The app, Knowledge Organisation
How Mynd organises what it keeps
Most apps store information. Few really organise it. Mynd was designed on a discipline almost a hundred years old, Knowledge Organisation, the same one behind great libraries and professional archives, and brought it to your phone.
The problem with folders
Folders force you to choose one place for each thing. But real things belong to several places at once. An invoice belongs to a client, to a year and to what is still unpaid. A folder forces you to choose one of these paths and hides the others. Later, whoever searches by the path not chosen does not find it.
Mynd solves this with a hybrid classification: a taxonomy to say what each record is about, and facets to say everything else. Four layers that cross each other.
1. The taxonomy: what it is about
A taxonomy is a tree from the general to the particular, like the classification of living beings or the classes of a library. In Mynd it is called the Core and it has three levels.
- Name, the root: a project, a client or an area of life.
- Type, the kind of thing inside that Name: cases, meetings, documents.
- Reference, the concrete piece: that case, that meeting.
Client Silva/Case/Inheritance
The meaning of each level is free. You can build it straight, with the Name as the person or the project, the Type as the category and the Reference as the piece; or the other way round, with the Name as the category and the Type as the person. A Reference always belongs to the Name where it was created.
So you do not start from zero, Mynd brings a library of 162 ready-made structures, 147 by profession (41 of them in digital technology and AI) and 15 for life roles, already in the order of importance designed for each. You apply a whole structure at once and adapt it later. See Professions.
2. Fixed facets: how you judge
Each record can get five judgement marks, each with its own scale: sentiment (from Hate to Love), priority (from Low to Critical), confidence, urgency and status (Idea, Open, Doing, Done, Dropped). And a timing mark: now, today, week, soon or later. None forces you to decide on the spot. Over time they form an honest portrait of how you think and decide, and let you find, for example, everything that is urgent and not yet closed.
3. Free facets: the Context
A facet is one of the faces of a thing. In faceted classification, created in 1933 by the Indian mathematician and librarian S. R. Ranganathan in his Colon Classification, each record is described by several independent characteristics at the same time, instead of sitting in one place. Today it is everywhere: the filters of an online shop, brand, size, colour, price, are facets.
In Mynd, the Context is the layer of facets each person creates. Each category is a question, and its phrases are the possible answers.
- What kind of thing is it? A decision, An idea
- What state is it in? Open, Decided
- Who is it for? For a client, For third parties
- What does it ask for? Answer, Read later
The vocabulary is all yours. A record can have several phrases, and the phrases cross the whole tree without repeating it.
4. Automatic facets
Some characteristics Mynd records on its own: the date and time of each record, and its type, text, photo, voice, video, document or file. They take no work, and enter search like the others.
The crossing: funnel search
The power of a system like this is in the crossing. With one layer, you find everything on a subject. With two, you find the crossing. With three, you find exactly what you are looking for. Each filter narrows what the previous one left.
- All records2,500
- Name: Client Silva140
- Context: A decision23
- Status: open6
- Last three months2
Example search, illustrative numbers.
No folder structure can answer this question, because it would have had to foresee it. Text search is complete too: whole word, start or end of word, exact phrase, nearby words, excluding words, title only, and no need for accents. The text in photos of documents is read on the phone itself and enters the search.
The order of tags: manual or TagIntell
With hundreds of tags, an alphabetical list makes you search every time. So Mynd decides in which order tags appear, and the person chooses how. Everything is adjusted on the Organize screen.
- Two places. The Core panel, where you put tags on a card, shows first what weighs most in your structure. The Name, Type and Ref buttons on the home screen show first what you have been searching for. Each can be manual or automatic.
- Manual. You drag Names, Types and References up and down, and that order stays. It is never lost: automatic does not erase it, and it breaks ties.
- Automatic, TagIntell. It orders by real use. It uses no AI: it is calculation, always the same for the same data, run only on the phone, and on by default.
- What it counts. What you have already classified, with recent uses worth more (a use from today is worth double, and that extra halves after 14 days, a half-life as in physics); and what you usually choose, the clicks, each losing half its value every 30 days, reaching full strength only after about 50.
- Different mixes. By default the Core weighs 60% classification and 40% clicks; search weighs 40% and 60%. When filing, your structure counts more; when searching, your habit counts more.
- Bold Blue. Writes in bold blue the 30% of Names you used most in the last 30 days. It only highlights; it never changes the order.
- The pilot in command. Whoever wants adjusts weights and periods, or switches to manual. The Proof button shows side by side today's order and the order with clicks. The order never changes while you are touching it, so the list does not run away under your finger.
Where these ideas come from
None of this was invented for Mynd. They are ideas with a history, brought together in one place and on your phone.
- The taxonomy, the tree from the general to the particular, comes from Linnaeus' classification of living beings in the 18th century and from library classes such as Dewey's Decimal Classification of 1876. In Mynd it is the Core.
- Facets come from S. R. Ranganathan, who published the Colon Classification in 1933, the first faceted classification. In Mynd they are the ratings and the Context.
- Free tags, created by each person in their own words, were named in 2004, when Thomas Vander Wal called them a folksonomy. In Mynd they are the Context phrases.
- Half-life comes from physics: the time something takes to lose half its value. In Mynd it is what makes old use weigh less than today's.
- Trails between records come from Vannevar Bush's Memex, in 1945, and Niklas Luhmann's Zettelkasten, the box of numbered cards linked to each other. In Mynd they are the Trails, #3598 and [[subjects]].
- Relevance ordering is the family of techniques of search engines. In Mynd it applies only to your data, inside the phone.
Why it matters
Organising this way is not just a search technique. Each tag and each phrase is a small decision about what something means. Over time, the tree becomes the map of your world, and the facets become your judgement about it.
That structured portrait, kept on your phone, is the context that, later, an AI will be able to read to understand you, instead of guessing. In today's version, none of this goes through any AI: it is all organisation and calculation, done on the device itself.
In short: a taxonomy for the subject, fixed facets for judgement, free facets for everything else, automatic facets for date and type. Search crosses them all, and the order puts first what matters.
