Thinking and writing in a decentralised collective intelligence ecosystem
What would a Tool4Thought designed to support decentralised collective intelligence look like?
Part of a bigger picture.
This is one of several posts exploring how a decentralised ecosystem for collective intelligence could be bootstrapped into existence. As set out in the introduction to the series (A Minimum Viable Ecosystem for collective intelligence), I’m exploring these questions by re-imagining MyHub.ai, a publishing platform I alpha-launched in early 2020 to explore the interlinked challenges of collective intelligence, decentralised social media, and community-owned AI algorithms.
I now see MyHub.ai evolving into an open-source toolkit, with templates for not just Hubs but also blogs, Substack-style newsletters and traditional websites. Moreover, this would just be one toolkit among many, with all toolkits interconnecting via a decentralised-social, open-source and self-sustaining ecosystem for collective intelligence.
The best way I can explore all this is nevertheless to use MyHub.ai as my lens, and describe how it could evolve and grow in this ecosystem.
Content Management System or Tool4Thought?
The first upgrade is to replace each Hub’s content management system with a super-simple Tool4Thought designed to support not just note-making, but also writing and publishing across social knowledge graphs.
There are as many Tools4Thought (Tana, Obsidian, LogSeq, Roam, etc.) as there are names for this software class: while I use Tool4Thought, Tiago Forte calls them Second Brains, Cory Doctorow prefers Memex, others note-taking apps or digital gardens (my tag, however, is #2ndbrain).
With such a variety of tools available and more on the way, I don’t really want to build another one. Instead, I’d prefer an ecosystem from which users can choose which Tool4Thought they want to use, which is why some friends and I are mapping Tools4Thought using massive.wiki via this site.
For a Tool4Thought to support a collective intelligence ecosystem, however, there are two requirements:
Your public website as seamless extension of your Tool4Thought
The MyHub Tool4Thought does more than help Hub Editors organise their notes: it’s designed to help them turn their notes into publishable content, and is seamlessly integrated with both their public website and the wider collective intelligence ecosystem.
Of course, this assumes that users of Tools4Thought want a public website, although this is not strictly necessary — users can also use the tools described below to think without publishing anything, and can even collaborate with others on content which will never be made public.
However, the tools and ecosystem described in this whitepaper are tailored for people — I call them Hub Editors — who want to not just think, but also share their thoughts with the world.
Whatever Tool4Thought they use, today’s creators are supposed to do their thinking in them, and perhaps prepare an initial draft of whatever they intend publishing. But until recently they had to finalise and publish that draft using a completely separate tool — their website’s Content Management System (CMS). And then they need to share the published piece via yet more tools — their social media platforms of choice.
But why not make your public website a seamless extension of your private Tool4Thought?
But why not make your public website a seamless extension of your private Tool4Thought? With a Tool4Thought for your Hub’s CMS, you can manage everything you have — your bookmarks, ideas, notes and drafts — in one place. Set any content to ‘Public’ and it appears on your public site.
Today, of course, Tools4Thought are beginning to allow just that: users of tools like Obsidian, Dendron, Logseq and others either feed some of this content into their public website, or simply publish some or all their notes as a “smart notebook” or “digital garden” (for example: Andy Matuschak’s working notes, Boris Mann Digital Notes Garden).
Personally, I enjoy exploring digital gardens. However, they publish by design thousands of interconnected “atomic notes”. Many people, however, don’t want to share their raw notes. More importantly, most people — including me — want to read (and write) more than just atoms, as humans generally absorb knowledge better when it is packaged into narratives.
To stretch the metaphor: truly interesting pieces of content are like living organisms, composed of many molecules, each composed of atoms. Weaving a new one into being means publishing more than its constituent atoms, particularly if we want to network both these private Tools4Thought and public sites together to create shared Social Knowledge Graphs. In an ideal world, therefore, we need systems which help us seamlessly publish both:
“Here is my current thinking about this, and behind this link are the notes, ideas and resources I built it from”.
Super-simple support to the writing process
Most Tools4Thought focus on helping people organise their notes rather than transform them into publishable content, and are not easy for newcomers to adopt.
This may be a controversial opinion for those who live and breathe Tools4Thought, but most if not all of them seem designed less for writers and more for productivityporn geeks obsessing more over customising their Tools4Thought than actually using it to get something published.
"The newcomer is not only asked to learn a new software, they are asked to change the way they work and think." - Mapping the Tools4Thought landscape
Tools4Thought quite deliberately claim to mimic how the mind works, with every note linked to many others without hierarchy. As others observe, however, this may be useful for note-taking, but less useful for writing:
“**everything in your mind is already… a graph… big and complicated, with way too many connections everywhere. There’s no good in replicating that in digital form…” - PKM is bullshit
At the end of the day, the only person who can read your knowledge graph is you. For almost everyone else you need to provide a linear text, not a “hairball” of unstructured notes, if they are to learn from you.
But getting from a messy graph of interlinked notes to a linear narrative is hard — it’s commonly called writing, although I also like the phase change metaphor, with atomised notes forming narratives like a gas cloud desublimating into highly structured crystal.
IMHO if we are to achieve collective intelligence we need to make it easy for everyone to contribute without having to learn complicated software. An easy-to-use Tool4Thought which supports the creative flow from idea to research to draft to publication is therefore essential.
How might it look?
Supporting writing with the content pipeline
The MyHub Tool4Thought helps Editors move ideas through a “content pipeline”, which also helps integrate them into the wider, decentralised collective intelligence ecosystem described in Social knowledge graphs for collective intelligence.
The content pipeline is deliberately simple, based on two key principles:
- filter to focus: spend 90% of your time on the best 10% of the content coming at you through the information firehose
- use progressive summarisation to embed that knowledge in your mind, as it’s rarely enough to simply read a piece of content to absorb its content— you will learn more if you summarise it in your own words. The more you iterate — ie, summarise an article, then summarise the summary, etc. — the more you learn.
Like any pipeline, the content pipeline illustrates a process: inputs appear left, and outputs leave to the right (any similarities to ActivityPub are a good example of convergent evolution):
The Content Pipeline supports the sensemaking journey by helping Editors transform a cloud of unstructured notes and ideas into fully publishable texts via a series of intermediate steps.
It’s also a funnel (not shown as I have zero design skills), because there’s a lot more content beyond the left end than there is on the right.
The information firehose, far left
That’s because to the left of the pipeline - ie, outside the fiture - lies the Internet: a structureless cloud of content, aka “The Information Firehose”. Only some of that content is valuable enough to be read. An even smaller proportion is worth learning from, incorporating into one’s own thinking and perhaps even published content. Managing that firehose efficiently is therefore essential.
Inside the pipeline
Inside the pipeline is where Editors perform sensemaking to create something valuable enough to push out the right end: weaving their own and others’ ideas to create something original.
Most Tools4Thought don’t help with this off the shelf, and simply provide an unstructured cloud of atomised notes. The Content Pipeline supports the sensemaking journey by helping Editors transform this cloud into fully publishable texts via a series of intermediate steps:
Filtering and processing
To begin with, the pipeline helps filter and process the information firehose coming in from the left:
- (Inbox curation) Priority Sources — newsletters, RSS feeds, social accounts — are chosen by the Editor because they provide particularly valuable content. Inbox curation is thus the frontline of filtering the information firehose, ensuring that the only content you even consider looking at is produced by the very best sources.
- (auto-aggregation) Each piece of content published by a Priority Source appears in the Editor’s Inbox. Content will not build up as by default each item is flushed if it has not been moved “Queued” (next) in time.
- (Scan & Queue) The Editor Scans the Inbox to select the best content, which is “Queued” — ie, moved to the Reading Queue for later, in-depth reading. The Scanning process is thus the second phase of filtering and is pure GTD (Getting Things Done). Again, content is by default auto-flushed if it hasn’t been moved out of the Reading Queue (next) in time.
- (Read) The Editor turns to the Reading Queue to read the “best of the best” content — i.e, content s/he previously identified as the best content recently published by his/her Priority Sources.
- (Annotate and Store) If, during or after reading, the Editor sees that the content is valuable, s/he starts annotating it, in the process Storing it in the Library. This is the first round of progressive summarisation.
Welcome to your Library
The above processes create a Library of what Zettelkasten aficionados may recognise as “bibliographic notes” — notes about other peoples’ content.
But the Library is also where Editors add their own notes, collaborate with others and create content for publication. This Library therefore contains:
- bibliographic notes on the high-quality content the Editor, following the above process, has curated from elsewhere;
- the Editor’s original notes, spanning from fleeting, atomic notes and ideas through to drafts and actually published posts;
- the Editor’s Overviews: also known as Zettelkasten Overviews, these represent a second round of progressive summarisation of the above notes. These are a form of Collection (see #Collections, below).
Everything in the Library is tagged using the same tag set, and processable by the ecosystem’s AI-supported writing tools.
Not shown: notes added to the Reading Queue or Library via the browser bookmarklet and other #Curation tools, below.
Published content, right
At the right end of this funnel, finally, are the Editor’s published articles, blog posts, etc, live on the public web.
While Editors can of course share wherever they wish, all content is also automatically shared to all Fediverse accounts (other Hubs, Mastodon, Friendica, etc. ) Following the Hub. Each Hub also:
- supports WebMention, and so pings every site it references;
- provides the Editor with a newsletter;
- and automatically creates an RSS feed for every collection of content (as is the case today).
These features are not shown in the above figure because we’ll explore it in Social knowledge graphs for collective intelligence.
Two content types: Notes and Collections
The entire Tool4Thought (and public Hub) is built from two content types.
Notes
Notes are the core building block of the Tool4Thought, as well as the principal content type of the public Hub. They:
- can have different levels of visibility (Private, Public and Friends)
- can be curated (ie pointing to a URL somewhere else, with the Editor’s notes and tags) or hosted (existing entirely within Tool4Thought and public Hub)
- come in three different Types: Like, Think and Do (if the public site is a Hub — this is easily customisable)
- are composed within the Tool4Thought from individual, taggable blocks (these tags are applied to the Note as a whole when made Public)
- can be Highlighted (a simple Boolean field), putting them in the Library’s “Favourites” collection and ensuring they appear (if published) in the Hub’s “Best of” resources.
- combine the best characteristics of blogposts and wiki pages.
Collections
Collections are a subset of your Notes created by a tag-based search. They come in two basic varieties:
- tag menu: the simplest collection, this is created automatically for each tag as soon as that tag is created in the Tool4Thought, and collects every note block with the tag, as well as any Overviews (next) using the tag;
- Overview: Those familiar with Zettelkasten will recognise Overviews as Zettelkasten Indexes: a summary of a field of interest, linking to all relevant notes. They therefore combine a Collection with a Note about that Collection, and are created by adding a Body text summarising the Editor’s insights from the content collected by a tag menu. The Editor can also use AND, OR & NOT operators to refine the Collected content.
The simplest way of creating an Overview is therefore to take a tag menu and click the ‘Edit’ button, providing another layer of progressive summarisation as the Editor summarises multiple ideas and resources. Collections are always Hosted and have the same different visibility levels as Notes. Obviously, Public Overviews do not include links to Private notes.
Key writing features
The following features are essential to making the most of the content pipeline.
Writing modes
Each writing mode combines selected stages of the content pipeline with specific tools to help push content through the content pipeline.
While this section sets out a few Writing Modes available off the shelf, they are templates which can be both customised and created from scratch by users to support their own personal productivity processes.
Scan mode: from InBox to Reading Queue
**Scan Mode supports a very specific productivity technique: Scanning. This is an important part of Getting Things Done (#GTD): you are not reading, you are shortlisting the most readworthy content into your Reading Queue.
Scan mode automatically calls up both Inbox and Reading Queue and closes everything else. The Editor “Queues” interesting content by right-clicking or dragging/dropping something in the Inbox into the Queue.
Read/Annotate mode: from Queue to Library
Rather than doomscrolling social media and reading whatever comes up, a more productive reading habit is to Scan (above) and then — either straight afterwards or some other time — read the contents of your Queue: the best content you’ve already selected from your most important sources.
read only the best content selected from your most important sources
While you can one-click any Resource into the Library, leaving the AI to autotag it, you will learn more if you annotate it. Read/Annotate mode supports Editors as they read their Queued content and summarise in their own words how it relates to their lives, work and other Library content.
Zen Write mode
Zen mode is for when you need to write up an idea or post without distraction. It opens a New Note and closes everything else.
Research and Write mode
When the Editor wishes to write something based on existing notes in the Library, this mode opens a new note and an advanced search panel. Using the latter, the Editor can search their Library, pin the most relevant Library content onto a pinboard (see below), preview or fully open and edit each note, and move selected content from existing notes into the new one, with back-references automatically added.
AI-supported writing modes
When an Editor wants their Hub to help write up an idea, Autoprompt Write mode replaces the Research and Write advanced search panel with dynamic search results: as the Editor writes, the AI surfaces relevant notes and overviews, which the Editor can pin, copy from, etc. However this is unlikely to be as popular as your own personal AI chatbot:
"an AI with ChatGPT’s language abilities which provided quotes and citations to back up its response ... and favoured, in its quotes and citations, your own notes in your Library, followed by the content shared with you by your trusted Friends, then content published by your Priority Sources, and then content published by sources you’ve annotated to your Library frequently" - How Artificial Intelligence will finance Collective Intelligence
Solving the “Zoom problem” with multi-level visualisations
**All such tools need to solve the “Zoom problem”, as the Editor needs to easily switch between:
- ‘ground level’: when the Editor is reading and/or editing one or a few notes or overviews, each taking up significant screen real estate
- ‘orbital view’: where the Editor can see many notes and overviews, albeit with almost no detail
- Intermediate levels such as ‘card view’ (similar to the card-based interface of a public Hub), ‘helicopter view’ and/or ‘airplane view’.
Notes and Overviews therefore need to be presented at different levels of detail using knowledge visualisation techniques, complemented by tools such as ‘note preview’ (mouseovering a link pops up a preview).
Publish & Share mode
Any note in the Library can be published by setting its visibility to Public, whereupon the Editor can share it wherever they wish. The dedicated Publish & Share mode, however, streamlines the procedure.
When the Editor deems that a draft is ready for publish, this mode sets its visibility to Public and opens a Share panel, which supports the creation of social media posts, the inclusion of the published article in the next newsletter edition, etc.
Curation tools
Bookmarklet and social media extensions
These allow one-click additions to either Inbox or Reading Queue. The browser bookmarklet, moreover, also allows the Editor to annotate, tag and store an external web resource directly into the Library (see MyHub FAQ).
Pinboards
Pinboards, on the other hand, allow the Editor to temporarily store and arrange notes and overviews within their Library: while searching for useful content to write a new Note, for example, the Editor can ‘pin’ potentially useful notes to a pinboard. There is always a default pinboard available to allow “one-click pinning”, while it is easy to create a dedicated pinboard for each writing project.
Notes are by default arranged on the pinboard in the order they were pinned. While other sort options exist (date of note creation, alphabetical by title, etc.), the most useful feature of a pinboard is to allow the Editor to manually rearrange the notes to create a sort of rough narrative for their writing project before they start writing.
Conclusion
A lot of the above wishlist can be assembled from various Tools4Thought today, particularly when connected together: spend any time in this space and someone is sure to tell you that all you need to do is chain together Hypothesis and Obsidian, Obsidian and GitHub, GitHub and Jekyll and so on, and if you need help there are scores of Discord servers available.
I’ve done similar things myself: I built my first Hub in 2013 from pocket, delicious, IFTTT and Tumblr, for example, and am currently helping build the site you're currently reading with massive.wiki, which combines any Markdown editor, GIT and a dedicated sitebuilder (see About this project).
As stated earlier, however, widespread adoption is crucial, so a Tool4Thought which does the above off the shelf is in my view essential. However it should have the “usual suspect” features seen in most Tools4Thought (bidirectional links, simple and faceted search with Related Tags, etc.), and be completely open source, allowing others to build additional features for power users.
Dig deeper
This post is part of a series, summarised in A Minimum Viable Ecosystem for collective intelligence. The other posts:
- Social knowledge graphs for collective intelligence
- How Artificial Intelligence will finance Collective Intelligence.
And if you're reading this here, chances are you already also like our first pilot project, described in About this project and the announcement post on Medium.
Comments welcome on all of them (How to comment), along with any contributions to the map (How to contribute).
Revision Notes
- Version 1 finalised on 1/1/2023 and reposted to Medium as a permanent version (see Wiki practice meets blogging).