XaiJu
Andy Matuschak
Andy Matuschak

patreon


[Preview] Latticework: unifying annotation and freeform text editing for augmented sensemaking

Longtime patrons may remember the research fellowship I launched at the start of 2023. I spent a year working with the winner, Matthew Siu, on a new sensemaking tool. Today I'd like to share a preview of our report on that work!

Read report preview ✨

Comments and feedback are welcome.

I've excerpted the introduction below, so you can more easily decide whether you'd like to click through:

Introduction

You’re in the middle of a large project, and you’re stuck. You’re staring at a pile of long, unstructured documents (journals, lab notebooks, meeting notes) with a sense that the clues you need are buried somewhere within. But there’s simply too much to keep in your head as you scroll. Instead, you hatch a plan. If you could find the important snippets and gather them into one place, it would be easier to see everything at once, to notice patterns, and to figure out what to do no ext.

So you create a new working document, and as you read through your unstructured notes, you copy and paste key passages into it. But this process is cumbersome and disorienting: you’re constantly moving both your attention and your cursor between these documents, deciding where to put each new snippet, losing your place in your unstructured notes. The snippets in your working document are isolated from their original context, so it can be hard to remember what each means and why it matters. And as snippets pile up, your fresh document can easily produce the same kind of overwhelm you were trying to escape.

If you’d been working with books or academic papers, rather than your own notes, you might have used annotation tools instead. It’s easy to enter a satisfying flow with these tools. When some phrase strikes you as important, a highlighter lets you capture that reaction immediately, without shifting your focus. Likewise, marginal comments let you record observations as they occur to you, in context. The trouble with these tools is that your annotations end up trapped in the pages of the original documents. But 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.

Annotation tools feel great to use, but they don’t support the follow-up thinking you need to do. Text editors give you a flexible canvas for making sense of snippets, but their design is often cumbersome and disorienting when used in this way. If you could move fluidly between these tools, you could use each where it excels and, perhaps, get the best of both worlds. In this paper, we present Latticework, a system which unifies annotation with freeform text editing, in the context of personal knowledge management tools.

Comments

About the idea nd the research: The ideas sounds reasonable and interesting enough to help with modern information overload and information process problems. Here are my thoughts on reading the article. - almost for the entire article I was thinking about something like an OS-level implementation that lives outside of application such as Obsidian (or any other). At the end I saw that those are part of the ideas for future exploration. I think that getrecall.ai is a tool that tries to create platform for collecting information, but lacks the Latticework-driven outcome (boards and backlinks) - one of the comments already mentioned Heptabase which is a great tool, but tries to do too many things at once. However, it provides canvas based approach to organising notes for "sensemaking" - I can't stop wondering if there are any thoughts on creating a bridge between "Orbit" and "Latticework" or any other Question bases product (Anki,etc). - highlights and copying snippers into new document is great, as I try to apply something similar when I do a research for a work project. However, in Latticework demo I lack the highlight and bold on top of the collected knowledge. I know that this has been mentioned as a pottential improvement - so I'm thinking that something like a Progressive summarization technique (from BASB - Tiago Forte) can be useful. Personally I see that I create a lot of research documents where I paste the same snippets but I bold different parts of the text as the context is different - again on highlight the highlight, as the problem was stated I was thinking a lot about making each new highlight read-only (technique similar to: Incremental Note-taking) and Concurrency Locking mechanisms (maybe each next highlight can be locked for read-only by another set of highlights with backlinks and lock counters.... - and again on highlights - I also have an parallel/alternative idea that each highlight may create "Draft" or new note that is read only. This reminds me to some degree to a database relationship (many-to-many) - or maybe you can think of implementing something iike the Readwise interface. I really love and like the product, but what I don't like is that they don't give me a proper text editor for the highlights. Once all is collected together they can live within the document, but on-top of that they can create new relationships and copies with backlinks with thought expansion (this somehow reminds me about the way you create you Evergreen notes and the idea of Zettelkasten for creating conversations with your past notes. In the ZK is even more read-only as I can discuss my past notes, referring to them but not changing them - only expanding notes - also incremental note-taking I think is stepping as an idea to a similar approach). What lacks in the ZK notes and indices is the backlinking and outlining of the Latticework. However, I think that in your YouTube videos where you presented reading and note-making/taking you have shown similar approach to collecting highlights with multiple notes open for easy referencing (but lacking the bi-directional links) For me the most exciting idea is a child between getrecall.ai and Latticework within Obsidian. Having an system that I can simply collect excerpts and notes that makes auto-links to Cloud files or downloaded pdfs, and makes discover and search easy with AI. Also, Latticework backlinks and note-taking approach can further augment the experience in my oppinion.

Марио Павлов

The most interesting aspect is the Chain-of-Notes, CoN. Different notes relating to different non-contiguous pieces of the text linked by the logical thread of an idea and aggregated into a single CoN, potentially a new object of knowledge. It would be natural to extend the note with the inclusion of images or videos taken outside the text. The two panels always open could be an element of distraction, loss of focus, while you are reading. In some situations you just want to be focused on the text.

Piero Rivizzigno


More Creators