I’m perpetually adding to it and looking up specific things I need, but I much less often have time to read through much of it, let alone organize. A day ago I spent an hour reading and organizing my (org-roam) knowledge graph. It felt both wonderfully productive and yet at the same time such a small drop in the bucket. I wish I could do that every day.
It makes me wonder what others’ habits are. How much time do you put into it? Is it random or methodical?
Just got notified about this, my own thread, in the weekly summary. I wonder whether that will bring in any more eyeballs …
no? time to switch to scrach and M-x emacs-lisp-mode. Actualy,
chatgpt…
[Try to visit this node for next 7 days to see how it works. ]
For the last week, it’s quite a failure in terms of building this habit,
so I can vaguely remember that I just opened this note to check
things, let alone to reconnecting anything. The main issue or the main
cause I feel is the stress from the outside world that kind of forcing
me to respond in a way that I forgot this habit. Put it simply, this
habit isn’t ready into my system.
But I do want say one positive thing that I still come here to post
some observations for myself. And I also change the default reaction
when I prepare my next week’s template. So I consider it as a win for
this point, for myself.
I find that whenever I’m creating a new node, I only need to answer a single question: “In what context do I want to re-discover this stuff?” I connect it to whatever I think it relates to. I don’t concern myself with the total number of nodes in my system, or with the idea that at some point I will be overwhelmed by the sheer volume of information I’ve accumulated. I’m not worried about that at all; my database is already home to almost 2,000 records. Context can be anything: your colleague’s name, a place, or a project. I also liberally make use of aliases to prevent accidentally re-creating the same topic under a different name. With proper context association duplicate nodes are not a big problem. I had to merge conflicting nodes a few times, it takes just a few seconds to wgrep and replace the IDs.
I think you are on a good point. The essential part of this system is to answer the question, in what context do I want to rediscover this note again? If I reflect myself, it’s probably only a very small percentage of the time I tend to think that way. What I do find something that is useful to make something non-trivial is to create those sort of index files.
I have to confess myself that a lot of time when I think of a topic, when I think about what’s the context I’m going to think about, I always think, okay, this is probably going to be a small structure of a bigger system, and that’s it. So I tend to create lots of MOC, but really that’s not the case here.
Also, most of the notes I’ve written are in response to a document that I’m supposed to write. I was just checking something interesting I found online, and I think about that in the future I might use it. Then I create that note. But that’s not how it’s supposed to be. Well, I do write manuscripts, but I would say it’s more confined to a narrow topic. So basically, the content of that manuscript is well-defined, so it’s rare for me to go to a far distance to write something that is not strongly connected to the current work I’m doing.
And I often ask myself, okay, what can I get out of those notes? Maybe I can create something real, physical real, I mean. Or I can write a piece of code that is workable, or even better, I can share with other people that might find it works. Other than that, the most thing I can think about right now is kind of management of the concept that I encounter on a day-to-day basis. Another thing is definitely about log, so if I’m working on a non-trivial ellipse function, it’s really helpful to keep a log file to go through some paper solutions to a certain question or even just to interact with LLM, it’s really helpful in that way.
I do want to improve this habit, maybe it’s about building that habit of reviewing the past notes or past things that is already in the system but is not fully developed. It’s not an easy thing, so every time I revisit old notes, there’s not too much of an attraction there for me to expand on ideas. I’m not sure if it’s human nature or it’s just me being too greedy of writing things down. The thing I can think about to attack is to build that consistency to act in a certain way. I think it’s more manageable than to just have all those type of things that I can choose to look back on those notes.
when I think about what’s the context
I don’t dwell on it too much, possibly because my notes are typically quite straightforward. I’ve developed a method to easily connect things. For instance, I’ve bound the “[[” sequence to org-roam-node-insert and if I need to type a left square bracket, I’ve assigned it to “[ SPC”. Let’s say, if I’ve just had a discussion on a subject with John Doe, a colleague or friend, I would create a new node linked to “People” or “Colleagues”. I have capture templates to quickly add people’s names. I also sometimes add dates using org-timestamp commands, but not for the purpose of searching. My brain simply doesn’t pattern match on calendar events. I typically don’t sit around reminiscing: “I remember we talked about this thing last year in October…”
create those sort of index files
That turned out to be not so useful for me. Over time, I realized that categorizing things into broad classification topics is not conducive to locating items. Nowadays, I deliberately break larger items into smaller, more granular levels. However, I must take care not to get too specific in order to avoid being stuck in a state of “analysis paralysis”, where I cannot determine where an item fits. Today, I maintain only a handful of substantial category groups. Additionally, it is essential that I periodically check for orphaned nodes that have no linkage. Both of these tasks are easily manageable with the help of roam-ui. I also think org-ui may be beneficial in generating new ideas from existing nodes, but since I do not write papers, books, or essays, I cannot say for certain.
One thing currently missing in my workflow is a way to find related notes that sit at the intersection of different nodes. For example, I have notes on Dr. Julia Shaw’s books, but to find them, I have to navigate through ‘Neurobiology’, ‘Books’, or ‘Julia Shaw’ nodes. Each of these topics has a considerable amount of back-linked and forward-linked sub-nodes. It would be nice if I could somehow search for ‘Neurobiology books’ or ‘Books by Dr. Julia Shaw’. I’ve been hoping for someone to propose a practical method for such a search. There’s project Khoj - “A Natural Language Search Engine for Your Org-Mode Notes”, but I have yet to try it, and I’m not entirely sure it would work for this specific use case.
Oh wow. I just installed Khoj and tried it. It is really great. And very simple to set it up.
For me, all those details are primarily serve as a hint for better
searching because That’s a natural way for me to pull a certain
information out of my memory.
I remember we talked about this thing last year in October
The time stamp is useful. I used it to find files that I have more
impression on when I created it. So, I also intend to leave more
information around that file/folder. I also use people to search,
mainly in the email.
Now I’m still trying to searching for a potential structure of a
random stuff and I can’t feel I can and in this process I feel
something special.
For example, Abstraction with Figures.
I try to kind of synthesize a template for a certain topic because
once I’m familiar with a template, then I know what’s gonna be next
things and it will help me to make momentum, build habits. Structure
notes is what I are looking for. But during the process, It is usually
not that obvious. It’s like searching happeness. If someone sakes for
happyness, he will not find it. But instead of focusing on small
things, focus on the moment, that’s where the happiness is going to
come.
Here’s some high-level structure. I tend to make a template on top of
those categories.
When I have no idea where I put a piece of information. Usually
searching the entire database, or add more advance techniques:
Khoj
Org-similarity
ada-embed
It seems to me that the first two things are based on those two
algorithms to find similarity between an input call rate to a given
database: TF-IDF and BM25.
I’m in the process to use OpenAI, their API
(`text-embedding-ada-002’), to achieve a similar task. this time I
didn’t try aux similarity or Khoj. I’m also interested to see
differences, pros and cons, performance, cost…
I recently saw a discussion about how people, how to organize their
nodes, or the idea that is similar in what we discussed here, as how
to build a better hack to maintain all of those information, or
something that we want to do out of our notes. I still like the way to
view these things from a point of view of yin and yang.
Yin: Separation Based, Structured, Research and Professionalism.
Yang: Integration Based, Freeform, Personal Growth and Insight
Again, from this brain map, clearly, I see a structure but at the same
time, I also see those random things embedded into system, mainly the
journaling part or log part, or diary part.
I can have an idea about revisiting those things in my database, but I
don’t have a clear mind. For example, I have a vague idea that I have
to make things clear. If I see tens of headline just within the first
subheading, I feel stressed and I want to try to minimize their
numbers. Also, I want to extend the idea into a deeper level. I also
try to find when I revisit those things. Basically, it’s when I feel
relaxed, when the environmental trigger is there. Based on those
ideas, I try to re-enter that kind of situation. Another thing is that
I want to reach out to the outside world to discuss the idea. That’s
where I find ideas, or that’s when I find most of the time I revisit a
certain topic. On a personal level, I tend to reread books, but I
don’t have a fixed schedule. Mostly, it’s when things pops up, I have
to do that and then I revisit the book I read. Or sometimes I just
randomly visit it and know that it’s just a note for a book I read. At
that time, I’m probably going to revisit it. Also, when I try to think
about it in my habit, I’ll probably revisit tons of notes I’ve just
written or discussed with other people. And of course, when I need to
write a program, I have to revisit a lot of things. The log file, the
idea that I want to achieve.
I like to think about “idea, memory, moment, emotion, skill” as a
target for old things that I want to visit. For example, when I read a
book, when I saw a book I want to revisit, I probably think about
three ideas that I’ve learned from this book. Maybe I have some memory
about a particular idea that I actually used in my life, and that’s
going to bring some memory, moments, back to the time I revisited
those notes. And of course, that will bring a certain emotion. Also I
definitely think about if there are any skills that I, it really
depends on this idea or this set of ideas.
I also like the idea that when you revisit old notes, you potentially
want to add some stuff, some action to that notes. For exmaple, adding
a tag, then probably adding a link, and probably adding new questions
I’m not clear about, original ideas. And maybe during the process, I
have some new ideas I can capture and then process later.