Memory

What is in it

Memory holds short, self-contained pieces rather than documents or transcripts. Each one carries what kind of thing it is, because the kinds are not equally trustworthy and should not be read as though they were.

• Fact. Something that is true and expected to stay true.
• Event. Something that happened, with its date.
• Plan. An intention of yours, with a status and a next step. A plan is not a thing you have done.
• Idea. Unvalidated. An idea repeated back to you as though it were a fact is a fault.
• Note. What a source says, credited to that source.

Every piece also carries where it came from, whether that is you, an import, or a named assistant, and the date it was learned.

How things get in

Nothing is typed straight into Memory. Pieces arrive from your conversations, from documents you give Gubbins, and from what your connected assistants propose, and every one of those waits in Review until you accept it.

That is the whole shape of the room: it is a record you keep rather than a notebook you write in, and the only thing you do to a piece already in it is correct it or remove it.

What you can do

Find something

Search reads across everything. The filters narrow by kind, and the scope list narrows to one area. The List, Map, Areas and Changes buttons switch how the same memories are drawn, and the search and filters narrow all four.

Correct anything

Open a memory to see its source, its date, and everything connected to it. You can edit the wording or change the kind.

Remove means remove

Removing a memory deletes it. Gubbins forgets it, it does not go to an archive, and it cannot be brought back. The confirmation says so before you do it.

That is deliberate, and it is your call rather than ours: somebody removing a memory has decided they do not want it recalled and do not want it stored, and a product that quietly kept a copy would be answering a different question from the one they asked.

It has one consequence worth knowing. The change log is worked out from the memories themselves, so a removed memory takes its own history with it. A forgotten thing has no past either.

The archive, which is the other thing

Retiring a memory is not removing it. A retired memory leaves the active picture and stops travelling to any assistant, and it stays in your archive where you can read it and restore it if it turns out to still be true. Where one memory replaced another, the archive says which.

The tidy panel nominates memories that look stale or doubled, and retiring is what it offers. Nothing there changes anything unless you say so.

The change log

The Changes view is a log of how your Memory got to be the way it is: what was learned, what replaced what, what was retired and what came back, newest first. It is one of the three ways of looking at the room, beside List and Map, so it is in the same place whatever you were doing.

It is not a separate store. Every line is worked out from the memories themselves, from the dates they already carry, which is why a memory you remove takes its lines with it.

It also answers a second question: what Memory held on a particular day. Pick a date and the room shows what was there at the end of it, and how much has arrived and gone since.

The room's search and filters narrow it, so you can ask what changed about one kind of memory or one area. It is the one view that includes your archive, because a log of what was retired would be empty without it.

Green and amber

Every memory is marked green or amber. Green travels to an assistant whenever the key allows the scope it sits in. Amber does not travel unless the key explicitly includes amber, which is a separate decision made when the key is granted.

The mark is a judgement about sensitivity rather than about accuracy. You can change it on any memory at any time, and the change takes effect on the next read.

It is a boundary and not a grade. Green and amber say nothing about how much a memory matters to you, which is a separate thing you set separately, below.

How much a memory matters

You can pin any memory. A pinned memory is offered to your assistants sooner, and it stops fading with age the way an unpinned one does. It is also left alone by the tidy panel, which will never suggest retiring something you have said matters.

It is a pin rather than a slider or a score, and that is deliberate. Nobody can feel how much something matters to two decimal places, and a number on every memory would be read as a mark out of ten by somebody.

Pinning is not a claim that something is true. It says this matters to you, which is a thing only you can know about your own life. Whether a memory is accurate is a different question, answered by where it came from and by you correcting it, and no button here grades it.

The same is true from the other side. Gubbins works out how likely each memory is to be wanted right now, from how recently and how often your assistants have asked for it and from what it is connected to. That is a prediction about what is useful, never a judgement about what is correct: a memory can be reached for constantly and still be wrong, which is exactly why everything in here stays yours to correct.

Areas and keys

Areas are the named parts of your Memory that a key can be narrowed to. A new bank starts with a handful of them covering the ordinary areas of a working life, and you can rename those or add your own. A key granted to one scope reads that scope and nothing else.

Areas exist for sharing and for keeping things together. They are not folders, and how your memories relate to one another is drawn by their connections instead.

The words

A memory is one thing you know: a dot. An arrival is a proposed memory not yet accepted: a hollow dot. An area is a place you keep and name, the only level you own; an area may hold areas, and a key granted on an area reaches the areas inside it. A cluster is a grouping the map works out from meaning, never stored: a theme of up to 36 memories, or a subject of up to 9, where the dots are named. A connection is a stored link between two memories, one of seven kinds; it is a line, and never moves a memory.

The map

The Map view draws your Memory as circles inside circles. Far out, each of your areas is a disc with its name above it. Zoom into one and its subjects appear, each a smaller disc named from the words its memories share. Zoom into a subject and every memory is named beside its dot, with its connections drawn. Tap a name to zoom to it, pinch or scroll to zoom freely, and use the trail above the picture to come back up. You can drag a memory, a subject or a whole group into another group to file it there: the group you are over lights up and the words say what the drop will do. Dragging a group onto empty space just moves it for a look, and the next redraw puts it back. Every named memory carries a small plus on its edge: drag it to another memory to draw a connection, and a sheet asks which kind. Drop an area onto another area to put it inside; drop it on empty space to take it out again; "New area" makes an empty area to drag things into. Drop a memory into an area's open space and it starts a cluster of its own; drop others into it to connect them and make the cluster last. Arrivals can be dropped anywhere a memory can: they are refiled, and any connection you draw on one is kept and made real when you accept it. Drop a memory into a subject and it joins it for this visit, drawing "relates to" connections to its nearest memories there; name the cluster to make it an area if the join should last. Double-click any name to open its overview in the middle of the screen: what it is, its description, what sits inside it and its memories. An area can be renamed and described there; a cluster cannot be renamed, but giving it a name makes it an area of your own, inside the area it sits in. In the Meaning grouping, dropping into a group or renaming it makes that group one of your areas.

A bigger dot is a memory your assistants have been handed more often; a brighter one was handed out recently. Grey is settled, amber is going stale, and a count like "+12 waiting" on an area is arrivals waiting for you in Review. Lines are real connections, and the accent-coloured ones are proposed connections waiting for you: when a memory gets its fingerprint, the memories closest to it in meaning are suggested as "relates to" connections, which you accept or decline. Nothing about the picture is stored: it is worked out from your memories each time, so the same bank always draws the same map.

The chips under the map choose which kinds of connection are drawn, and "None" puts every line away. When there are many lines they draw faint so the memories stay readable; the lines to and from anything you have just moved draw at full strength for a moment.

Drag a memory or a cluster out of its area and let go on the mat, and it comes out: into the area around it when there is one. At the top of the map everything lives in an area, so a cluster let go there becomes an area of its own, named as the cluster was, and a single memory goes to Not sorted yet. An area let go on the mat comes out of the area it was inside. The words at the pointer say so before you let go.

Areas

The Areas view draws each named part of your Memory as its own frame, with every memory a part hanging in the order it was learned: hollow while it waits for you, amber as it goes stale, still attached to the frame the week it arrives. What has not been sorted yet is the loose-parts bag at the bottom. Tap a frame's name to open the area: one sentence you can copy that says what it holds, which keys read it and what tools ride with it, then doors to grant a key to it and to add a tool.

An area is what a key is narrowed to and what a tool is attached to, so "give somebody a key to the Kilter area, with the Jira connector" is one drawer. New area makes one; nothing moves into it until you say. Under the bag, Gubbins may point out loose memories that look like they belong somewhere, or a name the pile keeps repeating; every suggestion is a button you press or dismiss, and Stop suggesting turns the card off on this device.

Where it is

Memory is the first room in the app, and the one Gubbins opens on. Open Gubbins.