PeopleThinkin Links.

Somewhere in Iraq right now there is a group chat holding the best notes ever taken on a particular medical course. A student wrote them, with love and at great effort, and posted them at two in the morning before an exam. By next year nobody will be able to find them. The student who needs them will start from nothing, and will write them again.

This goes on everywhere. A team at a university solves a problem, and the solution sits in a thread. An engineer explains how a local system really works, on a voice message. An NGO produces a good report as a PDF, sends it to a list of partners, and it is never seen again. The knowledge of an entire society is passing through chats, files and pages that never connect.

The dossier states this as a problem of Iraq and Kurdistan, and I think it is sharper there than in most places. We have few strong archives and libraries, and institutions have been interrupted by war and politics, again and again. People here are remarkably good at talking to one another. Our memory has always been oral, and it has served us well. But a culture that lives in conversation forgets everything that is not repeated. Digital chat has inherited that habit and sped it up.

A culture that lives in conversation forgets everything that is not repeated.

liiink starts with a claim about the mind. People think in links, not folders. Nobody remembers a thing by its location in a filing system. You remember it because it is connected to something else: a person, a place, a problem, a song. Memory is a web of associations, and that is how understanding works too. To understand something is to know what it is connected to.

Yet nearly all of our software was designed around the opposite idea. Folders. Feeds. Files. Each of those is a way of cutting knowledge off from whatever gave it meaning.

I find it useful to think about knowledge tools along two questions. Is the knowledge private or shared? Is it disconnected or linked?

The folder is private and disconnected. It feels organised, which is its great deception. People spend hours arranging notes in a hierarchy, and then never open them. A folder is a tidy grave for ideas.

The group chat is shared and disconnected. It is warm, fast and alive, and it is where our real intellectual life takes place. It is also a river. Whatever is said in it floats past and out to sea. Search barely helps, because you cannot search for what you have forgotten exists.

The personal graph is private and linked. In recent years a small community of researchers and writers has discovered how powerful this is. Each note connects to others, and over the years a person builds a second memory that can surprise them. These tools have mostly stayed with a technical, English-speaking minority.

The top-right corner is the one that barely exists, and it is where liiink is aimed: knowledge that is both shared and linked. A community graph. A group that can remember what it has learned, see how its ideas connect, and hand that on to whoever arrives next.

A group that can remember what it learned is a different kind of group.

Consider what that would mean for a university department. Each cohort leaves behind a connected body of notes, questions and sources. The next cohort starts further along. Or a community of doctors comparing cases across Iraqi cities. Or a civil society network that does not have to rediscover the same lessons every time its funding changes.

This is the principle of compounding that I keep returning to in all of my work. In other essays I argue that talent in my region is wasted because each generation begins again. liiink is the same argument at the smallest scale: a tool to stop one group of people from beginning again.

The third principle in the dossier is simple: communities think together. I take that literally. Intelligence has always been collective. No individual invented medicine. What we call a civilisation is a very large number of people connected by very good records.

A community graph cannot begin with the community. That is the lesson of nearly every failed knowledge project. Nobody contributes to a shared wiki out of duty. People write for themselves first.

So the product begins with one word and an exclamation mark: ME! Every person gets a workspace. It holds notes, files, links and projects, and it is useful on the first day, even if nobody else ever joins. Share later.

The growth of the system then follows the growth of a person's trust. Your notes become a personal graph as you link them. You open a part of that graph to a group. The group's graph develops into a community space. Some of it becomes public: linked pages that anyone can read and cite, rather than isolated posts that vanish. Each ring stands on its own, and each one feeds the next.

The second principle in the dossier is own your space, and I want to explain why it matters so much here.

For most people in my region, digital identity means a profile on a platform owned far away. Your friends, your writing and your reputation live on someone else's property, under rules that can change overnight, and in an interface designed to sell your attention. People in Iraq and Kurdistan have seen accounts disappear and pages removed, and they have watched a decade of writing become inaccessible. Your digital self should not be a tenancy.

Your digital self should not be a tenancy.

The word ME! is playful, and the idea underneath it is serious. Everything connects to a person. You are the centre of your own graph, and you decide what is linked, what is shared and what stays yours. For a young person in a society that gives them few private rooms of any kind, a space that is genuinely theirs can matter a great deal. It is somewhere to think before you are ready to be seen.

Language is part of this. The product is designed for Kurdish, Arabic and English from the first day. Almost every knowledge tool in the world treats our languages as an afterthought, with broken right-to-left layout and poor search. A Kurdish student should not have to think in English in order to organise her own mind. Tools shape thought. If the tools for serious thinking only work properly in English, then serious thinking will quietly migrate to English, and our languages will be left with the group chat.

The arrival of capable AI has changed what this kind of product can be, and it has made the underlying argument stronger.

The old weakness of personal knowledge systems was that they asked too much. To get value, you had to link everything by hand, and only obsessive people did. AI removes much of that labour. A system can now suggest connections between a note you wrote today and one from three years ago. It can find what you meant, even if you have forgotten the words you used. The dossier calls this smart search: find by meaning across your graph. A community's scattered notes can finally be asked a question in plain language.

But notice the dependency. A language model on its own knows the general internet. It does not know what your department learned last spring, or what your community has decided, or how things really work in your city. That knowledge sits in your own material. The most useful AI for a community will be the one grounded in that community's own graph.

A model knows the general internet. It does not know what your community learned last spring.

So a linked, well-kept memory stops being a nicety, and it turns into the thing that makes intelligence useful locally. Institutions that keep their knowledge in scattered chats will get a generic assistant. Institutions that keep it in a graph will get something that actually understands them.

There is a question of power here too. If the memory of a community is sent into a model owned far away, who benefits? Who can see it? Who can switch it off? I want a community's knowledge to serve that community first. For a society that has often had its archives destroyed or carried off by others, this is a practical matter as well as a principled one.

The AI should also stay honest about its sources. In a graph, an answer can point back to the note, the person and the date it came from. That is a real advantage over a model speaking in a confident, anonymous voice. People in a community know whose judgement they trust. A good system should show its work, and give credit to the people whose thinking it is using. Otherwise it is only a more polite way of taking something from them.

The furthest horizon on the roadmap is the idea that gives liiink its description as a spatial web: browsing knowledge as space, rather than as lists.

We are spatial animals. We remember rooms and routes far better than we remember abstractions. Ancient orators memorised speeches by walking through imagined buildings. The web began with a beautiful idea, the hyperlink, and then spent thirty years flattening it into feeds. A feed has no geography. You can never return to a place in it, because it has no places.

A living graph can be walked. You can stand at an idea and see what is near it. You can see where a community's thinking is dense, and where it is empty. That emptiness is interesting. It tells a student what nobody here has worked on yet. It tells an institution what it has forgotten to learn.

I should state where things are. There is an application codebase and a brand book. The next stage is a beta of the ME! workspace and a first university community, then a public launch in Iraq and Kurdistan. Universities are the obvious place to begin, because students are the people with the greatest need, the least budget and the most invitations to send. The moat, if it comes, will be the network effect of linked knowledge, and that cannot be copied by a larger company in a hurry, because it consists of a local community's own memory, in its own languages.

The name matters, as names do. The three i's belong to a family of ideas I have been building: liiist, miiind, liiink. liiist puts the facts of the world in order. miiind helps a person remember what they read. liiink connects people and what they know. They look like a typographic habit. They are really one thought seen from three sides: that knowledge is more valuable once it is structured, remembered and shared.

Knowledge becomes more valuable once it is structured, remembered and shared.

Let me return to that student and her notes at two in the morning. I would like her work to survive. I would like the student who comes after her to find those notes, improve them and link them to something new. I would like her name to stay on them. Multiply that one small act of continuity by a whole generation, and you have something our region has rarely been allowed to keep: a memory that accumulates.

Everything else follows from this. A people can lose a great deal and survive, provided that they do not lose the thread. A link is the smallest thread there is. It is one idea, holding on to another.