Building a Complete Knowledge Base From Zero

Almost nobody starts a new knowledge system voluntarily. The decision is nearly always made under some form of duress: a product has been discontinued, a drive has failed, a provider has changed its terms in a message somebody skimmed, or the accumulated sediment of six years of files has reached the point where locating something takes longer than producing it again. By the time a person sits down to begin, they have usually already lost something, and the loss is what supplied the motivation.

That makes the fresh start a peculiar mixture of opportunity and dread. The opportunity is genuine: an empty structure is the only moment at which the expensive decisions can be made cheaply, since anything settled before there is material costs nothing and anything settled afterward requires manual work across everything already created. The dread is also genuine, and it has a specific character. Somebody standing in front of an empty system knows they are about to make choices whose consequences will not be visible for months, using information they do not yet have, about material that does not yet exist.

The common failure at this moment runs in two directions and both are recoverable. Some people design elaborately in advance, producing a beautiful taxonomy that reflects how they imagine their work is shaped rather than how it actually behaves, and then spend more time maintaining the arrangement than using it. Others create nothing at all, on the reasonable grounds that they do not yet know what they need, and accumulate several hundred items in an undifferentiated pile before discovering that retrofitting structure is considerably harder than establishing it.

What the first hour should actually produce is a small number of decisions that are expensive to change later, and nothing else. What counts as an item. What the top-level divisions are, using the categories the person already uses when speaking about their own work. Which category of material, if any, requires protection. Where the files physically live. Those four are worth deliberating over. Everything else, including how deep the hierarchy eventually goes and which capabilities turn out to matter, should be left until the material has said something about what it needs.

There is a second and less obvious property of a good first hour, which is that it should produce something usable immediately. A setup session that ends with an empty and admirable structure has produced an intention. A session that ends with three real entries, a genuine deadline recorded, and a document attached where it belongs has produced a working system, and the difference in whether it survives the following fortnight is considerable. People persist with things that have already been useful once.

The most reliable way to arrive at these conclusions, oddly, is to teach somebody else. Explaining a system to a person who has never seen it forces every implicit decision into the open. The teacher discovers which of their habits are principles and which are accidents they have been defending for years, and frequently learns more during the session than the student does. That dynamic is what makes a setup session worth describing: not as instruction, but as the moment when somebody who has been using a thing for a year finds out what they actually believe about it.

What follows is one such Saturday morning, in an office in Atlanta, between a minister who has been doing this for about a year and the youth pastor he has been meaning to help since a conversation some weeks earlier.

An empty folder and a man being careful not to lecture

The church office on a Saturday morning was the quietest it ever got, which David Achebe had said several times over the years and which remained true. The building had a smell it only had at that hour, of floor polish and cold radiators and the coffee somebody had made at half past seven.

Reverend Michelle Torres arrived at nine with a laptop, a notebook she did not open once, and the expression of somebody who has been told this will take an hour and does not believe it.

‘Before we start,’ she said, ‘I want to say two things.’

‘Say them.’

‘The first is that I have been putting this off for six weeks and it is not because I did not want to.’ She sat down. ‘It is because every time somebody has set up a system for me it has taken a day and I have used it for a fortnight.’

David nodded, entirely unoffended.

‘That has been my experience too, and it happened to me twice before this,’ he said. ‘What is the second thing?’

‘I am not bad with technology and I would like that established, because people my age get assumed either way.’ She was matter of fact. ‘I can use most things. What I cannot do is maintain something elaborate that somebody else designed, and I have stopped pretending otherwise.’

‘Then we are going to get along, because that is the entire lesson and you have arrived already knowing it.’

He turned the laptop toward her rather than taking it, which she noticed.

‘I am not going to touch this,’ he said. ‘You are. I will talk and you will do it, because otherwise you will leave here with a system you watched somebody build.’

‘James told you that.’

‘James told me that after watching me try to help somebody with a projector, and he was right and I have not forgotten it.’

They started with the file, which was the part David enjoyed explaining most because it was the part that had persuaded him.

‘Open it,’ he said. ‘It is one file. That is the whole application.’

Michelle opened it, and a first-run guide appeared explaining where her material would be kept.

‘It is asking me to choose a folder.’

‘It is asking you to choose a folder, and that folder is where everything you ever write is going to live.’ David leaned back. ‘Not on a server. Not in an account. In a folder on that machine, which you choose, and which you can open in a file window while this is running.’

‘Where should I put it?’

‘Somewhere you will find it in three years and somewhere your backup already covers.’ He shrugged. ‘Mine is in my documents folder with an obvious name, which is unimaginative and correct.’

She made one, named it plainly, and granted the access.

‘Now open it in a window,’ David said. ‘Not because you need to. Because I want you to see it.’

She did. The folder held a small number of files: the licence anchor, a repository file, and a directory for attachments.

‘That is it?’

‘That is it, and it is going to fill up as you work, and everything in it is going to be readable.’ He was definite. ‘The repository holds your entries and your structure. The attachments directory holds the actual documents and photographs you attach. And that is the whole arrangement.’

‘Why are you showing me this in the first five minutes?’

‘Because it is the question everybody asks in month three and worries about in month one,’ David said. ‘Where is my material. If you have seen the folder, you never have to wonder again, and I would rather you wondered now while I am sitting here than at eleven at night in six weeks.’

Thinking before creating anything

‘Now the part that matters,’ David said, ‘and we are going to do it on paper.’

Michelle laughed.

‘You have brought me here to set up software and you are producing a legal pad.’

‘I have brought you here to make four decisions and the software is how we record them.’ He was unembarrassed. ‘If we start clicking, you will build a structure that reflects what the screen suggests rather than what your work is. Twenty minutes on paper first.’

‘What is the question?’

‘What are the things you talk about when somebody asks what you do?’ He wrote nothing yet. ‘Not categories. The words you actually use to your husband on a Thursday when he asks how the week went.’

Michelle thought about it, which David let her do without filling the silence.

‘Youth group,’ she said. ‘Which is the weekly programme and everything around it. Camp, which is three separate things and which eats about a third of my year. Parents, which is communication and which is where all my anxiety lives. Mentorship, which is the newest thing and the one I care about most. Meetings, which is the staff and elder side of it. And there is something about my own development that I have never had anywhere to put.’

David wrote six words.

‘That is the structure,’ he said.

‘That is not a structure. That is a list of things I said.’

‘It is a list of things you said in your own vocabulary, without having to think about how to categorise anything, which means you will recognise it at seven in the morning in eleven months.’ He tapped the pad. ‘Every structure I have seen fail was somebody else’s vocabulary, including the beautiful one my own daughter built for a bakery in Toronto.’

Michelle looked at him.

‘That is a very specific example.’

‘It is somebody else’s story and I will tell you at lunch.’

They created the six pages together, Michelle doing the clicking, and David making her name them as she had said them rather than tidying them up.

‘Camp is going to have things underneath it,’ she said, almost immediately.

‘Say what.’

‘Summer camp, which is the enormous one. Winter retreat, which is smaller and colder and which I enjoy more. And weekend events, which are four or five a year and are their own kind of thing.’

‘Then make them.’

She created three sub-pages beneath camp, and the tree indented, and she made a small noise of satisfaction that she was slightly embarrassed by.

‘Now the question that will save you a year,’ David said. ‘When do you make a page and when do you not?’

‘I was going to ask.’

‘Pages are for things that will grow. Labels are for properties that cut across things.’ He said it slowly, because it had taken him a long time to arrive at. ‘Summer camp is going to accumulate. It will have twenty entries by August, and next year another twenty, and in five years it will be the largest thing you own. That is a page.’

‘And urgent is not.’

‘Urgent is not, because urgent is true of a budget document and a parent conversation and a broken minibus, and those things live in three different places.’ He shrugged. ‘If you make urgent a page, you have to choose between filing something where it belongs and filing it where it is urgent, and you will get it wrong at the worst possible moment.’

‘So the page says where and the label says what kind.’

‘The page says where and the label says what kind, and once you have that you will never make a serious structural mistake.’

They tidied the tree slightly, which produced the demonstration David had been waiting for.

‘Meetings is in the wrong place,’ Michelle said. ‘It should be at the bottom, because I look at it least.’

‘Then move it.’

She dragged it, and it moved, and the tree reflowed around it.

‘That is all?’

‘That is all, and it is the reason none of this has to be right today.’ He was emphatic. ‘Anything you can fix by dragging is a decision you should not agonise over. Anything you would have to fix by editing three hundred entries is a decision worth twenty minutes now, which is why we spent the twenty minutes on the page names and are spending four seconds on the order.’

Michelle right-clicked something and discovered the context menu.

‘I can rename them.’

‘Rename, move, delete, all from there.’ David nodded. ‘Which you will use in about six weeks when you discover that parents should have been called families, or something like that, and it will take you eleven seconds.’

The icons and colours came last, and Michelle was dismissive of them for approximately ninety seconds.

‘This feels like decoration.’

‘It is decoration and it is also the fastest thing in the sidebar,’ David said. ‘I do not read my page names at half past six in the morning. I recognise the colour and go there. That is about a second faster than reading, forty times a day.’

‘Which is not nothing.’

‘Which is not nothing, and I would not have believed it either.’ He watched her assign them. ‘Put the same colour on all three camp pages. You want the branch to be one thing.’

‘Why?’

‘Because when you are in camp season you are only in camp season, and a branch that reads as one colour is a branch you can find without focusing your eyes.’

One real thing, made properly

‘Now we build something real,’ David said. ‘Not an example. An actual thing you have to do, because a system with sample data in it is a system nobody trusts.’

‘The lock-in is in three weeks.’

‘Then the lock-in is our first entry.’

She created it under youth group, and named it plainly, and then sat looking at an empty body.

‘Now what?’

‘Now I show you the thing that took me four months to find and which is the most important part of this whole morning.’ David leaned in. ‘An entry is not one document. It has sections inside it, and each one has its own body and can hold its own files.’

‘Why does that matter?’

‘Because your lock-in is not one kind of information.’ He counted on his fingers. ‘There is what it is and when. There is the practical arrangement, the room, the food, the transport. There is who is helping and what they agreed to do. There is what it costs. And there is what happened afterward, which does not exist yet and which is the most valuable thing you will ever write about it.’

‘Five sections.’

‘Five sections, and I would make them now while you are calm, because in three weeks you will be adding a phone number at half past ten at night and you will not be designing anything.’

She made them: event details, logistics, volunteers, budget, and afterward.

‘The last one is empty.’

‘The last one is empty and it should exist anyway,’ David said, ‘because an empty section is a question. In three weeks it will say, plainly, that you have not written down what went wrong, and you will write it, and next year you will read it before you plan the next one.’

Michelle wrote the event details, and David watched her use the editor without saying anything until she had finished.

‘You used a heading and a list.’

‘Was that wrong?’

‘It was exactly right and I want you to notice why you did it.’ He was pleased. ‘You did it because the information had that shape. Times are a list. The description is a paragraph. Nobody told you a rule and you did not need one.’

‘I bolded the arrival time.’

‘Because it is the thing somebody will look for in a hurry, and if you keep bolding only that kind of thing your entries will be scannable in a year.’ He nodded. ‘One meaning per decision. That is the whole of formatting.’

She attached the budget spreadsheet next, and hesitated with the cursor.

‘Where does it go?’

‘Where does it belong?’

‘Budget.’

‘Then attach it to the budget section rather than to the entry.’ David watched her do it. ‘Which sounds like a detail and is the thing that will make this readable in two years, because an entry with nine files at the top is a list of names somebody chose in a hurry, and an entry with the spreadsheet in the budget section identifies itself.’

She attached the flyer to event details, and a small indicator appeared on the collapsed bar showing that it carried a file.

‘That is useful,’ she said.

‘That is the thing I look at first.’ He pointed. ‘I do not open sections to find out whether there is anything in them. The bar tells me.’

Michelle collapsed everything and looked at the entry as a whole, five bars with one line of title each.

‘It is tidy.’

‘It is tidy and it will still be tidy when there are fourteen things in it, which is the point.’ David sat back. ‘I have entries with six sections and thirty attachments and they are still one screen when collapsed.’

‘How do you decide what a section is?’

‘A section is a part of the thing that has its own paper,’ he said. ‘Which is a rule somebody in Denver gave me and which I have never improved on.’

The second axis, and a dashboard made of stars

‘Labels,’ David said. ‘And I am going to give you a number before you start, because everybody gets this wrong in the same direction.’

‘What number?’

‘Six. Possibly seven. Never twelve.’ He was firm about it. ‘Every label you add makes every other label slightly less meaningful, because the more categories there are the less carefully anybody assigns them, and that includes you.’

Michelle thought.

‘Urgent. Camp. Mentorship. Parents. Worship. Budget.’

‘That is six and they are the right six, and I will tell you why in a moment.’ He waited while she made them, assigning colours as she went. ‘Say what each one is for.’

‘Urgent is anything that has to happen this week. Camp is anything camp-related that is not inside a camp page, which is mostly parent letters and budget lines. Mentorship is the same. Parents is anything I have said to a parent or need to say. Worship is the music and service side. Budget is anything with money in it.’

‘And notice what none of them are.’

She looked at the list.

‘None of them are places.’

‘None of them are places, which means none of them duplicate the page tree,’ David said. ‘That is the discipline. A label that says the same thing as a page is a label that has cost you a click and told you nothing.’

He asked her to apply them to the lock-in entry, which took her four seconds and produced a small satisfaction he recognised.

‘Now the star.’

‘I saw that. What is it?’

‘Star that entry and look at the sidebar.’ He waited. ‘Those are your favourites. And I want to tell you how to use it, because it is the feature everybody thinks is decoration for the first four months, including me.’

‘What is it actually?’

‘It is your front door.’ David was definite. ‘Not a bookmark list. Not a place to keep important things, because everything in here is important. It is the three to six things you are actually working on this fortnight, and you open one of them within four seconds of sitting down.’

‘And when the lock-in is over?’

‘You unstar it, and the list gets shorter, and that is the only moment of completion this kind of work ever gives you.’ He shrugged. ‘I am serious about that. Nobody tells a youth pastor that a thing is finished. You have to build the marker yourself.’

Michelle wrote something in the notebook she had said she would not open.

‘Three to six,’ she repeated.

‘Three to six, reviewed every couple of weeks, and if it has nine things in it you have stopped using it as a door and started using it as a list.’

The due date came next, and produced the first genuine surprise of the morning.

‘Set the date of the lock-in on the entry,’ David said.

She did.

‘Now look at the tabs at the top of the sidebar.’

Michelle clicked through them: recent, due, expiring.

‘Due has my lock-in in it.’

‘Due has everything in the whole vault with a date coming up, regardless of which page it lives under,’ David said. ‘Which is the thing the tree cannot do. Your tree is organised by subject. Your week is organised by time. Those are different shapes and they never overlap.’

‘So this is my week.’

‘That will be your week once there is anything in it, and there will be nine things in it by Christmas.’ He pointed at the third tab. ‘Expiring is different and you will need it for a specific reason.’

‘Which is?’

‘Safeguarding certificates. Volunteer clearances. First aid qualifications.’ He said it plainly. ‘Those do not become late. They become invalid, and a volunteer whose clearance has lapsed cannot be in a room with a child, and finding that out on a Friday afternoon is a conversation I have had and would like you never to have.’

Michelle had stopped writing.

‘I have eleven volunteers.’

‘Then you have eleven entries to make and eleven dates to record, and it will take you about twenty minutes, and I would do it before you do anything else this week.’

‘Why is that different from a due date?’

‘Because a due date needs a week of warning and an expiry needs six months, since renewing a clearance involves other people and a process.’ He was serious. ‘Different consequence, different horizon. That distinction has caught almost everybody I know at least once.’

Putting the year into it

‘Half past ten,’ David said, checking. ‘We have twenty minutes of setup left and then two things that are not setup at all.’

‘I thought this was an hour.’

‘It was always going to be ninety minutes and I said an hour because otherwise you would not have come.’

Michelle laughed and did not disagree.

The calendar came next, and David made her enter the recurring items first.

‘Youth group is every Wednesday.’

‘Then enter it once with a repeat rather than as forty separate evenings,’ he said. ‘Because you will enter twelve of them by hand and then stop, and then in March you will wonder why nothing is in your calendar.’

‘That is a very specific prediction.’

‘It is what I did, and what everybody does, and it is the single most common mistake with a calendar.’

She entered it, then the monthly parent meeting, then the three camp date blocks, which took longer because she had to look two of them up.

‘Now look at the sidebar,’ David said.

A strip near the bottom showed what was coming.

‘That is always there?’

‘Always there, and you do not have to open anything to see it, which is the whole reason it works.’ He was definite about this. ‘On a good day I open the calendar. On a bad day I do not open anything, and on a bad day that strip is the only thing standing between me and forgetting something.’

Michelle switched between the views and stopped on the week.

‘This is what I wanted,’ she said. ‘I have been drawing this on paper every Sunday night for two years.’

‘Then stop drawing it.’

‘I will keep drawing it for about a month and then stop,’ she said, honestly. ‘That is how these things go.’

‘That is exactly how these things go and I would rather you said that than promised otherwise.’

David showed her the full day view last, scrolling the timeline through all twenty-four hours.

‘Why would I need that? I do not work at three in the morning.’

‘You work at three in the morning twice a year,’ he said, ‘on the second night of camp, when something happens, and again on the lock-in you have just created, which is the entire point of a lock-in.’

Michelle stopped.

‘I had not thought about the fact that my events go through the night.’

‘Your events go through the night and any calendar that only shows you nine until five is a calendar that hides half of the two hardest days of your year.’ He scrolled it. ‘This is where you plan a camp day properly. Who is on at which hour. When the quiet time actually is. When the staff shift changes.’

‘I have been doing that on a whiteboard.’

‘Photograph the whiteboard and attach it to the entry, and then build it here next year.’ He shrugged. ‘I am not going to pretend the whiteboard is wrong. It is a good tool for a room full of people. It is a bad tool for remembering anything in eleven months.’

The part that is not about software

‘Now,’ David said, and the change in his voice was noticeable enough that Michelle put the pen down. ‘The mentorship programme.’

‘I was wondering when we would get to that.’

‘I have been leaving it deliberately, because everything up to now has been arrangement and this is not.’ He turned toward her properly. ‘Tell me what those conversations contain.’

Michelle took a moment.

‘A sixteen year old whose parents are separating and who has not told anyone else. A girl who is not eating and knows I have noticed. Someone who has told me something about a family member that I have had to think very carefully about.’ She said it evenly. ‘Nothing about my job is more serious than that.’

‘And do you write any of it down?’

‘Some. Not enough, and badly, and in a notebook I take home in a bag.’ She was honest. ‘Which I have known is wrong for about a year.’

‘Why do you write it at all?’

‘Because they come back in three weeks and refer to something they said, and if I do not know what it was they will decide they were not important enough to remember.’ She looked up. ‘And that will be the end of it. Not just with me.’

David nodded slowly.

‘That is the correct answer and it is the reason I keep mine,’ he said. ‘The record exists for them. It is also the most dangerous thing in your bag, and both of those are true at once and neither cancels the other.’

He showed her the encryption then, and made her do it herself.

‘Create the entry. Then set a password on it.’

She did, and the entry locked, and the content became unreadable in the list.

‘What is actually happening?’ she asked. ‘I want to understand it well enough to explain it, not well enough to build it.’

‘Then here is the version I can give you honestly, because I am not a technical person either.’ David counted it out. ‘The contents are scrambled using a standard called AES-256-GCM, which is very well studied and which nobody breaks. Your password is not the key directly. It goes through a process that deliberately takes a fraction of a second, which is why there is a small pause when you open it.’

‘Why would a delay be good?’

‘Because somebody guessing passwords has to pay that same fraction of a second for every single guess.’ He said it as it had been said to him. ‘That is the whole security argument. The pause is the product.’

‘And if somebody takes my laptop?’

‘They have a file full of nonsense.’ He was definite. ‘They cannot read it and they cannot tell which of your entries resemble each other, which matters more than it sounds.’

Michelle asked the question David had been waiting for.

‘What happens if I forget the password?’

‘Nothing happens,’ he said. ‘That is the answer and I am going to give it to you plainly rather than softening it. There is no recovery. There is no reset. The password is the only key and nobody, including anybody who made this, can produce your material without it.’

She sat with that.

‘That is frightening.’

‘It is frightening and it is the reason the rest of it works,’ David said. ‘If somebody could recover it for you, they could recover it for anybody, and then the door is not a door. What you are being given is a genuine guarantee, and the price of a genuine guarantee is that it applies to you as well.’

‘So the password discipline is everything.’

‘The password discipline is everything and it is the only part of this morning where I am going to be prescriptive.’ He held up fingers. ‘Generate it rather than inventing it, because anything you invent has a pattern in it. Store it in a password manager, not in your head and not on anything in this building. And write down, somewhere with your other important documents, how somebody would reach it if you were unavailable.’

‘That is a strange thing to plan for at twenty-nine.’

‘It is a strange thing to plan for and I am sixty-two and I did it last month,’ David said. ‘And I have obligations that outlive me, and so do you, and there are eleven young people in that programme whose records should not become unreachable because something happened to one person.’

He showed her the caching next, which she was relieved by.

‘You enter it once and it holds for the session,’ he said. ‘So a conversation on a Tuesday afternoon does not mean typing it four times.’

‘And when I walk away?’

‘Then you use this.’ He showed her the lock, which covered the whole screen in one action. ‘My office door does not latch properly and people put their head round it. That control is faster than looking up, and after about a month your hand does it without you deciding.’

‘Do you use it for people you trust?’

‘I use it for everybody, and not because I distrust anybody.’ He shrugged. ‘A screen with words on it in a room with another person is a document that has been published. It is not about reading. It is about the fact that it could be read, which changes how I write for the rest of the afternoon.’

Michelle wrote that down.

‘One password for all of it, or one each?’

‘That is your judgment and I will tell you mine and my reason.’ David paused. ‘Mine is one password for everything pastoral. Not because it is simpler. Because tiers would mean deciding whose confidence deserves the stronger door, and I will not do that.’

‘I would not either.’

‘Then you have your answer, and you arrived at it in about four seconds, which is how you know it is a principle rather than a preference.’

The last thing, which is about not going stale

‘One more and it is the one I would have given you first if I were being honest about what I actually use,’ David said.

‘Go.’

He opened the reader, and Michelle looked at the folder list down the side.

‘This is feeds?’

‘This is a dozen writers and organisations I follow, arranged into folders, and they arrive here instead of my going round a dozen sites on a Saturday morning.’ He was cheerful about it. ‘Which I did for eleven months. By hand. Copying and pasting. While feeling extremely diligent about it.’

‘I do that.’

‘I know you do that, because I have watched you do it in this office, and I said nothing at the time because I had only just stopped.’

They set hers up together, which took about ten minutes and produced more discussion than David expected.

‘How should I organise them?’ Michelle asked. ‘By topic?’

‘Mine are by relationship rather than by topic and that was not my first arrangement,’ he said. ‘People I know personally. Institutions I am accountable to. The field generally. And one called elsewhere, which is entirely unrelated to work and exists to stop me becoming a bore.’

‘Why did relationship beat topic?’

‘Because when I open this I am not asking what is happening in theology. I am asking whether anybody I know has written something.’ He shrugged. ‘Which turns out to be the actual question, and it took me four months to notice that my folders were answering a different one.’

Michelle built hers: youth ministry practice, leadership and development, denomination and safeguarding, and one for the two writers she followed personally.

‘And what do I do when something is useful?’ she asked.

‘Do not save the article.’ David was definite about this. ‘Make an entry, put two lines about where it came from, and then write for as long as the thought runs, in your own words. Underneath that, a section listing what it connects to.’

‘Why not just keep the article?’

‘Because a saved article is somebody else’s thinking sitting in your folder, and in six months you will not remember why you kept it.’ He said it as somebody who had made the mistake. ‘Two paragraphs of your own reaction are worth more than the whole piece, and about a fifth of what I have written this year started that way.’

Michelle looked at the sidebar, which now had rather more in it than it had at nine o’clock.

‘This is a system,’ she said.

‘This is a system and you built all of it, and I want that noted because I have not touched your keyboard once.’

What each of them took from a Saturday morning

At about a quarter to eleven Michelle closed the laptop and did not immediately get up.

‘Can I say the thing I am actually thinking?’

‘Please.’

‘I have been waiting for the part where you tell me the rules.’ She turned to face him. ‘Every system anybody has ever set up for me came with a document. A convention. A thing I was supposed to do every Friday. And I have been braced for it all morning and it has not come.’

‘There are no rules and there are four decisions,’ David said. ‘We made all four before ten o’clock. The pages are named in your words. The labels describe kind rather than place. Your mentorship material is protected as a category rather than item by item. And your files are in a folder you have seen.’

‘And everything else?’

‘Everything else you will work out by getting it slightly wrong, which is faster and considerably more durable than being told.’ He stood, stiffly. ‘I am going to give you one instruction and it is not about the software.’

‘Go on.’

‘Do not ask me anything for a month unless you are genuinely stuck.’ He was entirely serious. ‘Every time I explain something to you, you will learn a fact and forget it. Every time you find something yourself, including by pressing something eleven times by accident, you will own it forever.’

‘Somebody did that to you.’

‘James did that to me, deliberately and slightly cruelly, and it worked.’

Michelle laughed and picked up her bag.

‘I want to say something to you as well,’ she said. ‘And it is going to sound like flattery and it is not.’

‘Attempt it.’

‘You explained all of that without once making me feel behind.’ She said it plainly. ‘Which is not a small thing, because I am the youngest person on this staff by fifteen years and I spend a substantial amount of my week managing an assumption that I do not know anything.’

David was quiet for a second.

‘I have spent this morning learning what I actually believe about all of this,’ he said. ‘Which I did not know until I had to say it out loud in order, to somebody who would notice if it did not hold together.’

‘So you got something out of it.’

‘I got a great deal out of it, and the specific thing is that I have been treating three of my habits as principles and one of them is not.’ He shook his head. ‘Which I would not have discovered in another year of using it alone.’

‘Which one?’

‘I will tell you when I have worked out what to do about it,’ David said. ‘Go and enter your eleven volunteers.’

Nineteen days later, one question

She lasted nineteen days, which David considered a considerably better result than he had managed himself.

She came into his office on a Thursday afternoon with the laptop under one arm and an expression he recognised from staff meetings, which was the expression of somebody who has decided to be direct and is slightly braced about it.

‘I am breaking the rule,’ she said. ‘I have one question and it is not a stuck question, it is a why question, and I have decided that those are allowed.’

‘Those were never disallowed. Sit down.’

‘The lock-in happened.’

‘I know. I heard about the fire alarm.’

‘Everybody heard about the fire alarm.’ She sat. ‘I want to tell you what happened afterward, because it is not what I expected and I think you will find it interesting.’

‘Go on.’

‘I wrote the afterward section at half past one in the morning, in the car park, while waiting for the last parent.’ She said it plainly. ‘On my phone, badly, about four hundred words. And I would not have written it at all if the section had not been sitting there empty and slightly accusing.’

David nodded slowly.

‘And?’

‘And I read it four days later and about a third of it is wrong.’ She laughed. ‘Not factually. Emotionally. I was exhausted and everything looked worse than it was, and there is a paragraph about a volunteer that is genuinely unfair.’

‘Did you delete it?’

‘I nearly did and then I did not, and that is my question.’ She leaned forward. ‘I added a second paragraph underneath dated four days later saying what I actually think now. And I want to know whether that is a reasonable thing to do or whether I am making a mess.’

‘That is not a mess,’ David said. ‘That is the single most useful habit anybody in this work can develop and I would like you to keep doing it.’

‘Say why, because I could not justify it.’

‘Because the version written at half past one is the accurate record of what it was like, and the version written on Monday is the accurate record of what it was.’ He was definite. ‘Those are different things and both are true and neither replaces the other. If you keep only the calm one you will conclude, in three years, that camp is easier than it is. If you keep only the exhausted one you will conclude that you cannot do this.’

‘And keeping both?’

‘Keeping both tells you something no single account can, which is the difference between them.’ He shrugged. ‘I have entries about hospital visits with three dated paragraphs in them written over a fortnight, and reading the sequence is more informative than any one of them.’

Michelle wrote that down.

‘Now my thing,’ she said. ‘Because I found something and I want to know whether I have understood it.’

‘Go.’

‘I clicked on something at the bottom of the sidebar by accident and it showed me an entry from three weeks ago that I had entirely forgotten writing.’ She was visibly still pleased about it. ‘A note from a parent conversation. And it refreshes on its own.’

‘Ah.’

‘I have been opening it first thing every morning since.’ She paused. ‘Which I recognise is exactly what you would have told me to do if you had told me anything, and I want to point out that I found it by pressing something I did not understand.’

‘Which is the entire argument I made nineteen days ago,’ David said, ‘and I would like the record to show that I did not say anything just now.’

‘You are saying it with your face.’

‘I am a minister. My face is the only instrument I have.’

She asked the other question then, which was the one she had actually come for.

‘You said there was one of your habits that turned out not to be a principle. Have you worked out what to do about it?’

David was quiet for a moment.

‘My reader folders,’ he said. ‘I organise them by relationship, which I told you was the right answer and which I still think is right for me. And when I said it out loud to you I heard myself and realised that one of those folders is not about relationship at all. It is a folder of people I feel I ought to be reading, which is a completely different thing, and I have not opened it in four months.’

‘What are you going to do?’

‘I have deleted it,’ he said. ‘Which took about six seconds and which I had been not doing for four months because deleting it felt like an admission.’

‘An admission of what?’

‘That I am sixty-two and I am not going to read everything, and that pretending otherwise costs me a small amount of guilt every time I open that panel.’ He shrugged. ‘Which is a considerable amount of guilt to be paying for a folder.’

Michelle laughed properly.

‘So teaching me cost you a folder.’

‘Teaching you cost me a folder and saved me four months of the next one,’ David said, ‘which is a good trade and which is why I would recommend that anybody a year into something like this go and explain it to somebody who has never seen it.’

The first hour is the whole thing

What that morning demonstrates is not a feature tour. It is that a functioning professional knowledge system can be established in ninety minutes, and that almost all of the value comes from a very small number of decisions made before anything is created.

Four decisions carried the whole session. Where the material physically lives, settled by choosing a folder and then opening it in a file window so the question never recurs. What the top-level divisions are, taken from the words somebody already uses when describing their own work rather than from any attempt at categorisation. What counts as a page against what belongs in a label, resolved by the rule that pages are for things that will grow and labels are for properties cutting across them. And which category of material requires protection, decided once as a category rather than item by item. Everything else in the session, including the depth of the hierarchy, the colours, the ordering, and which capabilities turned out to matter, was either reversible by dragging or discoverable later.

The second observation is that the session produced something immediately usable rather than an admirable empty structure. A real event three weeks away, with five sections holding the kinds of information that event genuinely generates, a spreadsheet attached to the part it concerns, a date recorded, a star applied, and a deliberately empty section waiting to be filled in afterward. That distinction determines whether a system survives its first fortnight, since people persist with things that have already been useful once and abandon things that have only been arranged.

The third is that the encryption conversation was not a technical one. It began with what the material actually contains, proceeded through why the record exists at all, and arrived at a straightforward explanation of the mechanism, a clear statement that there is no password recovery, and a set of practical obligations following from that. Nothing about it required either party to understand cryptography, and the absence of a recovery route was presented as what it is: the price of a guarantee that applies to everybody equally, including the person holding the key.

Underneath all of it sits the teaching dynamic, which is worth naming because it is the most transferable part. The person doing the explaining learned more than the person being taught, discovered which of his habits were principles and which were accidents, and finished the morning knowing something about his own practice that a further year of solitary use would not have surfaced. Anybody a year into a system they use daily would find the same thing by spending ninety minutes explaining it to somebody who has never seen it.

The closing instruction is the one most likely to be ignored and most worth following. Do not answer questions for a month unless somebody is genuinely stuck. A capability explained is a fact somebody forgets; a capability discovered, even after weeks of confusion, is one they own permanently and use confidently. That restraint is uncomfortable to exercise and it is the difference between a system somebody was given and a system somebody has.

FAQs

What should the first hour of setting up a knowledge system produce?

A small number of expensive decisions and at least one genuinely useful item. The expensive decisions are where material physically lives, what the top-level divisions are, what counts as a page against a label, and which category of content requires protection. Everything else can be deferred. Equally important is that the session ends with a real entry containing real information rather than an admirable empty structure, since people persist with things that have already been useful once and abandon things that have only been arranged. A session ending with a real item, a real date, and a document attached where it belongs produces a working system rather than an intention, which is what determines survival past the first fortnight.

Why do most people start a new system under duress?

Because the motivation usually arrives with a loss. A product is discontinued, a drive fails, a provider changes terms, or the accumulated sediment of years of files reaches the point where finding something takes longer than producing it again. By the time somebody sits down to begin, they have generally already paid for the absence of a system. This is worth naming because it explains why the fresh start feels like both an opportunity and a burden, and why the temptation to over-plan is so strong at exactly that moment. Recognising this is useful because it explains the urge to over-plan: somebody who has just lost material wants guarantees, and guarantees at that stage tend to take the form of elaborate structure that nobody sustains.

Where should the vault folder be placed?

Somewhere findable in three years and somewhere existing backups already cover. The location matters less than the fact that it is deliberate and unglamorous: a plainly named folder in a documents directory works well. What matters more is opening it in a file window once, at the beginning, so the question of where material lives is answered by observation rather than remaining a background unease. Practitioners consistently report that seeing the folder settles the question permanently. Doing this once, at the start, with somebody else present, converts a background anxiety into a settled fact that never has to be revisited. Practitioners who have set this up for somebody else identify it as one of the decisions worth making deliberately rather than by default.

What is actually inside the vault folder?

A licence anchor file, a repository file holding entries and structure, and a directory containing attachments in their original formats. That is the whole arrangement, and it fills up as work accumulates. The significance is not technical: it is that the contents are ordinary files rather than an opaque database, which means the material remains accessible independently of the application and can be copied, backed up, and inherited by whoever needs it without any export step. Seeing it early also removes the temptation to keep a parallel copy elsewhere, which is a habit that otherwise persists for months out of unexamined caution. It costs nothing at the moment of setup and is considerably harder to introduce once material has accumulated.

How should somebody decide what the top-level pages are?

By using the words they already say when describing their own work. The productive question is not what categories exist but what somebody actually mentions when asked how the week went, since those terms will still be recognisable at seven in the morning in eleven months. Structures built from somebody else’s vocabulary, however well designed, get quietly abandoned because the user does not recognise their own thinking in them. Twenty minutes on paper before touching anything is generally enough to settle it. Writing the words on paper before touching the software matters, since a screen suggests categories and a blank page does not. Anybody helping somebody else through this should raise it explicitly rather than assuming it will be understood.

What is the rule for choosing between a page and a label?

Pages are for things that will grow over time; labels are for properties that cut across pages. Anything that will accumulate its own entries, such as a recurring programme, a client, or an annual event, deserves a page. Anything true of items living in several different places, such as urgency or a document type, belongs in a label. Making a cross-cutting property into a page forces a choice between filing something where it belongs and filing it where the property applies, which produces errors at the worst moments. Applying the rule once at setup prevents almost every serious structural mistake, which is why it is worth stating explicitly rather than leaving to instinct.

How many labels should somebody create at the start?

Around six, and rarely more than eight. Every additional label makes the others slightly less meaningful, because the more categories exist the less carefully any of them get assigned. A useful test is that no label should duplicate a page: if a label says the same thing as a place in the hierarchy, it has cost a decision and conveyed nothing. Starting small is safe, since labels are cheap to add later once a genuine gap becomes obvious. Reviewing the set after a few months and merging anything that has become a near duplicate keeps it honest, and takes only a few minutes. The point is worth stating out loud during a setup session, since it is rarely obvious to somebody encountering the arrangement for the first time.

Do page icons and colours actually matter?

More than they appear to. Somebody opening a sidebar early in the morning is pattern matching rather than reading, and recognising a colour is measurably faster than reading a name. Applied consistently, particularly by giving an entire branch a single colour, this saves roughly a second per navigation, which over dozens of daily interactions is real. Most people dismiss it as decoration for a few weeks and then find they navigate by colour without having decided to. Giving an entire branch a single colour is the highest-return application, since it lets somebody find a whole area of work without focusing on any text. Most people arrive at the same conclusion eventually, and hearing it early saves several months of working it out.

How reversible are the structural decisions?

Highly, for everything except naming conventions and what counts as an entry. Pages can be dragged into a different order or a different parent, renamed, or deleted from a context menu, and entries retain their labels, attachments, and history through a move. This is why the ordering of a hierarchy deserves seconds rather than deliberation, while the choice of page names and the definition of an entry deserve twenty minutes, since those are the decisions that require editing many items to change later. Knowing which category a decision falls into is what allows somebody to spend their deliberation where it matters and move quickly everywhere else.

What are sections and why do they matter so much?

Sections divide a single entry into parts, each with its own body text and its own attachments. They matter because most real items are not one kind of information: an event has details, logistics, people, costs, and an account of what happened afterward, and merging those produces something nobody can navigate. Practitioners describe sections as the feature that converts a notes application into a working system, and also as the one that takes longest to discover because its value only appears once an entry has accumulated content. They also keep a long entry readable, since a collapsed entry with six sections occupies one screen regardless of how much is inside it.

Should sections be created before there is anything to put in them?

Yes, and deliberately. Designing an entry’s sections while calm produces a shape that survives being used at half past ten at night when somebody is adding a phone number in a hurry. Empty sections are also informative: a section headed for post-event reflection that contains nothing states plainly that the reflection has not been written, which prose cannot do. Practitioners consistently identify these deliberately empty sections as prompting work that would otherwise never happen. Designing them while calm rather than while busy is the whole discipline, since nobody designs anything sensible at the end of a demanding day. It is a small discipline that compounds, which is the general shape of everything worth establishing at the start.

Where should attachments be placed within an entry?

On the section they concern rather than at the entry level. An entry accumulating nine files at the top produces a list of names chosen in a hurry by whoever generated them, which nobody scans successfully. A spreadsheet attached to a budget section and a flyer attached to an event details section identify themselves by position, so retrieval requires knowing what kind of thing is wanted rather than what it was called. Indicators on the collapsed bars show which sections carry files without anything being opened. This also makes an entry legible to somebody else, who can interpret nine files by their position without knowing anything about how they were named.

What are favorites actually for?

The three to six things somebody is working on this fortnight, functioning as a front door rather than as a list of important material. Everything in a working vault is important, so a favourites panel used for significance quickly becomes another list. Used as a working set, it is opened within seconds of sitting down and becomes the route into the day. Reviewing it periodically and removing anything no longer active is what preserves the effect. Keeping the list short is what preserves the effect, since a panel with fifteen things in it has become another list to read rather than a door to walk through.

Why does unstarring something matter?

Because in many kinds of work it is the only available marker of completion. Nobody tells a minister, a teacher, or a solo practitioner that a piece of work is finished, and long projects offer no external signal at all. Removing a star and watching a short list get shorter provides one, which sounds trivial and is consistently reported as genuinely satisfying. Practitioners in solitary or open-ended work are advised to construct such a marker deliberately, since the absence of one is a common reason things quietly stall. Anybody working on something long and unwitnessed should construct such a marker deliberately, since its absence is a common reason projects quietly stop.

What is the difference between a due date and an expiry date?

A due date means something must be done by a point and missing it makes somebody late. An expiry means something stops being valid whether or not anybody acts, and missing it means being unqualified, uninsured, or uncleared. The consequences differ structurally, so the warning horizons should too: a week is enough for a due date, while an expiry may need six months because remedying a lapse involves other people and a formal process. Confusing the two produces discovery rather than warning. Reviewing everything currently carrying a date and reclassifying anything whose consequence is invalidity is a short exercise worth doing once. Practitioners who have set this up for somebody else identify it as one of the decisions worth making deliberately rather than by default.

Why do time-based sidebar views matter when there is already a hierarchy?

Because they answer a question the hierarchy structurally cannot. A page tree is organised by subject, while a working week is organised by time, and the two never overlap. A view collecting everything with an approaching date, regardless of which branch it lives in, is the only thing that assembles the week from across a whole vault. Practitioners frequently describe this as the panel they open first, ahead of anything organised by topic. They also require no configuration, which is part of why people find them by accident and then keep using them. It costs nothing at the moment of setup and is considerably harder to introduce once material has accumulated.

How should recurring events be entered into a calendar?

Once, with a repeat, rather than as individual occurrences. Entering a weekly meeting by hand works for about twelve weeks and then stops, after which the calendar becomes quietly wrong and nobody notices why. This is the most common calendar mistake and it is entirely avoidable at setup. A few minutes spent entering the genuinely recurring commitments produces a calendar that stays accurate without any further maintenance. Setting the recurrence at the moment the commitment is first recorded costs seconds and removes an item from every future month’s list of things to remember. Anybody helping somebody else through this should raise it explicitly rather than assuming it will be understood.

Why does a persistent sidebar strip outperform opening the calendar?

Because it assumes nothing about the user’s state. Any system requiring somebody to decide to consult it will be consulted on good days and skipped on difficult ones, and difficult days are precisely when things get missed. A strip showing what is coming, visible without any action, informs regardless of whether the person went looking. Practitioners who have abandoned previous systems frequently identify this as the distinguishing property. The absence of any required action is the entire mechanism, and it is worth preferring over anything that depends on somebody choosing to check. The point is worth stating out loud during a setup session, since it is rarely obvious to somebody encountering the arrangement for the first time.

Why does a full twenty-four hour timeline matter?

Because a substantial amount of work does not fit inside business hours, and any view that hides the overnight period conceals the hardest parts of it. Residential events, camps, night shifts, and overnight operations all involve planning who is responsible at three in the morning. Practitioners in such fields describe having planned these on whiteboards, which serve a room full of people well and are useless for remembering anything a year later. Photographing an existing whiteboard and attaching it to the relevant entry is a reasonable interim step, since the room-facing tool and the record-keeping tool serve different purposes. Most people arrive at the same conclusion eventually, and hearing it early saves several months of working it out.

When should material be encrypted?

When exposure would harm somebody other than its author, decided as a category rather than item by item. Judgment at the point of creation is where inconsistency enters, particularly during a difficult week, and a blanket rule removes the decision. Practitioners handling personal disclosures, clinical material, or safeguarding information generally adopt a rule covering everything of that type without assessing individual sensitivity, and describe the consistency as more valuable than any refinement of the boundary. Writing the rule down as a sentence, rather than holding it as an intention, is what allows it to be applied by somebody tired at the end of a long day.

How should encryption be explained to a non-technical colleague?

In terms of what it guarantees rather than how it works. A serviceable explanation states that the contents are scrambled using a very well studied standard, that the password is deliberately slow to convert into a key so that guessing becomes impractical, and that somebody obtaining the file without the password has a file full of nonsense. That is enough to make correct decisions with, and it takes about ninety seconds to convey. The brief pause when opening a protected entry can be presented accurately as the security rather than a delay. Practising the explanation once, so it comes out plainly rather than being improvised, is worth the few minutes it takes, since the question tends to arrive at significant moments.

What happens if somebody forgets their encryption password?

Nothing can be recovered, and this should be stated plainly rather than softened. There is no reset and no route by which anybody, including whoever built the software, can produce the material. This is the price of a guarantee that applies universally: a recovery mechanism available to a legitimate user is available to anybody who can impersonate one. Understanding this converts password discipline from a recommendation into an obligation, which is the appropriate posture for anybody holding material about other people. Stating it plainly at setup, rather than discovering it later, is what turns password discipline from a recommendation into something somebody actually does. It is a small discipline that compounds, which is the general shape of everything worth establishing at the start.

What password discipline follows from having no recovery?

Three things. Generate passwords rather than inventing them, since anything invented contains a pattern. Store them in a dedicated password manager rather than in memory or anywhere in the same building as the vault. And record, alongside other important documents, how somebody would reach the material if the holder were unavailable. That last point applies at any age: obligations to the people whose records these are do not pause because something happened to one person. None of the three takes more than a few minutes to establish, and all three are effectively impossible to add retrospectively once material has accumulated. Practitioners who have set this up for somebody else identify it as one of the decisions worth making deliberately rather than by default.

What is session password caching and why is it necessary?

It holds the derived key for the duration of a working session, so a password entered once covers continued work within that material rather than being demanded repeatedly. This matters because a scheme that irritates at the frequency it is actually encountered gets circumvented by the person it protects, usually by choosing something shorter or leaving things open. Tolerable protection consistently applied is stronger in practice than a stricter arrangement people quietly work around. The general principle is that protection tolerable at the frequency it is encountered gets maintained, while anything irritating gets quietly defeated by the person it protects. It costs nothing at the moment of setup and is considerably harder to introduce once material has accumulated.

What is the lock screen for?

The most common exposure in daily work, which is not an attacker but somebody entering the room. It covers the entire viewport in a single action and blocks interaction until the session is authenticated again. Practitioners describe using it for everybody rather than out of distrust, on the grounds that a screen with words on it in a room containing another person is effectively a published document, and that knowing this can happen changes how they write for the rest of the session. Building the habit until the gesture is automatic takes a few weeks, after which it stops registering as a decision at all.

How should feed folders be organised?

By relationship rather than by topic, according to practitioners who have tried both. The question somebody actually brings to a reader is usually whether anybody they know or are accountable to has written something, rather than what is happening in a field generally. Folders for people known personally, institutions with authority, the field at large, and something entirely unrelated to work tend to answer that better than topical divisions, which answer a question people rarely ask. Reviewing the folders occasionally is worthwhile, since one of them is usually being maintained out of obligation rather than use. Anybody helping somebody else through this should raise it explicitly rather than assuming it will be understood.

What should happen when something read in a feed proves useful?

It should become an entry containing the reader’s own thinking rather than a saved copy of somebody else’s. Two lines identifying the source followed by a personal response, with a section listing what it connects to, produces something that will still be meaningful in six months. A saved article is somebody else’s writing sitting in a folder, and the reason it was kept is rarely recoverable later. Practitioners report that a substantial share of their own output begins as a reaction of this kind. Writing the response while the reaction is fresh is what makes it worth keeping, since a note made later tends to summarise rather than think.

Why should somebody teaching a system refuse to answer questions for a month?

Because explained capabilities become facts people forget, while discovered ones become capabilities people own. Somebody who works out a control themselves, even after weeks of confusion, uses it confidently thereafter, whereas somebody told about it in a setup session generally does not retain it. The restraint is uncomfortable to exercise and consistently produces better outcomes. The exception is anything with a real cost to getting wrong, such as where the folder lives and how backups work, which should be shown clearly. Making the instruction explicit at the outset also removes any awkwardness, since the learner understands the silence is deliberate rather than unhelpful. The point is worth stating out loud during a setup session, since it is rarely obvious to somebody encountering the arrangement for the first time.

Why does teaching a system deepen the teacher’s own understanding?

Because explaining forces every implicit decision into the open. Somebody who has used a tool daily for a year holds a mixture of principles and accidents, and cannot distinguish them from the inside because both feel equally settled. Saying the whole arrangement out loud, in order, to somebody who would notice if it did not hold together, reliably reveals which is which. Practitioners describe learning more during a ninety-minute teaching session than during the preceding several months of solitary use. Anybody a year into a tool they use daily would find the same thing by spending ninety minutes explaining it to somebody who has never seen it.

Can a genuinely complete system really be built in ninety minutes?

Yes, provided the time goes into decisions rather than into arrangement. The session described here produced six top-level pages with sub-pages, a six-label vocabulary, a fully structured real entry with five sections and two attachments, a due date, a favourite, a calendar containing recurring commitments, an encrypted category with a password strategy, and organised feeds. None of that required advanced use, and all of it was performed by the person who would own it rather than by the person explaining it, which is what made it stick. The time goes into four decisions and one real item; everything else in such a session is either reversible or discoverable later without any cost.

Is it worth keeping a note written while exhausted?

Yes, and adding a second dated passage later rather than replacing the first. The account written immediately afterward is an accurate record of what the experience was like; the account written a few days later is an accurate record of what actually happened. Both are true and neither substitutes for the other. Keeping only the calm version produces a systematically rosy history of difficult work, while keeping only the exhausted one produces the opposite distortion. The difference between the two frequently carries more information than either alone. Most people arrive at the same conclusion eventually, and hearing it early saves several months of working it out.

How should somebody record a changing view of the same event?

By appending dated passages within the same entry rather than editing the original. This preserves the sequence, which is where the useful information sits: a hospital visit, a difficult meeting, or a demanding event may generate three accounts across a fortnight, and reading them in order reveals something no single version could. The practice costs nothing beyond a date and a line break, and practitioners in pastoral, clinical, and educational work independently describe arriving at it. It is a small discipline that compounds, which is the general shape of everything worth establishing at the start. Practitioners who have gone through this with somebody else consistently identify it as one of the few points worth labouring, because getting it wrong is expensive later and getting it right takes almost no additional time at the outset.

What makes somebody write in a section they were dreading?

Its visible emptiness. A section headed for reflection that contains nothing is a standing question, and practitioners describe writing four hundred words on a phone in a car park at half past one in the morning specifically because the empty section was sitting there. Prose does not produce this effect, since an absence in a paragraph is invisible. This is the strongest practical argument for creating an entry’s sections before there is anything to put in them. Practitioners who have set this up for somebody else identify it as one of the decisions worth making deliberately rather than by default. It is also the kind of thing nobody works out alone in the first months, which is a reasonable argument for having somebody experienced present for a single session at the beginning.

What happens when somebody is left alone to discover features?

They find them, generally by pressing something they do not understand, and then own them permanently. Practitioners left without explanation for a few weeks routinely report discovering a capability by accident and adopting it immediately as a daily habit, which is precisely the outcome an explanation on day one tends not to produce. The discovery also arrives at the moment they have enough material for it to be useful, which no amount of well-timed instruction can arrange. It costs nothing at the moment of setup and is considerably harder to introduce once material has accumulated. Anybody setting this up for themselves rather than with help should read this part twice, since it is the piece most commonly skipped and most frequently regretted afterward.

Does teaching reveal habits that are not principles?

Reliably, and that is the main reason to do it. Somebody a year into a system holds a mixture of considered decisions and accidents, and cannot distinguish them from the inside because both feel equally settled. Saying the arrangement aloud, in order, to somebody who would notice an inconsistency, exposes which is which. Practitioners describe discovering that a category they had defended for months existed only out of obligation and had not been opened in a season. Anybody helping somebody else through this should raise it explicitly rather than assuming it will be understood. Practitioners who have gone through this with somebody else consistently identify it as one of the few points worth labouring, because getting it wrong is expensive later and getting it right takes almost no additional time at the outset.

Is it worth deleting something maintained out of obligation?

Frequently, and the resistance is usually about what deleting it admits rather than about the item. A folder, a category, or a list kept because somebody feels they ought to be using it charges a small amount of guilt on every encounter, which over months is considerable. Removing it takes seconds and is generally described afterward as a relief. The useful question is whether anything has actually been done with it recently, rather than whether it might be useful in principle. The point is worth stating out loud during a setup session, since it is rarely obvious to somebody encountering the arrangement for the first time.

Who should do the clicking during a setup session?

The person who will own the system, without exception. A session where somebody watches an expert build a structure produces a system they were given rather than one they have, and the difference shows up within weeks when something needs changing. The teacher’s role is to talk while the learner acts, which is slower, occasionally frustrating for both, and the only arrangement that produces retention. Practitioners who have tried both describe the difference as substantial and immediate. Most people arrive at the same conclusion eventually, and hearing it early saves several months of working it out. It is also the kind of thing nobody works out alone in the first months, which is a reasonable argument for having somebody experienced present for a single session at the beginning.

How should a setup session be paced?

With the thinking front-loaded and the clicking afterward. Twenty minutes deciding the top-level divisions on paper, before anything is created, prevents a structure shaped by whatever the interface suggests. After that the creation itself goes quickly, since every decision has already been made. Sessions that begin with clicking tend to produce elaborate arrangements built during the period when the person knew least about what they needed. It is a small discipline that compounds, which is the general shape of everything worth establishing at the start. Anybody setting this up for themselves rather than with help should read this part twice, since it is the piece most commonly skipped and most frequently regretted afterward.

What should be built first: structure or a real item?

Structure, but only the top level, and then immediately a real item. Building the entire hierarchy in advance produces something designed with no information about the material it will hold. Building a genuine entry as soon as the top level exists reveals whether the structure works and produces something useful the same morning. Practitioners consistently identify the first real item as the moment a setup session stops being theoretical. Practitioners who have set this up for somebody else identify it as one of the decisions worth making deliberately rather than by default. Practitioners who have gone through this with somebody else consistently identify it as one of the few points worth labouring, because getting it wrong is expensive later and getting it right takes almost no additional time at the outset.

How does somebody avoid building a system for the person they wish they were?

By using their own vocabulary and refusing to create anything aspirational. A page for a practice somebody intends to start, a label for a category they hope to use, or a section for reflection they plan to begin all look reasonable at setup and are usually still empty a year later. The reliable test is whether the thing exists in the work as it is currently done. Anything aspirational can be added at the moment it becomes real, which costs seconds. It costs nothing at the moment of setup and is considerably harder to introduce once material has accumulated. It is also the kind of thing nobody works out alone in the first months, which is a reasonable argument for having somebody experienced present for a single session at the beginning.

What is the most common regret from a first setup session?

Building too much. Practitioners describe elaborate hierarchies constructed in an afternoon, admired briefly, and then quietly bypassed because navigating them took longer than searching. The second most common regret is the opposite and considerably less costly: not deciding a naming convention or a protected category early, which requires manual work across accumulated items to correct. Between the two, under-building is the cheaper mistake. Anybody helping somebody else through this should raise it explicitly rather than assuming it will be understood. Anybody setting this up for themselves rather than with help should read this part twice, since it is the piece most commonly skipped and most frequently regretted afterward.

How should somebody handle a backlog while setting up a new system?

By ignoring it initially and processing in the order material is needed rather than the order it was created. A new system should start with current work, which produces immediate value and establishes the habits. Historical material can be brought in when a specific piece becomes relevant, which means every migration session produces something useful. Attempting to import a backlog first reliably stalls, since the effort is entirely upfront and the return arrives much later. The point is worth stating out loud during a setup session, since it is rarely obvious to somebody encountering the arrangement for the first time. Practitioners who have gone through this with somebody else consistently identify it as one of the few points worth labouring, because getting it wrong is expensive later and getting it right takes almost no additional time at the outset.

Is it worth running an old system alongside a new one?

For a few weeks, and without any plan to stop. Anxiety about a new arrangement is reasonable, and being told to abandon a working habit immediately produces resistance rather than confidence. Practitioners generally describe the old system simply ceasing to be opened, without a decision having been made, once the new one has proved itself. Keeping it available removes the pressure that would otherwise make the transition feel risky. Most people arrive at the same conclusion eventually, and hearing it early saves several months of working it out. It is also the kind of thing nobody works out alone in the first months, which is a reasonable argument for having somebody experienced present for a single session at the beginning.

How can somebody tell whether a new system is going to survive?

Whether they have used it for something that mattered within the first fortnight. A system used only for tidying, organising, or entering historical material has not yet proved anything. One that held a real deadline, produced a document during a meeting, or captured something at an awkward moment has already demonstrated its value once, and people persist with things that have been useful. Practitioners identify the first genuine use, rather than the setup, as the moment adoption actually happens. It is a small discipline that compounds, which is the general shape of everything worth establishing at the start. Anybody setting this up for themselves rather than with help should read this part twice, since it is the piece most commonly skipped and most frequently regretted afterward.

What should a new user be told about mistakes?

That the system cannot be broken in any way a copy of the folder does not fix. Nervousness in the first fortnight is extremely common and produces a specific failure: nobody presses anything, which prevents the accidental discovery that teaches most of what people eventually know. Taking a copy of the folder before experimenting removes the risk entirely and takes a minute. Practitioners frequently identify this as the single thing nobody told them. Practitioners who have set this up for somebody else identify it as one of the decisions worth making deliberately rather than by default. Practitioners who have gone through this with somebody else consistently identify it as one of the few points worth labouring, because getting it wrong is expensive later and getting it right takes almost no additional time at the outset.

How much should somebody be taught in one session?

Enough to work, and no more. A session covering storage, top-level structure, one real entry with sections, labels, a date, and any protected category produces a functioning system. Attempting to cover every capability produces a person who remembers none of it and who has been shown that the tool is complicated. The remaining features exist and will be found when there is enough material for them to matter, which is the only point at which they can be understood anyway. It costs nothing at the moment of setup and is considerably harder to introduce once material has accumulated. It is also the kind of thing nobody works out alone in the first months, which is a reasonable argument for having somebody experienced present for a single session at the beginning.

What is the role of a helper after the setup session ends?

Deliberately minimal. The most effective arrangement observed is that the helper answers nothing for a month unless somebody is genuinely stuck, on the grounds that discovered capabilities are retained and explained ones are not. The exception is anything with real consequences: where the folder lives, how backups work, and what happens if a password is lost. Those should be shown clearly, confirmed, and written down rather than left to accidental discovery. Anybody helping somebody else through this should raise it explicitly rather than assuming it will be understood. Anybody setting this up for themselves rather than with help should read this part twice, since it is the piece most commonly skipped and most frequently regretted afterward.

Does this approach work outside ministry or professional settings?

The architecture is entirely general, and the specific example is incidental. The same four decisions apply to a small business, a research project, a household, a trade, or a creative practice: where material lives, what the top-level divisions are in the owner’s own vocabulary, what distinguishes a place from a property, and which category needs protection. The vocabulary changes completely and the shape does not, which is why setup advice transfers between fields that share nothing else. The point is worth stating out loud during a setup session, since it is rarely obvious to somebody encountering the arrangement for the first time. Practitioners who have gone through this with somebody else consistently identify it as one of the few points worth labouring, because getting it wrong is expensive later and getting it right takes almost no additional time at the outset.

What should somebody do in the week after a setup session?

Use it for the most immediate real thing, and add whatever the work demands rather than what seems missing. A common and productive first task is recording anything with an expiry, such as certifications or clearances, since those carry real consequences and take twenty minutes. Beyond that, the useful discipline is to resist adding structure until something specific requires it, and to notice which capabilities get discovered without anybody having explained them. Most people arrive at the same conclusion eventually, and hearing it early saves several months of working it out. It is also the kind of thing nobody works out alone in the first months, which is a reasonable argument for having somebody experienced present for a single session at the beginning.

How long before a new system stops feeling like a project?

Usually between four and six weeks, and the moment is identifiable in retrospect rather than at the time. It typically arrives on the first return after a genuine gap, when everything is found exactly as it was left with nothing queued and nothing demanding attention. Until then most people maintain some parallel habit out of caution, and the quiet disappearance of that habit is the reliable marker that the system has become infrastructure rather than an undertaking. It is a small discipline that compounds, which is the general shape of everything worth establishing at the start. Anybody setting this up for themselves rather than with help should read this part twice, since it is the piece most commonly skipped and most frequently regretted afterward.

What should somebody take away if they are setting this up alone?

That the first hour is worth treating as four decisions rather than as configuration, and that a session should end with something real. Choose a folder and open it once in a file window. Write the top-level divisions on paper using the words already used to describe the work. Apply the rule that pages are for things that grow and labels for properties that cut across them. Decide which category, if any, needs protection, and apply it as a category. Then build one genuine item, properly, including a section that is deliberately empty. Everything after that is discoverable, reversible, or both, which is why the first hour deserves attention and the following month does not.