I decided at long last to go with Obsidian. But there is something missing. Namely, linking to some concept, and then clicking it to make a page gives an empty page. On Logseq, the relevant linked blocks (and even unlinked blocks) are already populated there.

There are Obsidian plugins which improve on the native backlinks at the end of documents, namely Influx (no more development sadly) and Strange New Worlds. And those are pretty cool, but they don't actually populate a page.

Also, as someone posted on reddit, in Obsidian "There seems to be consensus that 'folders are unnecessarily restrictive', but… that same problem is just moved down to the note-level." This has been my experience as well.

Looking at Logseq, the link handling is great, but indeed it is a bit "clunky" and hard to navigate to information. On the other hand, the idea of daily journal being the only entry and atomic notes is really appealing. You could certainly make a detailed Contents page as they recommend.

Any thoughts on this? I am still leaning toward Obsidian, but the lack of unlinked references not appearing is very annoying - these require a specific search to find. I am aware I could make a tag tree, but this would probably end up worse than a straight folder structure in any other program. Lastly, when you have a topic, say "Obsidan" where you keep Obsidian tips, and then you link that inside pages on some aspect of organizing Obsidan (say "Obsidian for research"), you have the annoying issue that now you have to organize where the "Obsidian" link-page is shown, and the other notes pages. I could go completely flat, but some notes like "Meetings" and "Contacts" surely need some organization, right?

Does anyone have ideas on how to accomplish a more seamless collection and navigation of PKM in Obsidian? Or a compelling reason to use Logseq instead, where notes are interlinked by default but finding them again may be challenging?

You don't have to choose. You can run LogSeq and Obsidian on the same set of Markdown files. As long as you don't have LogSeq set to do things like timestamps, this dual approach seems to work well. A few tweaks in each program may be needed --- for example, determining where Obsidian should store daily notes and in which format --- but I have used the dual approach for a long time. YMMV.

Thanks, that's great to hear!

I tried the Logseq Contents as they intend it, ie. [[]] for topics which are simply tags with the #. It works very well to automatically create a cross-referenced knowledge base. And then the pages themselves (likely daily journals) become completely irrelevant. And if you forget a tag, no worries as unlinked references are also included! There is absolutely no friction with this method.

So for knowledge management features, Logseq is truly amazing. Obsidian is of course superior for look and feel, speed and size of db, plugins, dataview, canvas, drawing. It's actually a shame it doesn't offer the option of going block based with automatic cross-referencing, which really makes organization completely unnecessary - with the folder/link/tag structure it has, I see all the power being used to suck users into constant tweaks of organizational method.

I will use Logseq as the main PKM and daily driver. And then try Obsidian to see if I can get it working to any great extent. Perhaps that would be to "clone" the Logseq contents somehow, or just for the occasional long-form input.

It’s well worth trying Tangent on the same data repository; that works very well, and it’s much less “weighty” than Obsidian, while having some very nice knowledge management features.

Tumbleweed wrote:

Thanks, that's great to hear!

I tried the Logseq Contents as they intend it, ie. [[]] for topics which
are simply tags with the #. It works very well to automatically create a
cross-referenced knowledge base. And then the pages themselves (likely
daily journals) become completely irrelevant. And if you forget a tag,
no worries as unlinked references are also included! There is absolutely
no friction with this method.

So for knowledge management features, Logseq is truly amazing. Obsidian
is of course superior for look and feel, speed and size of db, plugins,
dataview, canvas, drawing. It's actually a shame it doesn't offer the
option of going block based with automatic cross-referencing, which
really makes organization completely unnecessary - with the
folder/link/tag structure it has, I see all the power being used to suck
users into constant tweaks of organizational method.

I will use Logseq as the main PKM and daily driver. And then try
Obsidian to see if I can get it working to any great extent. Perhaps
that would be to "clone" the Logseq contents somehow, or just for the
occasional long-form input.

Tumbleweed wrote:

Or a compelling reason to use Logseq
instead, where notes are interlinked by default but finding them again
may be challenging?

I recommend that you make use of tags and operator queries (which can be saved and generate an instant results page in table or list format), I think this will make it easier for you to control the search for information and the general organisation of the whole system.

You can also take a look at "namespaces", which allow you to hierarchise datasets. But you should use them carefully so as not to distort the spirit of Logseq as an associative system (you can find out more on the Logseq forums).

Wow, thanks @tberni the saved queries are indeed an amazing and powerful feature of Logseq! It's really akin to UltraRecall "Saved searches", Zoot "Smart folders" or InfoQube grid criteria. I'm still leaning toward going with Logseq as my daily journal and PKM, as well as for articles - the automatic intertwining is amazing and frictionless. And having the Contents open as you write Daily Notes, to add new topics of interest, is just an amazingly simple and powerful way to make sense of the structure - and they can also hold queries.

Regarding Tangent, it looks like a "light" version of Obsidian. Without the block structure and auto cross-linking of Logseq, I don't know why it would be needed.

Logseq is clunky though (although that could easily be improved). The results of Live query, or a [[]] tag page you create with Linked and Unlinked References, is just very blocky (pun slightly intended). The Obsidian plugin "Influx" for instance gives much more easily results to visualize. But no Unlinked References!

Obsidian is so tantalizing every time I try it. Looks amazing, tons of power and plugins to do everything. Influx mimics Logseq cross-linking to some degree. BUT, there are two massive problems:
1) Lack of block structure means either just using Daily notes (which is fairly messy) or organizing everything (which gives friction)
2) The organizational structure of folders/links/tags leads to too many organizational choices (for me anyway). Too much time and thought to the system instead of just using it.

Hence, Logseq seems like the obvious choice, for me anyway.

@Tumbleweed

Sounds like LogSeq is a good fit. But regarding what you mentioned about Obsidian, I assume you have tried tweaking the settings for the Backlinks core plugin, including the "Backlink in document" setting? "Unlinked mentions" should show up.

Also, regarding organization, if you haven't already seen it, you might be interested in:

https://github.com/makary-s/obsidian-metafolders

Hi, yes thanks I saw metafolders in a post on the Obsidian forums. I'm a bit wary of replying on 3rd party plug-ins for my methodology. Also, I am just not sure I want to have to have an organizational method, it's kind of why I'm moving from UltraRecall and InfoQube (which is awesome by the way).

I can't get Unlinked mentions to work properly. Or Influx, it just disappears a lot for no reason. Have blocks built in really means all notes can be from Daily Journal with no regard for organization. And if I import a doc or larger website clipping, then I can have a standalone note or enter it in Daily even - and let the linking of topics occur automatically.

These are the two reasons why I'm getting more sold on using Logseq. I tend to use a system for a while, get hung up on limitations in the setup or methodology, CRIMP hard, try to fix it, and move to another software. Hopefully Logseq will alleviate this!

Sometimes the only solution is not to play.

I realized that Logseq was not operating as I wanted it to, namely putting two links in one block would not link them together, but rather link each to the block parent. Although this is by design, it means that creating noteworthy links is a massive chore. I question the design choice, as the block centric view seems to me to take care of the "accidental linking" phenomena. Just add multiple links to new blocks under the main, and make new blocks for new thoughts or insights. But alas this is not implemented.

Even the way I was trying to do things had some fatigue in trying to keep linking things. And when the links became substantial, I couldn't see them well on the topics page, there was a large amount of scrolling so getting the value of linking was difficult. Another substantial issue was adding scientific articles via PDF, where all the many co-authors got added to the graph, making it useless after just 10 papers or so since the author lists would overpower my actual entries. Furthermore, the tools to rapidly cite based on your PDF notes are suspiciously absent. Obsidian wouldn't solve any of these issues, unfortunately.

So I made the radical decision to go back to a more traditional outliner. In these, I can see MUCH more information at once and actually get a broad overview much faster. It can be made into a Zettlekasten with links, which should be more deliberate.

My choice for what to use came down to this: I want to quickly and easily document my work - there are other uses, but this is the one that has the most friction. I have active licenses to UltraRecall, Zoot, RightNote, InfoQube, DoogiePim (V3), Heptabase, EssentialPim, Obsidian/Logseq (free), and have previously used Mybase and several others.

However, I decided to CRIMP and found a new program that is ideal for this use case: MyInfo 8. It is so fast and useful for my purposes! I'll make a review to point out the main reasons for using it and the few quibbles. To be clear, I'm also going to use EssentialPim for email, tasks and meeting notes, and Citavi for PDFs and academic research.

Tumbleweed wrote:

It can be made into a Zettlekasten
with links, which should be more deliberate.

I was never happy with zettelkasten in Obsidian etc.
Partly notes inclined towards long. The links carry no indication of a train of thought and, in practice, links on a note can reflect very distant trains of thought. And convenient workflows are rarely zettelkasten related.

And when it comes to it I came down on the folgezettel being important side of the debate.
And a traditional outline/mindmap automatically has a folgezettel structure. So long as the placement of a note is done with zettelkasten deliberation it's all good, and there's an inclination to short notes rather than long. And links and backlinks can still be done.

Dormouse, you make an interesting point about a traditional outline or mindmap program being better.

Would you agree that a traditional analogue approach - using a pen and index card in the spirit of Luhmann migh also be better than the digital approach?

Thanks.

Daly

Dormouse wrote:

Tumbleweed wrote:
It can be made into a Zettlekasten
>with links, which should be more deliberate.
>

I was never happy with zettelkasten in Obsidian etc.
Partly notes inclined towards long. The links carry no indication of a
train of thought and, in practice, links on a note can reflect very
distant trains of thought. And convenient workflows are rarely
zettelkasten related.

And when it comes to it I came down on the folgezettel being important
side of the debate.
And a traditional outline/mindmap automatically has a folgezettel
structure. So long as the placement of a note is done with zettelkasten
deliberation it's all good, and there's an inclination to short notes
rather than long. And links and backlinks can still be done.

Dormouse, I agree with your points 100%. I wished I had seen your post 20 years ago and found MyInfo! The folgezettel method of child notes is very natural. My mistake was in worrying about the exact location of notes, as if they might become lost. Occasional review allows one to clone them elsewhere, or link in new notes which refer to these older topics.

Deliberation as you mentioned is important, to avoid making a messy repository without useful links. Again, spot on with the issue I had with Logseq, as those links had no intention and were not useful. I do occasional like to break down long notes into shorter child notes, which then can be cloned where needed when they represent a new topic or subject area.

Daly, I see your point but really any tree software can be used for a Zettlekasten with the advantage of very rapidly moving from note to note. Just the time saved alone is worth digitization, not to mention the archival aspects. I do have a yearly journal I write in every evening, and both A5 dot notebooks (as an idea sketchbook) as well as A6 to carry. I plan to digitize important notes from the A5/A6 into MyInfo and keep the yearly journals for future reflection.

The idea of a mind map is convincing at first glance. Unfortunately, the structure is limited to a tree structure. A node (thought, note) can only exist in one place. This seems to be a disadvantage compared to programs that allow the note to be saved (cloned) in more than one place.

This is one of the things I found useful about TheBrain. Any node (thought) can have existing nodes as children, parents, or "jump" nodes.


Bernhard wrote:

The idea of a mind map is convincing at first glance. Unfortunately, the
structure is limited to a tree structure. A node (thought, note) can
only exist in one place. This seems to be a disadvantage compared to
programs that allow the note to be saved (cloned) in more than one
place.

Daly de Gagne wrote:

Would you agree that a traditional analogue approach - using a pen and
index card in the spirit of Luhmann migh also be better than the digital
approach?

I believe that will depend on the individual and the definition of better.

It would certainly allow a process more precisely similar to that of Luhmann. Though he was supported to have an almost clockwork like daily routine, and I doubt that can be replicated today. And the general speed of media and requirement for output to be digital creates a different environment for the mind.

Tumbleweed wrote:

The folgezettel method of child notes is
very natural. My mistake was in worrying about the exact location of
notes, as if they might become lost. Occasional review allows one to
clone them elsewhere,

Implementing a Luhmannesque zettelkasten requires worrying about the exact location of notes though.
They have to be entered in the sequence in the right place. Which means review is daily at least.

Bernhard wrote:

The idea of a mind map is convincing at first glance. Unfortunately, the
structure is limited to a tree structure. A node (thought, note) can
only exist in one place. This seems to be a disadvantage compared to
programs that allow the note to be saved (cloned) in more than one
place.

Many mindmap programs allow cloning or mirroring.
But Luhmann's methodology is about the tree structure, with links and references to notes in other places in the tree. That may be good (or not) but it was his system.

Amontillado wrote:

This is one of the things I found useful about TheBrain. Any node
(thought) can have existing nodes as children, parents, or "jump" nodes.

Many programs allow "jumps" and they can be simulated even in a single tree outliner like Workflowy by adding new top level notes (which was what Luhmann did in his card index).
tbh I'm unconvinced by using a mindmap for zettelkasten, conceptually it's the same as an outline, but working with it is quite different cognitively

Dormouse wrote:

Many mindmap programs allow cloning or mirroring.

Can you name some of them? Thanks!

Both Freemind and its fork Freeplane have cloned nodes.

Regarding the limitations of a tree structure and the implementation of node "cloning" in mind map programs:

Personally, I find it much more helpful to have a "concept map", where an item can have multiple parents. So, rather than cloning nodes and showing multiple copies in the map, only one copy is needed and one can view its various parents.

The ideal, in my view, is to have software that has both an outline view (that can show multiple parents) and a "concept mapping" view (i.e., a map view that can show multiple parents). Options for this are extremely limited. A few that come to mind are: CMapTools (a bit out of date), InfoQube with its Surface view, and Fibery (online only, a bit finicky for this purpose in my experience, and requires a slightly complex set-up).