collect into Readwise Reader (before 2023, into Instapaper) for later reading.

Under what circumstances would Ernest choose to read the article:
  • Utilizes fragmented time (waiting, waiting for someone, waiting for transportation, waiting for dishes, etc.)
  • Low acquisition cost (but high mining cost, so it needs to be combined with a scoring mechanism to filter sources and improve conversion efficiency)
Ernest PKM workflow:
  • Collecting
    • Encounter new articles on the phone or computer
    • Mobile (including tablets)
      • Use the phone’s operating system sharing feature to import articles from browsers, communication software, and social software into the read-it-later app.
      • If some mobile apps cannot share articles into the read-it-later app, import them into the to-do app.
    • Computer
  • Read-it-later
    • I currently use Instapaper which allows me to easily collect article URLs for later reading on various platforms (computer browser, various mobile apps).
    • I further use Instapaper Premium, mainly for its “Unlimited notes” feature. 👉 Changed to Readwise Reader](https://readwise.io/i/ernest06) in 2023.
    • Import from Instapaper to Readwise for consolidation.
  • To-do
    • In some cases on mobile phones, the URL cannot be sent to Instapaper, so it will be temporarily stored in the ToRead Inbox of the to-do tool.
      • (Sometimes it’s just for peace of mind XDD)
      • (But after accumulating more and more, I’ll start to feel guilty and begin to face whether I really need such a blooming (buffering) area (blooming is correct!))
      • I currently use Todoist, which is cross-platform and has a web version.
        • Set up a Project named Stuff Inbox, place it at the top of all Projects (so you don’t have to scroll down when buffering, you can quickly buffer🌹)
        • Periodically clean up the Stuff Inbox. Complete reading, summarizing, and note-taking.
        • If the Stuff Inbox hasn’t been cleared for a while, such as a quarter or half a year, you can check:
          • Perhaps there is no incentive or not enough objective, and these situations belong to cases where there is no real need, but you still want to collect, so you can continue to stuff the Stuff Inbox. After a long time, see if you give up collecting.
          • It could also be that the process design didn’t allocate time for organizing, which belongs to cases with a real need (such as product design, thesis research, etc.), so you can go back and review the overall process and make slight adjustments. For example, schedule it in the morning routine of a certain day of the week.

1.1.3 Videos

Learning new things through videos seems easy and accessible, but from the perspective of time conversion efficiency, it’s not necessarily the best solution. (I know some friends who can only learn through watching videos and achieve fruitful results with some tips, so don’t worry about other people’s best solutions, just try to find your own optimal solution, which is more practical.)

Under what circumstances would Ernest choose to watch a video:
  • Situations that require image or video explanations, such as learning recipes (watching a dozen different chefs interpret a dish or ingredient using different techniques), new software interface operations (e.g., learning how to set up thinkorswim’s software interface), how to grow plants, etc.
  • Suitable for learners with visual memory, simultaneously deepening memory impressions through images, sounds, and even subtitle text
  • Moments when the brain needs relaxation
  • Adjusting the video playback speed according to personal status
Ernest PKM workflow:
  • Watch-it-later
    • Most video playback platforms allow you to bookmark planned or unfinished videos or add them to a playlist.
    • Readwise Reader also supports YouTube videos and generates subtitle text files for certain languages. This can save 1:1~1:2 of video watching time.
  • Note-taking
    • If watching tutorials or lectures, I use a dual-screen environment on the computer, and split-screen on the iPad. I watch the video on one side and take notes on the other.

1.1.4 Podcast

Podcasts are input into the brain through audio, and most of the time, they free up both hands to perform different tasks simultaneously, usually tasks that don’t require using audio to receive information. I spend a lower proportion of time listening to podcasts, mainly because the information density is lower (it’s a bit like… listening to a shorter self-introduction time in Clubhouse?!), but I still use the “scoring mechanism” to help filter information sources.

Under what circumstances, Ernest will choose to listen to the podcast:
  • When it’s possible to free up both hands for other tasks, but not while cooking, as I need to listen to sounds in the kitchen
  • During more relaxing moments
  • To collect stories
Ernest PKM workflow:
  • If I want to take notes while listening to a content-rich podcast, I currently use the Airr app on my phone to listen to podcast content.
    • When I hear a segment I want to take notes on, I can press the “previous” headphone function key, which defaults to capturing the previous 45 seconds of audio called “AirrQuotes”. After listening to the entire podcast, I go back and organize these segments and input them as text notes.
    • Then I can import them into Readwise for consolidation.

1.1.5 Courses and Lectures

I initially wanted to place the importance of courses after books, but decided to introduce articles, videos, and podcasts (audio) first before discussing courses.

Courses often present a process and environment to learners through the integration of multiple senses.

In high school, I was exposed to Project Adventure (PA) Experiential Education course design in an open school environment, which opened the door to course design and public speaking. Subsequent experiences in college photography club, elective courses, film and stage, and numerous opportunities to organize information and practice public speaking in AIESEC, left me with a deep affection for teaching and giving courses.

The origin of courses is similar to books, where instructors (authors) have accumulated extensive research on a topic or knowledge domain and present it to learners (readers) in a certain way. With this measure, we can classify which courses are worth studying, which are assembled, and which are pieced together. Other measures may be discussed in the “scoring mechanism”.

Under what circumstances would Ernest choose to participate in the course:
  • Like books, courses provide structured, systematic knowledge frameworks, which can greatly save time spent on self-exploration. (The cost, like books, is pre-selection.)
  • Courses offer more interaction with the instructor and interaction among peers than books. Short-term interaction is achieved through Q&A to address immediate issues or doubts, while long-term interaction may result in lifelong exchange of ideas and friendships.
Ernest PKM workflow:
  • The handling of course and book notes is similar, generally divided into physical and electronic forms.
  • I’ve observed that after leaving a course or lecture, other work tasks tend to pile up. So, I always try to summarize the content and extend my thoughts on-site or within 24 hours at the latest, and then organize them into my note-taking system.

1.1.6 Conversations and Thoughts

I categorize all short-term input sources into this category, where short-term refers to a timescale of seconds or minutes.

Under what circumstances, Ernest will choose to generate conversations and thoughts:
  • This category covers formal meetings, informal meetings, social media (Facebook posts/comments, Twitter tweets, Instagram posts, etc.), most of which are fragmented pieces of information. Some proportion of these require post-processing and reassembly, and this proportion usually varies depending on personal preference.
  • As conversations are mentioned here, it actually includes a portion of output, which is related to connections and creation. The output aspect will be discussed in later chapters.
Ernest PKM workflow:
  • Most short pieces (one sentence or shorter paragraphs) are first stored in my daily notes. (These are fleeting notes.)
  • If the insights from others have passed the “scoring mechanism,” they are temporarily stored in insights notes. (These are also fleeting notes.)
  • If I’m in a meeting and have the opportunity to use my computer, I record the key points of the meeting in the corresponding card on Heptabase.
    • Use the @today feature for quickly linking time continuity.
  • In some situations and environments where it’s not appropriate to use a computer, such as during dinner parties, sports events, or while on the move, I first remember keywords in my mind and then take a quick note on paper or Apple’s cross-device Notes.app when I have a break. Later in the day, I update the Heptabase note-taking system.
  • In some situations, using paper or a notebook may be more polite.

1.2 Tools

Tags: Fleeting Notes

When initially designing a PKM system, I was very concerned about which tools to use. I got a period of confusion, wondering why others have great productivity with these tools, but not myself.

Eventually, I began to pursue: dissecting the essence, integrating the workflow, and achieving the objective.

I adjusted my approach to be more “objective-driven,” working backward, re-examining the workflow, and choosing tools suitable for each specific objective.

In choosing tools, I prioritize the structure and workflow, combined with some experimental trials, iterating and piecing together what I want step by step (💰the money isn’t gone, it’s just…

I still use some manual tools or analog (non-digital) tools to achieve a deeper (or more relaxed) experience (for my brain), which helps in breaking down steps and attributes.

graph LR; R(1.2.1 Reading) --> T(1.2 Tools) ; RL(1.2.2 Read-it-later) --> T; A(1.2.3 Aggregation) --> T; class T mermaidTier1; class R,RL,A mermaidTier2;

1.2.1 Reading

There are various ways to acquire physical books, which will be skipped here.

My current e-book reading tools all have highlighting note-taking features. O’Reilly Safari Books (O’Reilly Media), Amazon Kindle, and Kobo can be imported into Readwise for aggregation, moving highlights downstream in my workflow, and automatically integrating them into my note-taking system.

O’Reilly Safari Books Online

Now called O'Reilly Learning, but I still call it Safari Books because otherwise, I wouldn’t remember it’s a place for reading books (early aging symptoms…).

  • I primarily use it for reading technology and technical books.
  • I started using Safari Books around 2012, initially sponsored by the Mozilla Foundation (thanks to the Reps Program and the MozTW community!). After several years of finding it helpful, subsequent companies I worked for continued to sponsor various reading sources, including Safari Books.
    • I started subscribing at around $40 per month, and the current monthly fee has increased to $49 (annual fee $499), with access to recorded sessions from O’Reilly conferences (saving a lot on ticket fees).
  • When exploring new technology domains, I engage in extensive reading. If I read more than two original books in a month, such as two O’Reilly books, subscribing to this platform is cost-effective.
    • Occasionally, friends ask me how I can find time to read all the sponsored reading sources.
    • My approach is the opposite; my goal is not to finish reading but to solve the problem at hand.
    • As mentioned in the first sentence of this paragraph, I use “exploring new knowledge domains” or “encountering problems” as triggers and then engage in extensive reading from available reading sources.

I’m very grateful for everyone’s sponsorship (and looking forward to more?!

(New books are available quickly on Safari Books, usually within a week or two after publication. There are also previews for new books.)

(Reading environment and highlights in Safari Books on a computer browser.)

(Readwise can handle importing highlights from Safari Books.)

Amazon Kindle

As an AWS Community Hero, it seems reasonable to own a Kindle, right? However, I started using the Kindle App to read AWS developer documentation, then purchased a physical Kindle Paperwhite before eventually becoming an AWS Community Hero :p

  • Chinese books vs. English books
    • Although Amazon Kindle has a Chinese bookstore, the number and variety of books are still accumulating.
    • Therefore, I let Amazon excel at its strength: original books and magazines are comprehensive, and I primarily use Kindle to read English books (and occasionally flip through beautiful recipe magazines for food photography inspiration).
  • Kindle plans
  • Kindle devices
    • In 2022Q1, I purchased the 2021 Kindle Paperwhite (5th Generation), which is fully-flat and easy to carry. I didn’t bother adding a protective case or cover. It’s perfect for carrying around, whether I’m taking a bus, flying, or need to stash it somewhere. It’s more convenient than an iPad, and the battery life is rarely a concern.
    • Kindle readers can take screenshots, but as of 2023, you need to connect to a computer to retrieve them.

kindle-paperwhite-2022 (Using the Amazon Kindle Paperwhite (11th) in various scenarios: cafes, airplanes, breakfast, and more. Lightweight and portable.)

(Although my Kindle Library contains some AWS documentation, to be honest, the Kindle version updates slower, so I later switched back to the web or PDF versions.)

Kobo
  • Based on friends’ recommendations, I currently purchase Chinese e-books on Kobo and then partially convert them to read on my Kindle device or use the Kobo app on my iPad.
  • My original 6-inch Kobo Clara HD e-reader is still placed on my desk for easy access. However, I hardly take it with me when going out.
Readdle PDF Expert

In some cases, the obtained document information is in PDF format. Among the numerous PDF reading tools, I currently choose PDF Expert by Readdle. It seamlessly continues reading and organizing highlights between iPhone, iPad, and macOS. There are other good choices on other platforms, with the selection criteria being:

  • Can handle highlights conveniently and pass them downstream in the workflow.
  • Can read file synchronization spaces (cloud spaces are more abstract; any space with synchronization and permission functions is acceptable), allowing seamless cross-device continuation of reading. If you are used to completing reading a file on the same device, you can omit this.
    • PDF Expert supports iOS Files, iCloud, Dropbox, Google Drive, Box, OneDrive, WebDAV, SMB, FTP/SFTP, etc., so in theory, it should work with Amazon S3 as well (are there any super AWS fans out there? XD). Just choose one that’s easy to maintain.

Please feel free to comment and recommend your favorite PDF reading tools.

Heptabase
  • New feature in 2023, you can directly read PDF files and highlight notes in the note-taking software Heptabase. Still adapting.

heptabase-card-pdf (Heptabase started supporting direct highlighting and commenting on PDFs in 2023-02, and you can further extend it into cards for integration.)

1.2.2 Read-it-later

Readwise Reader

Readwise Reader started small-scale closed testing in 2021 and opened for public beta testing in Q4 2022.

(Readwise Reader user interface. The left and right sides provide more guidance and features than Instapaper, but it’s relatively more distracting, so choose according to your preference.)

Instapaper

During commutes or short breaks, I often use Instapaper to read accumulated articles. When browsing social media or reading articles shared by friends, I may not have time to read them right away, so I have set up both iOS and Android platforms to share URLs to Instapaper, and then read them when I have time.

  • Highlights and annotations in Instapaper are considered notes and can be set up to import into Readwise.
  • Instapaper generates revenue through the Instapaper Premium plan, a single, straightforward plan.
    • $2.99 per month or $29.99 per year, if you can make use of unlimited notes, consider supporting Instapaper.
    • That said, after the introduction of Readwise Reader, I temporarily unsubscribed from Instapaper Premium in February 2023.
  • Pros:
    • Consistent reading interface for focused reading.
  • Cons:
    • Since Instapaper uses crawlers to fetch specified URL articles, it cannot access content from subscription-based websites.
    • For public content on non-subscription-based websites: * If the article does not follow HTML standards or lacks proper context, Instapaper crawlers cannot parse it. Many Chinese websites have beautiful layouts, but their source code lacks context. * If the webpage loads images upon scrolling, Instapaper crawlers may cause missing images. * However, the original URL is still preserved, and the worst-case scenario is going back to the original webpage for reading. Then use the next tool Hypothesis

instapaper-with-sushi (Reading articles with Instapaper while eating sushi. Wait, this is the wrong focus!)

instapaper-screenshot (Instapaper’s user interface is very clean, emphasizing focused reading of articles.)

Hypothesis

Hypothesis

Rarely used after 2022.

1.2.3 Aggregation

Readwise

Readwise can import highlights and annotations from various sources, process them centrally, or act as an aggregator to pass highlights and annotations downstream in the workflow. For example, many people export their data from Readwise to Evernote, Notion, or Roam. Most cloud-based services with API endpoints should be compatible. Don’t worry about local note-taking systems; Readwise provides an API for retrieving data regularly from your local setup.

  • Currently (2022-), my Readwise downstream is the Heptabase note-taking system on my local computer.

hepta-highlights-2023-02-14 (Heptabase added the Highlight app in 2023, making it easy to receive and organize highlights from Readwise. Recorded on 2023-02-14.)

  • Before (2021-2022), my Readwise downstream was Obsidian on my local computer, which didn’t have a cloud-based API endpoint to receive regular exports from Readwise. Fortunately, there are two or more Readwise plugins in Obsidian that can be set to periodically retrieve data from the Readwise API and import it into Obsidian as individual .md files.
  • Readwise full version costs $7.99 per month. Considering the time saved in aggregating notes from various sources, I decided to support Readwise. There’s a free 30-day trial when you start, which is well worth trying out.
    • Currently (2021-), Readwise is an essential part of my fleeting notes process.
graph LR; K(Kindle) --> RW; AB(Apple Books) --> RW; M(Medium) --> RW; RSS(RSS Feed) --> RWR; A(Articles) --> RWR; A --> IN; RWR(Readwise Reader) --> RW; IN(Instapaper) --> RW(Readwise); IN --> RWR; H(Hypothesis) --> RW; GR(Goodreads) --> RW; Airr(Airr) --> RW; RW --> O(Heptabase); class IN,K,AB,M,H,GR,Airr,RWR mermaidTier1; class RW mermaidTier2; class O mermaidTier3;

1.3 Scoring Mechanism

  • We face many information sources (from multiple different information providers, media, authors)
    • We have already addressed “information format” earlier; here, we focus on “information content” filtering.
  • Too many information sources can cause problems
    • Overwhelming amount of information
    • Information needs filtering and organizing before storage
    • Is this fact? Inference? Or false information?
    • Verify it yourself? Or have someone else help verify? What’s the accuracy of self-verification? The accuracy of others’ verification? Everyone has their own areas of expertise.
  • Assign scores to each information source
    • Prioritize high-scoring sources for use or processing
    • Handle low-scoring sources when you have time
    • Discard sources with zero scores, don’t waste your life
  • Example approach
    • Use a simple scale of 0-10 or 0-5 for scoring.
    • Use familiar tools such as Google Spreadsheet, Microsoft Excel, AirTable, or other spreadsheet tools, or simply record with note-taking tools.
    • Start with the information provider granularity, and if you have the capacity or need, process down to the author, editor, recommender granularity.
    • For example:
      • For article classification, start with the media name for the scoring mechanism, and if you have the capacity or need, score reporters and editors.
      • For book classification, use both publishers and authors for the scoring mechanism, and if you have the capacity or need, score recommenders and series.
      • The score of content sharing “How-I” with supporting information is higher than the score of “How-To”.
  • Alternatively, you can think of this “scoring mechanism” as parameters for various models such as AI/ML/ChatGPT.
    • Different parameter settings and choices may produce similar or different results.
    • Parameters may be adjusted based on different contexts or personas.
    • The working backwards method still applies here.

2️⃣ Inbox

Tag: knowing, Slip box, Literature Notes

  • Have an Inbox readily available for immediate note-taking, to be categorized later.
    • I want to emphasize “not categorizing” at the input and storage stage, before digesting. It’s recommended not to spend time categorizing at this stage.
    • If you still want to categorize at this stage,
      • Consider the goals you want to achieve during the application and output stages, and categorize with the end in mind.
      • Or identify “facts”, “inferences”, “emotions”, and “false information”. Be aware of dependencies; when using “inference A” for the “3️⃣ Distilling” step, you may get “inference B” based on “inference A”. It’s not prohibited, but remember to cancel “inference B” when “inference A” is no longer valid.
        • This judgment method can also be applied to add a “scoring mechanism” logic to the participants in a meeting.
  • Don’t worry too much about the format. (Check out Tiago Forte’s note-taking format)
  • Use tools that are popular among many users, as there’s a higher chance of utilizing ready-made, well-integrated automation tools for upstream and downstream processes (time-saving).
  • Before starting any action, establish a goal and write it down (place it somewhere where you can check it off when completed).
  • Don’t trust your brain and don’t use it as a temporary storage area for your Inbox. (I assume a human is reading this article by default? AI may choose to skip this?! Huh :p)

2.1 Tools

  • Various bookmarks, temporary storage, and playlist features provided by the SaaS, products, and services you use can serve as temporary storage Inboxes.
  • Pair with a periodic cleaning habit, such as a morning routine or weekly review, to regularly clear content scattered across multiple Inboxes, and send them to the appropriate tools in 2.2 based on the categories in 2.1.
  • Always be prepared with several tools for immediate note-taking and temporary storage.

Paper Notes

FIELD NOTES (A portable paper notebook with a magnetic notebook cover that can hold a pen on the right side. Recorded on 2023-02-27 in the Taipei office.)

  • I usually carry several pens and at least one FIELD NOTES with me.
  • Paper notes have an additional benefit: they can be transcribed.
    • Transcribing to another piece of paper or notebook allows the brain to be stimulated again, deepening impressions or making edits while transcribing.

Note-taking Software

  • I currently (2022-) use the Heptabase Journal app for conveniently recording daily notes. It can also serve as the foundation for the subsequent Distilling process, using the same operation methods and data formats to reduce data conversion (ETL) processing time.
  • Similar features are also available in other note-taking tools.
  • Alternatively, using the built-in note-taking app on computers, smartphones, or tablets with a simple date as the note title can be an easy-to-use tool.
    • For example, Apple devices come with the Notes app, which allows for cross-device synchronization, increasing time flexibility.
  • (2023-03-01) Heptabase Quick Capture mobile app was just released for testing.
    • You can quickly capture your thoughts on your phone, save them in the Heptabase Quick Capture mobile app, and then send them to the Heptabase Card Library.
Todoist
  • Refer to the to-do (Todo) workflow in section 1.1.2 of the article.

Digital Handwriting Notes

Notability (Handwritten notes produced with Notability are popular among friends and clients. You can also read or listen to lectures while writing or drawing notes.)

  • I (2021-) started using Notability, which allows handwriting on an iPad, then taking screenshots on a computer to integrate into note-taking systems, presentation slides, etc.
    • Regardless of which software you choose, consider spending some time trying out all the strokes, brushes, thicknesses, and colors to familiarize yourself with the available features.
    • Then, try organizing some note-taking plans to write or draw your thoughts, practicing these features to become proficient. This way, when it’s time to use them in real situations, you can quickly switch to the desired features or strokes. Otherwise, it’s easy to forget what you want to write while thinking about which feature to switch to or how to switch.
    • The downside is that if you need to carry various devices at the same time, the weight can become quite heavy. (Time for some weight training…)

Camera Reproduction

  • Any type of camera, such as smartphone cameras, can be used.
  • If you find yourself temporarily away from your preferred notebook, you can record your sudden ideas on any paper, whiteboard, or other tools. Then, take a photo with the camera and send it towards your note-taking system.

Whiteboard (Capture a whiteboard with a camera and develop a knowledge map to create consensus among the team during a collaborative exploration process, ensuring common and consistent definitions in subsequent work, and eventually becoming public information, saving time for those who join later in the process. Adapted from The Measurement Units of Dots, Lines, and Surfaces that Qin Shi Huang Did Not Have Time to Unify - From Letterpress Printing, Desktop Publishing to Mobile Devices and Display Systems.)

3️⃣ Distilling

Tags: Permanent Notes

(Originally “Organizing”, later changed to “Distilling”)

  • Review and internalize the ideas in your own words.
    • The above “1️⃣ Input Sources” and “2️⃣ Storage Inbox” sections may be accelerated or replaced by various tools or AI in the future.
    • However, if we want to process the information within our own minds, I tentatively call this step “3️⃣ Distilling”.
    • In this step, I aim for repeatability and consistency.
  • Output: Pattern = Trigger + Action + Parameters
  • Output: Organized, digested knowledge map, also known as a framework.
  • Process
    • During regular reviews (e.g. morning routine), review daily notes or globally organize Fleeting Notes on specific topics.
    • Usually, the content produced for the first time is not the final content to be used. Assume that multiple iterations, implementations, and adjustments are needed, and there is always room for drafting another version for next time.

3.1 Patterns

  • Considering repeatability and consistency, I tend to make some attempts in the problem-solving process and identify patterns or different parameters within the patterns.
    • This helps break down complex problems into smaller, more manageable actions that are relatively easier to advance and achieve results.
  • If the term “pattern” sounds unfamiliar, I’ll give some examples for better imagination.
    • First, try to imagine a “pattern” as a combination of: if and simple English sentence (S+V+O; Subject+Verb+Object).
    • In some common situations, you can omit “if”, but it’s better not to.
    • In some common situations, you can omit the “subject” (S), but it’s better not to.
    • Repeatedly apply and test each identified pattern, make actual tests, and adjust parameters.
      • If the pattern suits you, consider automating and scaling it 👉 integrating it into one of your frameworks.
    • Different domains may have different patterns.
      • Find patterns that suit you and use them repeatedly.
      • When switching or expanding domains, try to identify new patterns and appropriately eliminate existing ones.
  • 🌰 Example: If you encounter “My ____ is not necessarily equal to your ____”, you need to verify.
  • 🌰 Example: If you encounter an “adjective”, ask for the definition.
  • 🌰 Example: “He didn’t come.”
    • Cause: Don’t use pronouns, use complete sentences to describe to reduce misunderstandings.
    • If there is a pronoun in the sentence, replace it with the appropriate noun or title.
    • Common pronouns are “you, me, him” and “you all, we, they.”
    • In our team, for easy remembering, as soon as someone mentions a pronoun in a meeting and the pronoun can be replaced by two or more nouns according to the context, we remind each other “he didn’t come” and “who is he?”
  • ℹ️ Other common alternative keywords for “patterns”:

“One static and one dynamic, they can iterate with each other.”

3.2 Frameworks

Frameworks can be divided into “Concept Frameworks” and “Process Frameworks.” One static and one dynamic, they can iterate with each other.

3.2.1 Concept Framework

  • “Concept Framework” is used to describe the relationships between nodes (nodes).
    • Nodes can be anything, any concept. If a node is abstract, multiple concrete nodes can be used as supplementary descriptions to make it concrete while maintaining its abstract concept. This allows for practicing “abstraction.”
    • In the process, synonyms and antonyms of nodes will gradually be discovered. 👉 If you encounter “My ____ is not necessarily equal to your ____”, you need to verify.
  • When expanded and visualized, a “Concept Framework” looks a bit like an organizational chart or a biological taxonomy diagram.
  • “Concept Framework” corresponds to the next section “4.1 Extend Knowledge.”

heptabase-whiteboard-aws (Caption: Ernest, as an AWS Community Hero, it’s reasonable for him to organize the AWS product architecture classification tree by hand :p. Using Heptabase for organizing. 👉 Extended Reading: AWS Cloud Product List)

heptabase-whiteboard-architecture (Caption: Concept framework for Architecture. Organized using Heptabase.)

3.2.2 Workflow Framework

  • Multiple patterns can be connected to form a “Workflow Framework,” which describes the flow direction of information or certain nodes.
  • “Workflow Framework” corresponds to the next section “4.2 Build Business.”

heptabase-whiteboard-payment-ecosystem (Caption: Workflow framework for the Payment Ecosystem. Organized using Heptabase.)

4️⃣ Executing

Tags: Knowing, Doing

  • Execute = Trigger + Do

4.1 Extend Knowledge

  • The “3.2.1 Concept Framework” section is based on the “1️⃣ Input sources” and “2️⃣ Inbox” sections, and the result of digestion and organization.
  • Next, in “4.1 Applying Knowledge,” a periodic review process should be designed to allow you to access this knowledge at any time.
    • Personally, the challenge lies in “accessing it at any time.”
    • “At any time” and “access” can be considered separately.
      • How “at any time” does “at any time” need to be? 👉 e.g.
        • “At any time” like “responding within seconds during a customer’s call”?
        • Or “at any time” like “responding within minutes during a meeting discussion”?
        • Or “at any time” like “responding to a customer’s inquiry email within a few hours”?
      • How should “access” be “accessed”? 👉 e.g.
        • “Accessing only memory in the brain and only verbalizing,”
        • “Accessing only memory in the brain, in addition to verbalizing, using a whiteboard or pen and paper to enhance description,”
        • “Accessing not only memory in the brain but also a note-taking system on a computer, and only verbalizing,”
        • “Accessing not only memory in the brain but also a note-taking system on a computer, in addition to verbalizing, using a whiteboard or pen and paper to enhance description,”
        • “Having no memory in the brain but having a record in a note-taking system, trying to access the note-taking system from a device, …”,
        • And so on and so forth…
      • Design corresponding processes and tools for the “access at any time” scenarios you will encounter.
  • 🌰 Example: Develop a habit of organizing weekly or monthly reports, even if customers, bosses, or colleagues don’t ask for them. Accumulate reports from each period in one place for easy access, reference, and reminders.
  • 🌰 Example: When reviewing on your own, practice talking while drawing. Carry a notebook or A4 paper with you. When it’s time to present and explain, if there’s no whiteboard, take out the prepared paper and pen and start explaining, talking while drawing.
  • 🌰 Example: From time to time, friends ask me how to get started with AWS. I eventually organized it into an article AWS Learning Path and Strategy for easy access by myself and friends.
  • Whether through self-learning or traditional education systems and methods, the action of “knowing” to expand knowledge can be achieved.
    • However, traditional Asian methods rarely discuss objective, so we often hear statements like “knowing is not the same as doing,” “the difference between theory and practice,” and so on.
    • If you don’t quite understand “objective,” you can use another way to break it down: “time allocation ratio.”
      • This can be achieved by recording time and using simple math to get the ratio relationship.
      • Measure: the ratio of time spent on “knowing” and “doing” for yourself.
  • Don’t forget, the process that suits me may not suit you.
    • To find the process that suits you, you need to try and get hands-on experience.
  • Q: What’s the difference between “4.1 Apply Knowledge” and “5. Teaching (Teach Others to Do)”?
    • A: The difference lies in whether the “object” is “yourself” or “others.”
    • After understanding the differences, find corresponding methods and tools.
    • For “yourself,” find “application” methods and tools.
    • For “others,” find “teaching” methods and tools.
      • “Others” can be humans, animals, plants, AI/ML, etc., nodes other than yourself.
      • “Teaching” is a form of “transfer.” In terms of information transfer between two software, you may face ETL challenges.

4.2 Build Business

  • After “knowing,” it’s time to tackle “doing.”
    • Transform “knowledge” into “action.” Apply it to work and life, and achieve tangible results and outcomes.
    • After “doing” it yourself, you can update your resume and your “About Me.”
    • If it’s a company or organization, you can update the “About Us” section.
    • Don’t just ask “Who am I,” also ask “What have I done.”
    • Don’t just ask “Who is he,” also ask “What has he done.”
  • The term “Business” here is used conceptually, and any activity can be considered a business.
  • Clarify the objective, set goals (S.M.A.R.T.), and define the scope.
  • Continuing from the “3.2.2 Workflow Framework” section, but actually invest resources (e.g., resources in PMP terms = people, machines, materials) to execute the process, obtain output, adjust parameters, and decide whether to continue investing resources.
    • Accumulate experience through practice, review, and iterative improvement.
    • “Experience” can be “process” (different approaches) or “parameters” (different values).
    • In terms of PDCA:
      • Plan = Generate “3.2.2 Workflow Framework.”
      • Do = Try implementing the plan, and in most cases, some problems will arise, whether big or small.
      • Check = Regardless of whether problems are encountered, collect, measure, evaluate, and inspect various outputs based on the objective as much as possible.
      • Act = Analyze and improve. Move towards the goal or adjust the goal.
  • 👉 Further reading: Tupao-DIY-style Product Development Framework

4.3 Workflows

  • Start with “manual” then move to “automatic.”
    • In the beginning, it will be “manual operations.” After confirming feasibility, alignment with objective, accumulating experience, and stabilizing work processes, attempt to “automate” and transform into “automatic operations.”
    • Once the process is validated, you can update “3.2 Frameworks.”
  • Workflow steps
    • Use the same description method as “3.1 Patterns”: combinations of simple English sentences (S+V+O; Subject+Verb+Object).
  • Dependencies within the workflow:
    • Certain workflow steps may require roles, permissions, or authorization.
    • Some workflow steps may require raw materials or machinery.
    • The connection between processes, accompanied by “time,” generates “inventory.”
    • After each process, there might be a change in “status” and the creation of artifacts.
    • The output of a previous process might serve as raw material for the next process step.
    • Normal processing vs. exception handling.
  • Since “workflows” often require collaboration with others, we’ll briefly introduce it here in the personal version of Ernest PKM and expand on it later in the enterprise version of Ernest EKM (if possible).

5️⃣ Expressing

Tags: collaboration, teaching, teaching others to do

Our output could potentially become someone else’s input.

Therefore, “Types” are common between these two “1️⃣ Input sources” and “5️⃣ Expressing” are mutually applicable. However, the tools and scoring mechanisms are different. Here, “5.2 Tools” refers to tools for expressing your own knowledge and processes, while “5.3 Scoring” refers to how your expression of knowledge and processes corresponds to other people's scoring mechanisms.

5.1 Types

As mentioned earlier, please refer to the “1.1 Types” section.

5.2 Tools

I’m very keen on making the whole IIDEE cycle more and more efficient. So, from using pure Markdown format notes (plain text files, using git on GitHub) in the early days, to using software like Joplin and Obsidian, and now mainly using Heptabase, I’ve been focused on efficiency. I want to be efficient. “Efficiency” means increasing output within a unit of time. One approach is to reduce the transition time between steps, or find ways to eliminate the need for transitions, allowing for direct use when pasting.

Here, I share some tools I am currently using and some that I have experimented with in the past.

Heptabase

Currently (2022-), the tool I use more often is the Heptabase Whiteboard feature.

  • During discussions with clients or when giving presentations, I try to turn on the camera and screen sharing, so both parties can see which node we are focusing on. When listening to someone sharing, others can observe the relationship between this node and other nodes, or their relative positions.
  • After the discussion or presentation, if circumstances permit, I will share the Heptabase Whiteboard I used during the meeting or presentation with the attendees, allowing them to use the same tool to disseminate information (e.g., presenting the same Heptabase Whiteboard to their colleagues or bosses).
  • 🌰 Example: In the “Postscript” of this article AWS Learning Path and Strategy, there is a knowledge map portal, which is an example of a Heptabase Whiteboard. Everyone can click in to play around and freely zoom the map.

Hugo

Hugo is a static website generator developed using the Go programming language.

Marp

Marp is a tool for converting Markdown format to slide presentations.

  • 🌰 For example: In 2020, during the pandemic, it was a great time for geeks like me to engage in various online learning and exchanges. One of them included participating in a cross-disciplinary marketing reading group. The book chosen for the reading was “Influence: The Psychology of Persuasion” by Dr. Robert B. Cialdini, published in 1984 by Arizona State University. This book has had a significant impact on the marketing field for more than three decades. I was assigned chapters 5 and 6. At that time, I was playing with Marp, so the notes: Influence: Chapter 5-6 you see in this post, whether it’s the HTML version or the downloadable PDF version, were all created by pasting my reading notes (the Markdown files at that time) into a Marp format file in VSCode and then using the Marp extension to output HTML or PDF files.
  • Unfortunately, due to the need for format conversion and the added dependency of the Marp extension, I haven’t used it for a while. However, the entire process is based on open formats and open source, which is still quite special, so I’ve recorded it here.

Whiteboard

If we can meet in person, I really like venues with whiteboards, the more the better, like our Taipei office.

  • Discussing and unfolding various nodes on the whiteboard simultaneously makes it easy to focus and confirm that “the terms you use are the same as the ones I use, and the same as the ones he uses.” After the discussion, taking a photo with a smartphone and sending it to the participants is almost equivalent to having a meeting record.

5.3 Scoring Mechanism

  • The origin of the scoring mechanism is actually to establish trust between people, giving higher trust scores to information sources you trust and checking them regularly.
  • If we are discussing the scoring mechanism between humans and machines, to some extent, it may be mechanisms like SEO.
  • Whether it’s human-to-human, human-to-machine, or machine-to-machine, I think originality is a good starting point to some extent. Let’s all give it a try.
  • Further reading:

✳️ Bottom line

Thanks to the growing popularity of various large language models (LLM) (e.g., ChatGPT), I thought maybe I wouldn’t need to write this conclusion paragraph specifically (getting hit XDD).

What I mean is, the original intention of sharing this long article is to write for friends who are willing to try implementing it themselves. I hope that if you have the opportunity, you can ask ChatGPT to summarize and then try to build your own workflow, such as trying to incorporate ChatGPT into a specific step in the process, which may achieve twice the results with half the effort.

If you have any thoughts, feedback, questions during the building process, or are interested in the corporate version of Ernest PKM, you can leave a message or ask questions for Ernest here, and I will collect and organize them from time to time, updating the FAQ section below or replying to you via email.

If you’ve referred to the Ernest PKM Workflow or IIDEE framework and started building your own knowledge system and shared it on your website, feel free to leave a message for Ernest to exchange links with each other. I will link everyone’s shares in the “References” section at the end of the article, looking forward to various positive shares and modifications.

Buy Me a Coffee If this sharing has been helpful to you, Feel free to buy Ernest a cup of coffee ☕ 😍

ℹ️ Other Common Alternative Keywords

  • Personal knowledge management
  • Knowledge management workflow
  • Personal self-learning database

✳️ Frequently Asked Questions (FAQ)

Q: What’s the difference between highlights and annotations?

Ernest:

When in doubt, ________…

Oh, I mean… When facing a “comparison” situation, look up the definitions of the terms:

Q: How did you start designing your PKM workflow?

Ernest:

I actually started with various handwritten notes, documents, and meeting records. As time went on and the scale of the organization grew, I had to handle more and more projects simultaneously. As a technical manager, I also had to continually learn new knowledge, which led to a new level of chaos.

So, I began to observe how others dealt with large amounts of data across contexts and domains. I realized that I needed to handle not only “data” but also “emotions”, “timing”, “process”, “structure”, and so on. Therefore, I tried to find some spare time to sort out my workflow and define it. At first, it was quite rudimentary, but then I discovered the keyword “PKM” and searched for it extensively. By breaking down various fragments and steps, my own framework gradually emerged.

My approach is more makeshift (tupao). When I feel overwhelmed, I find a piece of paper, a whiteboard, or use Notability on an iPad to write (or draw) things down (the “vomiting method”). I try to find the relationships, assumptions, definitions, parameters, etc., between the things I’ve written down, and then attempt to infer a version. I then look for various scenarios or extreme examples to overthrow my inferences, iterate to the next version, and continue this process.

The simplified version looks like this:

graph LR; IN(Input) --> P(Process); P --> OUT(Output); class IN,OUT mermaidTier1;

After adding Context and Inbox, it looks like this:

graph LR; C1(Context) --> T1(Input Trigger); T1 --> PP1(Pre Process); PP1 --> IN(Inbox); PP1 --> T1; IN --> PostP1(Post Process); PostP1 --> Out1(Output); PostP1 --> IN; class T1,Out1 mermaidTier1; class IN mermaidTier2;

When you start to combine multiple different contexts, information explosion may occur, making you feel out of control. However, continuing to try and refine your own PKM workflow can help keep the complexity of the situation under control and maintain its functionality.

graph LR; C1(Context #1) --> T1(Input Trigger #1); C2(Context #2) --> T2(Input Trigger #2); C3(Context #3) --> T3(Input Trigger #3); T1 --> PP1(Pre Process #1); T2 --> PP2(Pre Process #2); T3 --> PP3(Pre Process #3); PP1 --> IN(Inbox); PP1 --> T1; PP2 --> IN(Inbox); PP2 --> T2; PP3 --> IN(Inbox); PP3 --> T3; IN --> PostP1(Post Process #1); IN --> PostP2(Post Process #2); IN --> PostP3(Post Process #3); PostP1 --> Out1(Output #1); PostP1 --> IN; PostP2 --> Out2(Output #2); PostP2 --> IN; PostP3 --> Out3(Output #3); PostP3 --> IN; class T1,T2,T3,Out1,Out2,Out3 mermaidTier1; class IN mermaidTier2;

✳️ Reference Materials

Tools mentioned in this article

Ernest PKM development history

Mix of Smart Notes, Bullet Journal, Productivity

Others sharing their PKM personal knowledge management systems

  1. Knowledge - Wikipedia, the free encyclopedia ↩︎

Loading comments…