Have you ever designed for Unlinked References?

Hello, I am noticing, as I make notes in the daily journal, that, often times, I am Linking and Tagging some blocks that are not really meant to be of prime importance or they are somewhat duplicate info.

I’ll try an example from the top of my head. I have a meeting with Bob and we discuss a Logseq Feature I’d like to be implemented, say Context-Aware TAB (sic) and I make a block note about this. Later on, when I would have researched the matter more closely and would have discussed it with other team members, I would make a block note about the actual details of this proposal. From the two captured notes, only one is a candidate for a prime spot for a Logseq Page Link, while the other would have Bob linked to it as the primary tag-receiver. But I also want to link my discussion with Bob to the actual matter just I don’t want it to be a result for a query on the topic at the same level as the actual KB I write or the proposal document. So i thought that I might use the Unlinked References for tangent-Links like this.

How would that work in practice:

- Had a meeting with [Bob]( [[Bob Penn]] ) and we discussed the feasibiliy of implementing a context aware system in Logseq for the TAB keystroke;

Normally one would create a Logseq Page, say [[Logseq Context-Aware TAB]] and would write the above note as:

- Had a meeting with [Bob]( [[Bob Penn]] ) and we discussed the feasibiliy of implementing a context aware system in Logseq for the TAB keystroke. #[[Logseq Context-Aware TAB]];
OR even better, using the Markdown Link Syntax:
- Had a meeting with [Bob]( [[Bob Penn]] ) and we discussed the feasibiliy of implementing [a context aware system in Logseq for the TAB keystroke]( [[Logseq Context-Aware TAB]] );

But that would be “Hard”-Linked to the Logseq Page, which would be on the same level in results with the more important kb note on the topic. So, instead of “Hard”-linking to the page and have it show up in Linked References or in Queries, a “designed for Unlinked References” would have the name of the Logseq Page with no brackets syntax so it would still resurface somehow, but not as a prime-importance query result:

For this I would use the Title of the Markdown Link Syntax so that I am free to write any text as the link description.

- Had a meeting with [Bob]( [[Bob Penn]] ) and we discussed the feasibiliy of implementing [a context aware system in Logseq for the TAB keystroke](.# "Logseq Context-Aware TAB" );
OR, using an invisible Markdown Link:
- Had a meeting with [Bob]( [[Bob Penn]] ) and we discussed the feasibiliy of implementing a context aware system in Logseq for the TAB keystroke. [](.# "Logseq Context-Aware TAB" );

The technical implementation is of second importance here, I am asking if `Designing for Unlinked References" is something that you think would improve note-taking, linking, resurfacing or it is flawed somehow and I would better stick with linking everything to show up at the same level of importance.

1 Like

Yes. The answer is yes. That is to say, what works for you.
But I have bumped into this issue in a similar way in a completely different context.

Call me stupid (that’s fine!) but I note down data a bunch of different times.

  1. during the day I will list the things I do. (eg. Played [[game name]] #games) on the journal page
  2. Then for each day I make a summary (still on the journal page) This would be a block starting with: [[📓dagboek/verwerken]] : and so that would continue with played [[game name]].
  3. Then for my week review I will look at all the blocks linked to [[📓dagboek]] and write my week highlights on that week’s page. These would again include played [[game name]].
  4. On my month review… same as my week review, but now distilled from my week highlights. So would include played [[game name]].

So eventually my linked references… yeah…

Right now I at least try to not add links in the [[📓dagboek]] section. However quick filling is also worth something.
I’m still honestly not sure what I prefer to do. :woman_shrugging:t4:
Games is a bit of a silly example for this, but it’s not sensitive personal data! For games I don’t really use linked references. But I do for those sensitive stuff as well and it suffers from the same as the games.

In my situation I would have sometimes the same information in the daily journal 3 or 4 times in different contexts: in the MIT of the day I would state a goal. In the journal I would have an entry while doing that MIT, observations, etc. In the Accomplishments of the day I would state again -if that was the case- about the same subject, the same action but different context. Sometime I will have a phrase about it even in the gratitudes of the day and, often times, the same issue would appear in the “I learned Today” section of the daily journal template. Now talk about duplicating the same data :). I clearly have one journal entry that is no1 in importance for the same topic and that one I want to resurface. Most of the times it is either the Journal (“Log”) Entry or the I Learned Today section, depending where I have input the most data.

I haven’t used this system/strategy extensively yet, I am just now thinking of having a system to move redundant or lower weight-score notes from the Linked References to the Unlink References. And, because I am using extensively the Markdown Link in all forms and shapes, I will make use of it when Hard-Linking the Page - [some descriptive text]( [[Some Page Name]] ) and also add an invisible “Softer”-link it to the same topic via another trick of Markdown Link syntax - [](.# "Some Page Name") so that the note remains clean.

I am quite glad that this idea is also on some other people’s mind. Thanks for your input.

I cannot counter that (at least not at the current maturity of Logseq). However, I would not do it myself nor I would suggest it:

  • Unlinked references are not secondary linked references.
    • It is good to have them, but not to depend on them.
    • When they are not empty, they usually contain false-positives, which can hinder discoveries.
  • Conceptually, a tag is already a softer link than an inline reference.
  • Filtering is more flexible than grouping.

I would rather separate the information into more blocks (in this case three):

In journal:
- Meeting with [[Bob Penn]]
  - Topic: Feasibility of implementing [[Logseq Context-Aware TAB]]

In page Logseq Context-Aware TAB:
- Description: a context-aware system in Logseq for the TAB keystroke

Points of interest:

  • Secondary statements should be indented.
    • Ideally only one non-auxiliary verb (even if implied) per block.
  • Shortening the text is more beneficial than shortening the references.
    • The eye can skip known patterns, while it gets tired with long sentences.
  • A description belongs to the page of the described concept.
    • It is ok to begin its life in the journal, but after the concept’s page gets created, it should be moved there.
  • Repeating mentions is (usually) ok, repeating info is not.
1 Like

I get now what you are saying about the maturity of Logseq, and, if this is the case (haven’t been scrutinizing the Unlinked References so far), i see why it is not a good idea;

In Logseq it’s treated the same so I wanted a way to separate the Hard-Links (Logseq Internal Page References) from Soft(er)-Links, which would be some other way to pointing to the same content but from a tangent perspective. It’s like tagging a TODO A, TODO B, etc. However, in Logseq I would rather have the Unlinked References show me a hierarchy of probabilities so if the exact Page Name is found, without the double square brackets, it would be a result at the top of the Unlinked References, then, if all the words are there, at one-two word distance from one another, then the probability of this being related to the page name is second to first, etcetera. I aggree here, again, that the algorithm for this should be very mature and well thought and implemented.

On a side note, I am not keen on Linked References either as it’s often times a mess, not organized by entry date nor some other obvious criteria.

Completely agree. But then I wish Logseq would let me write a custom query for the Linked and Unlinked References. I believe this should be an option at least.

I haven’t yet been sold to the idea of Pages the way Logseq kind of intends them to be. For me Zooming-In on a bullet gets you a page and I would very much like Logseq to treat it as one, to the point of showing the “Page” Graph of all the sub-content of that bullet. I see Pages as Dashboards for these Linked Topics / Tags.

Overall, I agree that Unlinked References in the form of “same-page-name-but-without-double-brackets-so-they-show-up-in-the-Unlinked-References-in-Logseq” is not the most robust way to Soft-Link. Now, that I think of it better, having paid much attention to what you said, I realize they are very fragile actually because I might change the name of the page but the “text” in the Journal Entry that was supposed to be designed for the Unlinked References remains the same.

My original intention to come up with a softer linking system was because of the Search (Ctrl+K), which gets very ugly if the blocks it finds are quite large and it seems to show all duplicate blocks even if I am not looking for “just a mention or a tangent on the topic but which was tagged or hard-linked”.

The solution to my problem actually relies on being able to do simple logic queries straight in the search bar and be able to write AND and OR in there. ATM it seems that the words you type in there are OR-ed and that’s why the more words you input, the more results you get, which is very counterintuitive and not ok from my pov. For me the Search and also the Auto-Completion of Internal Pages and Tags is broken in Logseq. When I start typing [[... and don’t exactly know the exact name of the page I want to reference, I can, in theory, write some words that I remember but as I type more words, especially connection words, the more results I get in the popup-list. But this is another topic altogether.

Now, because of the renaming of pages which won’t update the text if it’s not Hard-linked by Logseq Internal Page syntax ([[]]) and also because I take a more of an Atomic Note approach to every block, I have to re-think this.

1 Like

I do agree. I often would go into unlinked references to make those linked references either.

Conceptually yes, but actually no. For linked references it doesn’t matter whether it is a tag or inline. So have to agree with @FlorianF would be nice if we could differentiate.
Though I also appreciate that I don’t need to think whether to use [[ ]] or # for linking. I have custom css to change the look of tags, but not double brackets.

I believe the journal pages are at least in reverse chronological order, but after that it is indeed a mess.

I think this is also why I often opt to use custom queries over looking at linked references. We can get much more precise results.