- Sep 2024
-
www.matthewsiu.com www.matthewsiu.com
-
One key motivation for Latticework was how wonderful it feels to stumble upon a past moment of shining clarity, to point and revel. We want to be able to carry those moments with us, to see them all at once when we’re lost, and to use them as landmarks as we navigate our messy notebooks. We’ve used Latticework to do this in small ways so far, and we’re excited to see how our upcoming projects might feel different with its extra affordances.
this paragraph reads like making commonplacing navigable in a new way. Also turns 'snippets' into potenital entry points without them being separate notes, and pivots like tags. Note the clear spatial overtones (landmarks, being lost, navigate, ways, stumble upon, point).
-
We had a strong personal motivation for this project: we often find ourselves stuck in our own creative work. Latticework’s links might make you think of citations and primary sources—tools for finding the truth in a rigorous research process. But our work on Latticework was mostly driven by the problems of getting emotionally stuck, of feeling disconnected from our framing of the project or our work on it.
Again the important distinction, here in the context which itch Latticework scratches, between 'evidence' and 'kindle' perspectives. The latter is an emotional thing, where knowledge is not an external thing, but a internal network of meaning.
-
Our test users were largely quite enthusiastic, but our sessions with them usually lasted less than an hour. Future work should be informed by extended use in demanding situations.
I recognise this. We spent an hour in Feb. Which was fun, and useful because it was a real effort on actual notes and for an actual purpose (for me a workshop design). Then afterwards I didn't use it much due to tech hurdles, so I didn't get to experience ongoing value. Reinstalled it now because of this article. (Which [[Maarten den Braber]] pointed me to.
-
Adjustable snippet ranges. After working with a snippet link, some test users found that they wanted to shift its endpoints, to include more context or to tighten its focus. Latticework doesn’t currently allow this, but one could create an interaction which modified its current snippet links accordingly
adjustable snippet ranges, letting your emergent insight impact your original highlighting/annotation sounds like a very interesting idea. Not because you're pinpointing the info at source more accurately, but because the emergent purpose of your sensemaking reflects back on your source material. It shifts the exact point where your Surprisal originates around.
-
Giving a cluster a name can impose formality prematurely, adding friction to the process.
Naming clusters can be incorporated into sensemaking efforts though, when not used as result but as intermediate step. As in [[2 step archetype extraction 20121130152904]]
-
we rarely know the shape of our categories in advance. Often we’re just reacting: “this seems important”; “this is related to that”; “this makes me think of…”; and so on.
exactly. Tags are emergent structure, and are not per se to describe the information stated nor to be used as a taxonomy. Vgl [[%On Tagging 20200818120917]] as associative emergence, as search/find history, as pivots in an exploratory path etc.
-
Text as a medium for sensemaking. In QDA tools, the “working document” where you make sense of your excerpts might be a spreadsheet, or a database query, or a whiteboard. By contrast, Latticework emphasizes a textual canvas, where freeform notes and snippets can mix arbitrarily. That mainly comes from a difference in the role of the snippets: we view them less as “evidence” or “data points”, and more as “kindling” which might be consumed and discarded on the way to insight. In the latter setting, when even the problem being solved is undefined, the only way forward is often to write in circles, until some sense starts to emerge. This writing may weave chaotically between new observations and snippets from old documents. Some QDA tools, like Dovetail, include freeform text editors, but their affordances emphasize communication to stakeholders, rather than sensemaking.
This is good way to make the distinction with qualitative research tools, incl. those where the narrators of qualitative bits do their own signification which then serves as filters. Tagging like those two types serve a different purpose, spotting patterns 'out there' rather than provoking thinking 'in here'. Both useful and not unrelated, but different activities. The 'evidence' vs 'kindling' metaphors make sense to me. Diff points of application.
-
Latticework’s portal-based marginalia allow commentary to be created and viewed from either “side” of a snippet link, interchangeably. Snippet links can be serialized to standard URLs in standard Markdown plaintext. (Modern systems only support links to blocks, not arbitrary text ranges. For block links, Obsidian uses non-standard anchors; others use proprietary database formats.) Snippet links can be quickly created (either from source or at the destination) using key commands. Transcluded snippets can be collapsed to increase density, and re-expanded when needed. Pane-based navigation allows users to preview and visit links while maintaining a consistent view of the linking pane.
5 reasons L's bidirectionality is different to pkm tools / Xanadu. The first two are most key imo: they can be initiated from both sides of a link, and result in the same thing. (e.g. diff from [[Webmention 20200926203019]], and two the links are standard URLs (in standard Markdown), whereas Roam / Notion abstract them away in a db, moving outside their role as viewer, and Obs maintains its viewer role, but adds things to the notes not obviously interpretable outside of it.
I'd add that the linked snippets are a different unit of linking from what Obs et al support. Much closer to the granularity where the knowledge work is done. Surprisal I rarely find in a paragraph, mostly in part of a sentence. Questions latch onto a single word sometimes.
-
Snippet links are a kind of hypertext, and embedded snippet links are a kind of transclusion. Originally proposed by Ted Nelson as part of his Xanadu system, transclusions present part of one document within another, while maintaining bi-directional links for navigation and orientatio
Bidirectionality here explicitly tied to Ted Nelson's Xanadu. Call block transclusion in pkm tool transclusion "primitives" which sound like a right charcterisation. Say Latticework differs from both
-
Latticework’s design evolved through many iterations, driven in large part by user interview and observation sessions. Through our personal networks and the Obsidian forum, we recruited experienced Obsidian users who needed to distill some insight from a large collection of unstructured notes. We’d like to discuss some observations from those sessions, and from our own use of the system.
I did #2024/02/01 https://www.zylstra.org/blog/2024/02/matthew-and-andy-watched-me-test-the-obsidian-reference-plugin/ where my own similar observations are captured. - shifting from highlighting to linking as emergence occurs - structure is earned over time - keeping the copy commands straight was hard (I kept forgetting too) - preview and multiple panes is what I did too, but it seemed to clash with the plugin then - selection of snippets is not blocks, but phrases inside par's and sentences. Not whole blocks (and this is why I hardly use block links inside my notes) - after paraphrasing collapsing snippets keeps overview possible, or to collect examples, and treat list as tasks, collapsing when done.
-
We’ve been careful to implement Latticework’s features in the same spirit. Snippet links are stored as ordinary links using standard W3C Selector URL fragments to specify an arbitrary text range
Using Latticework does not break the 'Obsidian is only a viewer' principle cf [[3 Distributed Eigenschappen 20180703150724]]. It adds markdown style links according to W3C selector url standard. Nice, because it maintains readable plain text files.
-
Alongside disorientation, working memory overload is one of the biggest problems when distilling these large unstructured documents. We believe that’s why people in these situations so often try to collect everything important into once place: that way, everything can be viewed at once, and it’s possible to notice connections and themes without relying on working memory. Unfortunately, as snippets accumulate, the working document itself can become quite long—leaving you stuck scrolling around, trying to remember where everything is.
The processing document can get as unwieldy as the source material for which it is a solution. Latticework lets you collapse stuff therefore.
-
Latticework uses a similar pane-aware interaction
This pane awareness is what seemd to clash with some other plugin I run. At least it did in feb. I notice their code repo still warns about clashes with other plugins and to run it in a separate vault with no other plugins.
-
You’ll get the same result no matter which direction you go—a highlight in the source document and a snippet link in the working document. Conceptually, highlighting doesn’t actually modify the source document. Highlights are a dynamic style applied to all the snippets linked in your working document. So if you delete a snippet link, the corresponding highlight will disappear, too
Bi-directionality is a key feature in Latticework, which is great. At the same time, on the source end it is also ephemeral. If your remove a linked snippet from the sensemaking document, the highlight, which is just a styling element, gets removed from the source. The source is not modified to produce the highlight. (Any permanent link to source should be made consciously, which is right) Bidirectionality, other than linking seems to me a key affordance in #pkm, something that not now exists in either my annotations / reading / processing flows. #openvraag: where in my workflows would bidirectional trace leaving be useful.
-
Latticework’s main goal, then, is to enable fluid movement between these foraging and sensemaking stances. By extension, that means fluid movement between acting on source documents (which emphasize foraging) and on your working document (which emphasizes sensemaking)
Latticework sees foraging as tied to source, and sensemaking as tied to working document, and aims to make the movement between the two fluid, so you can shift focus between the two docs and thus the two activities. It leaves traces of your work in both (vgl [[Hoe emergence tot stand komt 20040513173612]] for the role of such longer traces to more easily stumble across in emergence.)
-
this process isn’t linear. It’s often convenient to do a bit of preliminary sensemaking in the midst of foraging; conversely, observations you uncover during sensemaking will often lead to another round of foraging, and so on, in a loop.
Making sense of material is not a linear process of ever more refinement, as e.g. Tiago Forte suggests with [[Progressive summarising 20200922080651]]. Siu/Matischak embrace the non-linear, recognising you go from 'foraging' (their term, great, K-garden style) to annotating, rearranging, noting an idea, back to foraging, back to rearranging etc. This is a key thing imo.
-
Latticework is built to support the workflow we described in the introduction
The video demo also shows adding comments to a snippet either in source or in the sensemaking document. I wonder if they put that in the released plugin (the say it doesn't do everything they did in the research project)
-
when you’re trying to make sense of a confusing situation, you need to get everything into one place, where you can see, rearrange, and elaborate the pieces into a new whole.
Intended effect of Latticework plugin (great name btw) Latticework allows you to fetch snippets in one note, and paste untalterable into another, with a link added in the original and copy. In the new note you can rearrange, paraphrase etc, purposefully add a link to source material, and otherwise do away with the snippet over time. Allows a better overview of what comes from where etc, preventing getting lost in the source material, which often happens.
As annotations already flow into my notes this helps reinforce their use.
-
This is the 'final' result of [[Matthew Siu]] [[Andy Matischak]] research into an Obsidian plugin for making sense of several sources into one, emerging an outline. I tested an earlier beta on #2024/02/1 [[Andy Matthew Obsidian plugin]] https://www.zylstra.org/blog/2024/02/matthew-and-andy-watched-me-test-the-obsidian-reference-plugin/ I stopped using it after a few weeks due to clashes with other plugins I could not pin down. At first glance this is a good description of the process and intended purpose. Re-installed this version of the plugin.
-