searchable blobs as unnamed groups enable unnamed formalization
It’s worth noting how people use analog tools – like pen and paper, whiteboards, and maps – to structure their thoughts. We’ve mentioned the implicit structure of clustering, and this idea can be carried through as a pattern for pulling together unnamed things into an unnamed, dynamic collection – a blob.
Analog spaces enable blobs you don’t have to label or name. At these early stages, you don’t know what form the group will take on. It is difficult to predict whether structure now will be worthwhile later. This enables unnamed formalization of collections of thoughts, which allows for later retrieval that doesn’t depend on names.
Most current tools require you to label relationships if you want to reuse them in the future. We posit that there users should be able to specify unnamed structure through loose methods.
Exceptions are in spatial tools like Miro or Figma. One of our interviewees made a variety of maps in Miro. They were able to look back at a collection of items with a lasso around them and the arrows the collection pointed to, and jump back in quickly.
This is also seen in most software we’re familiar with, too. When a new document is created in Google Docs or Obsidian, the file is initially labeled ‘Untitled (1)’, and a direct search can later find words within the file. This is to show our familiarity with this frame.
In our own work, Roam Research allows Joel to construct blobs by chaining things through block references; this linked exploration becomes a trail. Direct search doesn’t enable this in most tools. But faceted search, where you can filter and refine your search as you go, allows for blobs to form. This can be created through Roam’s filters on a page’s linked references.
Workspaces as a primitive could make sense here. We could infer that items in the same workspace are all related in some way, and enable users to search that sort of relationship.
Section below is Authored by:: Rob Haisfield
In his search for an answer to What is the data structure of a graph built to facilitate decentralized knowledge synthesis, Brian James Rubinton said that the XML for knowledge graphs will have URIs for triples, for links, for blocks, and pages. He said he only thought you should be able to link to pages, not unnamed blocks. But when I look at an unnamed blocks, I believe that’s where the act of synthesis happens. These unnamed blocks might reference multiple named ideas, but since the author is forming the connections in their head, they have not yet been able to put a name to it. [^1]
Jump will look at your statements, draw out a sort of molecular diagram of the concepts represented and their relationships, and then search for other areas in your knowledge graph with a similar shape. None of that requires users to name the idea.