"""
Github, for instance, is built on a tree-like structure. When Github arrived to the version-control arena, it quickly blew away the other systems (which were rife with horrid version conflicts and syncing issues; think technologies like SVN).[7]
"""
Ummm what? SVN is also built on a tree like structure. SVN is terrible but not because (git) is based on a tree and SVN is not. I honestly don't get his point. Is this just word salad?
The author seems very unimpressed with note taking tools like Evernote and Obsidian, then goes on to describe something similar - I may have missed some subtlety of the article in thinking that.
I'm using Obsidian and find it makes the Zettelkasten-style experience even easier, because you don't need any arbitrary ID on each note, just a name, and you can easily create new links to notes which don't yet exist. You can definitely use it to create an 'antinet'.
As a software developer it’s very confusing to read an article where a tree metaphor has been chosen because in the author’s view trees aren’t hierarchical! It’s a good exercise in cognitive dissonance.
The author needs to make up his mind: in one place he says that there is no "parent leaf, or child leaf", and then two sentences later says "The ID on each leaf refers to it position on a branch of a tree." He's just playing word games. Either you're proposing a hierarchical system, or you're not.
What he seems to be advocating is a taxonomy where each card can link to another card by ID, whilst taking a long time saying it.
I tried out zkn a few weeks ago for my programming notes. It was kinda OK, but it didn't preserve line breaks when I pasted in code, not even between the zettels.
So I decided to see if I could just use plain text files. The files are numbered sequentially. The first file is 01-index.md. The second one is 02-spi-irq.md. I can see straight away that the file beginning 02 tells me how to do SPI transfers using interrupts.
The index file is for, well, indexing. It only has 1 entry at the moment:
duty cycle, 4
So I know that stuff relating to duty cycles in in card 4.
It seems a reasonable system.
I also keep a written notebook. I'll keep entries in them, which are typically 1 page long, but could be longer. Each entry has a title, which I list at the back of the book as a table of contents. The big disadvantage of a TOC is that the contents are in no particular order, and scanning through them is a pain. Inevitably one's eyes glaze over and the TOC becomes a mist.
The solution? Indexing! The index is now my first port of call. I can cross-reference however I please. A tip I would add is to not just write page numbers, but add in additional descriptions. Here is my section on Pinouts:
So I can find pinout diagrams for my microcontrollers easily. Otherwise they're just mystery numbers, and not much help.
The disadvantage with this system is that one must guess how much space a letter of the alphabet will take. Things won't be in strict alphabetical order, either, as you don't know in advance what the entries would be.
I guess one could always keep a computer file for an index. Or perhaps forego an index and keep a TOC which one can grep.
The article is all over the place, but I think I might be able to summarize:
- ZK involves organizing your cards in a tree
- This tree is not a taxonomy. It merely reflects the way in which the information was consumed or the ideas were created.
- It is better to link cards together with ID numbers rather than by giving cards names and using wikilinks. Wikilinks imply some amount of taxonomic organization of the cards. (Remember, naming things is one of the two difficult things in CS, along with cache invalidation and off-by-one errors.)
- There shouldn't be a master table of contents that one organizes. This gives an implied taxonomic organization of the cards.
One idea from Memexes and early hypertext systems are "trails", which are special entries that include links to other cards. They're sort of special-purpose tables of contents into the hypertext archive. These are important for navigation, and they don't require that the entire archive be taxonomically organized.
Indexing is a good idea. It's basically a curated full text search.
> The index. That much overlooked invention.
I don't know if I'm just misremembering Cat's Cradle as a work of non-fiction, but I think I heard that Vonnegut sat next to an indexer on an airplane once, and they convinced him of the value of indexing. I thought I remember there being an index at the end of Cat's Cradle, but I can't find my copy to check.
The human brain is often described as the most complex object in the universe. Tens of billions of nerve cells-tiny tree-like structures—make up a massive network with enormous computational power. -- Trees of the Brain, Roots of the Mind
Our thoughts literally flow through and exist as tree structures. In the sense that we are the thing that is aware, we are each great forests of neurons. It does not necessarily follow that mapping the world in the form of tree-like hierarchies is conducive to thought but it is suggestive.
As social animals we are bred to be exquisitely sensitive to dominance hierarchies in families and tribes. The less sensitive stumble in the mating dance. The survivors of eons of such selection are good at remembering hierarchies.
The typical Memory Palace works by associating objects with our spacial memory. We seem to have a separate set of registers for recording dominance hierarchies, available for use as a separate memory palace channel.
I can't edit my other comment anymore, but on a more serious note I'd like to debate the claim that the brain has a tree like structure.
A tree is by definition acyclic, whereas the brain surely is not. A neuron can even connect from it's axon terminals to its own dendrites. Surely more cycles are present.
Hence I oppose the idea of hierarchical note-taking.
I think this follows your general thrust. Apéry's constant was mentioned on HN earlier, and to me it links the underlying nature of the vacuum to trees.
I share some similar sentiments with the author and I developed a personal note-taking system which has automatic linking based on the collocational word. You can think of it as sort of like an inverted mind mapping software.
I think a shortcoming of my software and other people’s note-taking software is poor integration of time.. if I could modify my system so I could in effect make some commentary of the changes in the word files I think that would be a much more productive process than simply being able to visualise the changes as a static thing.
I've been using Emacs Org mode backed by a private github repository for a completely free way to take tree-like notes. Did anyone find anything better?
Can self host & sync, bidirectional links, note transclusion (incredibly powerful), built on markdown files, nice interactive graph views and in the case of obsidian there is a good plugin community.
Ummm what? SVN is also built on a tree like structure. SVN is terrible but not because (git) is based on a tree and SVN is not. I honestly don't get his point. Is this just word salad?