> what solves the problem for us is not a knowledge base, but search

There are problems with search: you have to know to search for something, and you have to know how to search for it. In some cases this is an issue, in some it isn't. I have had many times over the years where I reviewed my notes and reminded myself of things I had completely forgotten. Search is useless in that case.

But in any event, there's no conflict between a knowledge base and search. They are different things and you can search a knowledge base.

Search can not be avoided. The alternative to search is to go through everything that has been read which is not feasible. Tagging is the only solution. The user has to remember the tags which is referred to as context in zettlekeaten method. I could not see any workaround to this problem

I ran a private system for a while that proxied all my browser traffic and indexed it into Lucene/Solr. This was cute at the time but with the rise of javascript apps it became increasingly useless. There wasn't anything worth indexing in the responses. These days something like Zotero Connector is probably what you want.

That’s a really cool approach though! Maybe an updated version would be to have a headless browser shadowing your browsing and indexing the rendered HTML - there would still be some gaps though.

Maybe a browser plugin?

> 1. If we have time to enter something into a knowledge base of any kind - then we have time to just jot it on a piece of paper.

There is a kind of generation gap here I think.

I have 3 devices at arms reach while writing this, but it would take me 5 min to find a working pen and a blank piece of paper.

If I used it pen and paper more it would be closer to me now, but it would be the same wherever I go. Then I’d need to capture that paper or bring it with me until it has no value anymore, which is a PITA.

Our current society is tending to a paperless state, writing stuff on paper is just not fast, natural or reliable act anymore.

Eh, not so much generation as different way people think. I'm not old, have multiple laptops, desktops and smart phones. More often than not, when I need to seriously think about a paper I print it out and annotate the margins, underline, highlight, etc. I have a notebook (and fountain pens!) that I use to write out plans or projects or sketch designs. I have pads of flip paper for short-term (weekly) TODOs, and bullet-journal longer term life planning.

It's just a different way of organizing one's life. I spend enough time in front of the computer, sitting at the kitchen table to jot down notes or sketch out a project idea works better for me than trying to try the latest XYZ digital notebook app.

Good points, especially point 1. About the second part here - I remember seeing a (rather creative) guy post a personal project based on that, on here, a few days ago. He wrote a comprehensive post about his addon which aims to remembers exactly in which context has a user encountered said article/author/keyword in the past. I'll see if I can link it.

---

quick update, found it: https://news.ycombinator.com/item?id=23668507

> 1. If we have time to enter something into a knowledge base of any kind - then we have time to just jot it on a piece of paper.

Great point. I often see people sharing strategies for organizing notes but I rarely see people discussing how they get stuff down in the first place. And there's a lot of improvement we can make here.

I wanted to add more modularity to my jot-downs so that it would be easier to organize them later, one small step at a time. I found that I can get this modularity by writing my notes as if I'm texting myself.

(Working on a tool [1] that has this sms-style interface with keyboard shortcuts for purging my notes and organizing. would love your feedback).

[1] https://bytebase.io

My system lately has been to write every note as a raw text file that I save to a git repository on the fly. Later I process these notes and give them Zettlekasten style IDs. This step forces me to organize the knowledge: either everything gets a fresh ID (e.g. "25"), or it gets a derived ID that relates it to whatever subject or task it's related to (e.g. "44a3b9"). But it also means I can stream thoughts into a lightweight inbox without having to do the organizing upfront.

I then have derived categories from the main stream of thoughts. If the thought is a task, I further process these into categories. I have recurring tasks and events that I handle on a "Time" page that is effectively a calendar, and I have one-off tasks that go into a Kanban style backlog.

What I really got from Zettlekasten is that trying to establish your system upfront is a mistake. Things inevitably leak through your categories and then you lose faith in your system. By just having a running stream of thoughts and then relating them after the fact and deriving categories afterward, you get the benefits of organization without a lot of its failure modes.

Awesome. Glad to see that separating the "getting stuff down" from organizing has worked for you too.

Do you create a new raw text file for each note? How long does it typically take you to do the organizing process?

> trying to establish your system upfront is a mistake. Things inevitably leak through your categories and then you lose faith in your system.

I don't know if it's a mistake. I think you need to embrace the fact that your current system won't be perfect, and you'll need to adjust as you go. It doesn't mean you shouldn't start with something.

I do create a new text file each time, although it's sort of arbitrary what I group. Sometimes I will put my entire day's notes for a given book in one file. Other times I'll have atomic notes for each thought.

The organization process is wholly dependent on how much I come up with in a day/week. I mostly read for fun now, so I don't have to rigorously organize lecture notes or anything like that for later use. On the other hand I'm pretty rigorous with how I process tasks. As a ballpark figure I'd say 30-60 minutes per week, or 5-10 minutes per night.

To your latter point, I think that's fair. The main idea is to be flexible and recognize when you need new categories or whatever (and when you can dump unhelpful ones).

Thanks for explaining! It sounds like you have a strong process going.

> I do create a new text file each time, although it's sort of arbitrary what I group. It's great that you do what works for you without obsessing. Too often people chase after a rigid/"perfect" system and that's just not realistic.

In my experience, the more modular you can go (without it stopping your flow), the better. Because it'll be easier in that organize step.

I've found that by thinking about my raw notes as a chat with myself, I'm able to increase the modularity without hurting my flow.

> I want a web browser that remembers every single page I have visited (#) and then lets me search them.

Did you check out WorldBrain's Memex <http://worldbrain.io/> browser extension yet?

omg, somebody built it! Will immediately try that.

I find it hard to get people to write down anything. And if they do, they often times don't put enough effort into it, like writing down some bullet points without context or screenshots without adding text. This is a personal and cultural problem at companies and in teams. A lot of people don't know how to generate value from taking notes/documentation and aren't good writers in general. I'm still thinking about how you can teach people what's important when it comes to writing and when to take notes.

I couldn't agree more with your second point. Search is probably even more important than taking notes in the first place. I find myself searching through my notes multiple times a day, even if it's something you can find at Google at the same speed. But my notes are mine, meaning I know what I can't remember or in which form I need some piece of information. Sometimes, I write a long article about something and other times some bullet points are good enough. When I share my notes on a team/organization I try to make sure the context is clear. It's a bit like code: it's read by an order of magnitude more often than written/changed.

We built Emvi [1] to make note taking more collaborative and with focus on searchability and ease of use. I hope I can write more about this topic on our blog. In case you have ideas on how to solve these cultural problems, please let me know.

[1] https://emvi.com/

I agree that search is a key direction for growth in this tooling ecosystem. I don't think a tool should index every visited webpage though. I think knowledge base construction should be intentional and methodology, as the process provides its own value.

I’ve begun building a knowledge base that is built around a Google-search interface but with closed-world assumptions, such that it can only answer questions and provide results that I should know myself.

In this sense it’s another take on “second brain” tools, but to me it’s potentially a novel angle, using search with closed-world restrictions (ie. much fewer documents, personally stored and indexed) and a question-answering focused interface.

Most other tools focus on the authoring side of things, but I want something that will superpower my memory.

It’s all closed source until I get it going well, but I’ve set up http://rememex.org.

I agree, the most useful thing these tools can provide is fast data entry and robust search for most users. That’s the focus of the tool I’m building, https://NoteBrook.com . Reach out if you want to be in on the beta, it has been delayed a bit as I’ve been refining the search and editor experience of the alpha version and building super fast apps on every platform, but it should be launching in the next week or two!

I have a suspicion that fast data entry and automated search might be a red herring for this sort of tool. I believe that the real utility of my notes comes from actively engaging with them, and automation runs the risk of damaging that.

My own system (which is an entirely physical one) is designed around this principle; each step is designed to make me engage with my own recorded thoughts in a different way:

- Every note gets written on an A5 sheet of cardstock, with a unique slug written in large print at the top of the card. The bottom ~(1/3) of the card is reserved for cross-references.

- After writing a note, I look through the file box of existing notes for 2-3 related ones; these become the initial cross-references and are notated on both cards.

- The new card gets photocopied, and the original placed in the file box, ordered alphabetically by the slug/title.

- The copy gets turned into a flashcard by blacking out the title and put in a Leitner box (a physical spaced-repetition system)

Thanks for your insight. I agree for a knowledge base, where you don’t want to persist all information, and precision is important. Most of my note generation tends toward writing down historical updates (project a is at stage b on date c) or recording minutes of a conversation (x person said y was ok to implement). Or perhaps I have this task to do and once it’s done I only need to know when it started and completed. There is no inherent ordering or importance to this note data until I need to recall it. That is the type of note I typically take, and many others too. It’s a different use case.

If we write it down on a piece of paper and there is a fire or natural disaster it easily lost or destroyed. If you factor in writing it down on a piece of paper, scanning and making it searchable then it is easier just to input it on an electronic device.

Indirectly related - you can try https://web.hypothes.is. You can annotate web pages (public, private, group) and have page notes.