What I Think, and Who I Read

One of the earliest people to prove that finding and surfacing what is new can be a real public service. Through Scobleizer, Fast Company and Building 43 he turned relentless signal-hunting into a visible media role: tracking fast-moving tech, interviewing founders, testing products before anyone had an opinion about them. More recently he built AlignedNews, a human-guided newsradar that combines his own source selection, his trust criteria and his taste with AI doing the running. He is where I learned to send people away from my own site, out to other people's pages, tools and writing. You go and look at those places first. You know they are good. Then you point, and you have saved someone a week of clicking around and guessing. That is what makes them come back to you.

She has been writing The Marginalian since 2006, back when it was called Brain Pickings. She does not write about her own ideas. She reads books, poems, old letters, science, and she pulls out the parts that matter and sets them side by side. That is the work, and she has done it for twenty years. She also credits everybody, not only the author but the person who pointed her to the author. From her I learned what crediting really does. You name the source, and you name the people or the places you came through to find it. The reader can go back and get it himself. He can see how you think and how you chose. He knows what is yours and what is not. And the writers and artists you name get found by someone who would never have reached them otherwise.

He built Seek Sense Share, an early framework that made curation understandable to an ordinary person. Before it, curation sounded either trivial or vague. Seek Sense Share shows you the actual work. You go and look for things. You sit with them until they mean something to you. You pass them on in a form somebody else can use. It takes real time, and the value gets built along the way, which is the part most people miss. I was in tune with it straight away, and it is practical enough that I could hand it to somebody else and they could start on Monday. He was also one more pillar of that small crowd of academics, writers and entrepreneurs who could see there was real value sitting in what is already out there, waiting to be found and made sense of.

He spent decades on how information gets laid out, and he showed that the same set of facts can be drawn a dozen different ways. The ones that work tell a story. They do it without decoration, without special effects, without colour thrown in to make things exciting. He wanted the data to speak for itself. He would count the ink on a page and ask what each mark was doing there, and if it was doing nothing, out it went. That has stayed with me longer than almost anything else. It carries straight into curating, because the hard part is knowing what to leave out, what to turn down, what to mute. Everybody else is busy making every item shout louder. When everything is screaming, nothing is really saying anything. For a curator, how you organise and present what you have decides whether the reader understands you at all.

He came up with the words hypertext and hypermedia in 1963, before there was a web to put them on. His Project Xanadu asked for things the web still does badly. Links that work in both directions, so a page knows who is pointing at it. Version history. Attribution you can see, that stays stuck to a piece of text wherever it travels. From him I took the idea that curation is partly plumbing. If you want people to understand something, the connections back to where it came from have to stay alive, not just get pasted in and forgotten. His 1974 book Computer Lib / Dream Machines is worth seeing too. It is a curated object in its own right, and he was picturing the personal computer years before most people had touched one.

In Deschooling Society, in 1971, he described something very close to the web twenty years before it existed. He called them learning webs. Instead of pushing everybody through the same institution, give people open access to the resources, let them find each other, let them pick their own guides and their own path. He spent his life arguing that institutions quietly take away a person's ability to do things for himself, and that a badge on its own is not a reason to believe anybody. From him I got the habit of being suspicious of packaged expertise, and the belief that a good filter run by a person you trust beats a funnel run by an institution. Good curation should leave a reader more curious and more able to go looking on his own, not more dependent on you.

She is not only writing about curation, she is building for it. Sublime is her attempt at a place where the things you save are connected to each other and to what other people have saved, instead of sitting in a private notes app where nobody will ever see them again. Her essays keep making the same argument, which is that as machines get better at producing more, faster, the things that go up in value are taste, noticing, and being able to tell what actually matters. From her I took the case for slowing down. Reading something properly, sitting with it, going deeper into one thing rather than skimming ten, is now a competitive advantage rather than an indulgence. She does not publish often. When she does, it is usually about something the rest of the field has not noticed yet.

Kei and Kazuki built Glasp together, and I have followed both for years. Glasp lets you highlight what you read and leaves those highlights out in the open, so what you save and what you underline becomes something other people can see. That turns a private reading habit into a public act of judgment, which is a small change that does a lot of work. They also grew the company mostly by paying attention to other people in public, praising them, quoting them, showing what they were reading. From them I learned that being generous in public pays off in a plain practical way, and that big things get built brick by brick. You do not need to be a great communicator to start publishing. They did not wait to be one.

He worked out early that the web pays off for whoever can spot a signal in the noise. He did it at TechCrunch and ReadWriteWeb, then built Little Bird, which found the real experts in a subject by looking at who the experts themselves paid attention to. He now runs two tools. What's Up With That helps you get more out of what you read. Hawkeye is a newsradar you build for yourself: you name the organisations, people and communities you want watched, you say what matters to you, and every day it hands back a sorted digest you can publish from. From him I took the idea that curation is partly a monitoring job. You decide whose signals are worth watching, you watch them steadily, and the pattern shows up because you were looking, not because you were clever.

He built Techmeme, and Techmeme proved that aggregation done properly is not secondary work. It is the infrastructure. Software finds the candidate stories, then a human decides what goes on the page and how it is framed, and it shows you the conversation around a story as well as the story. You get in minutes what would take an hour of going round thirty sites. He runs four of these on the same model: Techmeme for tech, Mediagazer for media and journalism, memeorandum for politics, and WeSmirch for celebrity gossip. What I take from him is that a very small operation can beat a large newsroom if the model is better, and that one page can become a habit for a whole industry when the signal is good enough. The same thing can be built for any niche, in any country, which is exactly what most people have not done yet.

He is not a curator, and I read everything he publishes anyway. He is an academic, an entrepreneur and a real technology enthusiast at the same time, which is a rare combination. He teaches this material to business executives. He took Karpathy's LLM wiki idea and built The Curator, the app I used to make my own knowledge base, and he put it out as open source. It is not the only thing he has built. He also keeps writing about what he builds: how AI agents lose track of what they were doing, how you get several of them working together, how you check their work, and how somebody who cannot code ships working software anyway, which is what he does himself. He keeps finding new things you can do with your own knowledge once a machine can read all of it.

He is a philosopher of information, and he gives you words for things you can feel happening but cannot name. His point about the infosphere is that information is not a set of tools we pick up. It is the place we now live in. The idea of his I keep coming back to is distant writing. You are not typing the sentences any more. You are directing, choosing, correcting, deciding what counts. That is still writing, and he is the one who made the case for why. He also argues that ethics has to be built into a system at the start, because you cannot bolt it on afterwards. Read him when you want to think about AI underneath the headlines.

He built the Apt Curation Model, and what it does is give a rigorous argument for something a lot of us had only felt. In the AI age the value shifts away from producing the text and onto judging it. In his model the human author is the evaluator, the architect, the integrator, and the one who carries the responsibility. That last part matters most to me. Whatever the machine did, the accountability stays with the person who published it. He matters because he put academic weight behind a position that practitioners were holding on instinct, and instinct is easy to dismiss. Read him if you want the philosophical case for why judgment survives automation.

He runs OLDaily, and he has run it nearly every working day for over twenty years. He reads a pile of things about learning and technology, picks a few, and writes a short note under each one saying what it is and why he thinks it is worth your attention. He saw early that the skill that matters is being able to find things, connect them and judge them for yourself, more than carrying a fixed body of material in your head. He describes his own work as an airport hub. Things arrive from everywhere, he sorts them, and he sends them back out. His piece The Network Second Layer was the first spark for me, and it is one of the places my newsmastering and newsradar ideas came from. He also taught me that knowing where to look is worth as much as knowing the answer.

He was the first to use the phrase “content curator”, on 30 September 2009, when curation still sounded like something that happened in museums. He also spent ten years on Non-Obvious, an annual trend report that started as a free PDF in 2011 and turned into a bestselling book, each edition naming fifteen trends he had built up out of small signals collected across a lot of different fields. He matters to me because he named the thing. I had been sensing for a while that a human editor of knowledge was going to be valuable, and nobody around me was saying it. He defined that role clearly and put it in print. That is what gave me the confidence that I was seeing something real, and that it was going to matter later, and not just a private hunch of mine.

He took personal knowledge management, which was a vague subject people argued about, and made it something an ordinary professional could actually run. Second Brain is the idea, PARA is the filing system underneath it. What he got right is that saving things is worthless unless you can find them again and use them for something. A note you will never retrieve is not a note. It is a feeling that you did some work. He also treats an archive as material rather than as a record. The half-finished thought you filed two years ago is fuel for the thing you are making today, if you built the shelves so you can reach it.

She is one of the youngest people I follow and one of the clearest signs that human filters are where this goes. Her publication is called Instead of Doomscrolling, and she does not just post links. She tells you why a particular essay or podcast or video is worth your hours. From her I took the point that curation protects a person's attention as well as informing it. She also reads all over the place, subjects with nothing to do with each other, and that is worth copying.

He writes Creative Destruction, and he is the king of reframing. He takes essays, articles, research and blog posts and looks at them from fresh angles, to imagine and explore alternatives the original piece never reached. His subjects are the systems most of us take for granted, regenerative futures, the stories a society tells itself. He uses curation to reopen imagination in places where the old mental models have gone stale. He is good for the kind of rabbit hole you come out of seeing your own field differently, and he is worth following if you care about where things may be heading.

She authors and contributes to three newsletters, Futuring Architectures, Thoughtforms, and Scenius Mag, in which she gathers several articles and excerpts around one theme. The name comes from Brian Eno. Scene plus genius, his idea that good work comes out of a whole circle of people feeding each other rather than out of one head. She calls herself a futuring architect, and that is the job. She takes a possible future and frames it in a way nobody else does. I did put her in my best curated newsletters of 2025 for unconventional sense-making and paradigm expansion.

Through Blockbuster Blueprint and everything since, he researches obsessively and in many directions at once, always chasing the same question. How do you learn best, share best, and produce knowledge and insight actually worth something. The range runs from video curation all the way to the most advanced AI systems, which he pushes well past what the mainstream suggests you do with them. He also does something almost nobody does. He runs a long conversation with an AI, or sets it a piece of reasoning and analysis to work through, and then publishes that conversation or that whole output as it stands. The thing itself is the artifact. From him I took the plain rule that better inputs make better outputs. If you want to make something worth reading, go and learn something first.

He is something like a Robert Scoble for creators. Scoble watched the technology and the founders. Greg watches what people are actually doing online. He runs Late Checkout, a holding company for community-based internet businesses, and he has been giving startup ideas away in his weekly letter since 2020, thousands of them by now. His method is the part worth copying. He looks for subreddits growing fast, then reads the comments for the same complaint repeating. Somebody saying they have been doing a thing in a spreadsheet for years is a business waiting to happen. That is curation used as market intelligence. Watch a community closely and you see demand forming before there is anything to sell to it.

He is the clearest example I know of somebody who built knowledge infrastructure that also makes money. Nomads.com, which used to be Nomad List, and Remote OK are the same act underneath. He gathered information that was already public and scattered across a hundred places, organised it, and filtered it for one specific group of people with one specific need. Where can I live, what does it cost, where is the work. None of it was secret. The value is entirely in the gathering and the arranging. People have kept going back to both for a decade. That is what infrastructure does.

He understands AI down to the machinery and can still explain it to an ordinary person, which almost nobody manages. He is as interested in what these systems cannot do as in what they can, which is why I believe him when he says something works. But the idea of his that changed what I do is the LLM wiki. Instead of a machine searching your documents every time you ask it something, you let it read what you give it and compile that into a wiki of its own, all interlinked, and keep the wiki up to date as more comes in. I fell for it straight away. For an expert it means you can finally see what you know at proper resolution, and pull value out of it you did not know was in there. That idea is what set Dr Tali Rezun off building The Curator, the open source app I used to build my own knowledge base, which is where everything on this site comes from.

He runs eBiz Facts, and what he does there is check. The field is full of people selling a course on how they made their money, and almost none of those claims ever get tested by anybody. He goes through them, names his sources, shows the criteria he used, and tells you what held up. From him I learned that curation can protect a person from a bad decision, and that is worth as much as informing him. I also like where his authority comes from. He shows you several paths, critically, and leaves the choosing to you.

He writes How to AI, and he does one thing relentlessly. He tests the tools himself, works out what actually functions, and writes it up in language so plain that anybody can follow it. Step by step, prompts you can copy, real examples. No jargon and no showing off. While he is doing that he is building the other half: archives, guides, frameworks and prompt collections that people keep going back to. He also runs a company that goes into organisations and helps them customise AI to do the specific things they need. The newsletter is what people see. The rest is where the money is.