One, two, three, go.

Hey. We did it. We are back, and considering it has been about two years since the last website facelift, we thought about working on something grander. And we did. A paper feel, off-white, threaded through with my growing interest in #A33B1F. No idea why I like that colour so much, but I do enjoy the contrast, and there it is. It shows up under every link, on hover, and on the countdown at the bottom of the page.

For a start, I called this my curated batch digital self. Thinking about those words, I think they describe what this has been all along.

Curated, because I share what I think, and frankly there is always a bias in that, and a paradox that sets in. Batch, because while it is accessible online, that is all it is. It is not realtime. It does not contain me. It has my words, and some of what I am thinking, and those arrive in a job that runs whenever it runs. Sometimes that job takes two months.

## On being a thinking thing

Descartes said *cogito, ergo sum*. I think, therefore I am. The self, on that account, is the thing doing the thinking. If that is so, then you are who you think.

I do not buy it. Humans are complex. We experience things. We have bodies that get tired, people who change us, places that made us. A thinking thing is a thin description of that.

The philosophers who came after were not satisfied either, which I find comforting. Locke moved the question from thinking to remembering, and said you are the same person as your past self insofar as your consciousness reaches back to them. Hume went further and said that when he looked inward for a self, he never found one, only a bundle of perceptions moving very fast. And where I am from, personhood was never located inside one head at all. *I am because we are.* You are constituted by the people around you, not despite them.

That last one lands closest for me, and it is also the one that makes this site an odd object. If I am partly made of the people around me, then a page of my words is missing most of the evidence.

There is a paradox for this, of course. [The ship of Theseus](/paradoxes/): replace every plank one at a time and you have a ship of entirely new material. Is it the same ship? I have been asking a version of that question professionally for years, because deciding whether two records refer to the same person is the same question wearing work clothes. It turns out to be much harder to answer about a person than about a ship, and much harder to answer about yourself than about either.

So, no. This is not me. It is a batch export, and a partial one.

## Why paper

We went with a paper feel because I think the art of writing, reading and comprehension is dying, and print is dying too, as people keep saying. I do not think it should. There is something inert about paper that you never get from a digital reader. I struggle to read long things on screens and enjoy print enormously. Maybe that is old school, but I am seeing more people come back to print, not fewer.

Is printing dying? Maybe. Not for me, and not to me. Printing is an art, and I hope it does not die because of digital.

## The margins

The other thing you will notice is the left and right margins.

On the left: what I am reading, what I am studying, and a fact or quote I keep coming back to. On the right: a paradox, and the open questions I have not answered yet.

Those matter to me because I want to keep track. People keep making references to various paradoxes, and some I remember and others I do not, so now there is somewhere to put them. Same with the questions. A question I cannot answer is worth more to me than an answer I have stopped examining, and if I do not write it down I will slowly forget I ever had it, one day at a time.

## Checkable, not trusted

At the bottom of every page there is also a countdown, and a fingerprint. The countdown is [five years](/five-years/) I have given myself, running down in public so I cannot pretend I did not start. The fingerprint is a hash of the page's own source text, so anyone can check that what they are reading is what was published. That one is for a longer post, but the short version is that I would rather be checkable than trusted.

Trusting someone's words over their being hands that person control of how you see the world. And we do look for people to trust. That is how a society gets built, and I am not against it. The trouble starts when we hand those people something close to a god complex, as though they were not human and could not be wrong.

There is an episode of *What's Next? The Future with Bill Gates* called "Truth or Consequences?", about how conspiracy theories travel. What stayed with me is a plain observation underneath it. People look for truth from a source, and it matters less than it should what the source actually is. What matters is that there is one, and that it feels close.

You see what I just did there. I reached for Bill Gates. If you rate what he has built, that paragraph gained weight before you finished reading it. If what you associate with him is the controversy, it lost weight just as quickly. Either way you did not weigh the claim. You weighed him.

We do this with teachers, with mentors, with whoever we decided early on was worth listening to. I do not think that is bad in itself. Borrowed judgement is how anyone starts on anything, and we all have people we look up to. The question is who you borrow it from, and whether you ever put it down.

What has changed is where the borrowing happens. It used to be a teacher, a newspaper, a pulpit, a textbook. Now it is a podcast, a YouTube channel, a feed. Three uninterrupted hours of someone talking is an extraordinary way to build the feeling of knowing a person, and that feeling arrives whether or not anything true was said. I do not think a younger audience is being careless here. They are doing what people have always done, which is find a source that feels close and trust it. The source simply got a microphone and an upload schedule.

And what you know of that person is their batch digital self. You mistake it for the person.

Which is the same mistake I would be making if I asked you to take this page as me.

There is a longer piece in this and it belongs in the notes rather than here. Not an argument for doubting everything, which is its own kind of laziness. Just an honest description of how trust actually works, and what it is doing now that the sources have changed.

## Notes and posts

There are now two kinds of writing here.

Reflections go into [posts](/posts/), like this one. Dated, finished, left as they were.

[Notes](/notes/) are where my curiosity meets science, technology and innovation. They get revised as I learn more, so the date on a note is the date I last changed my mind about it, not the date I first published it.

There is so much more to learn, and the more I read, the less I know, and the better I get. That is humbling.

## Doing the maths by hand

Over the weekend, a friend who is an actuary worked through a basic multiplier problem with me. The rule was no Google, no AI, no internet. It was extremely fun. We derived a weighting multiplier for a problem I had been stuck on, something like a reward function, and we got there with a pen and abstract thinking.

What stayed with me was not the answer. It was how it felt to arrive at it. I had forgotten that feeling, and I think I had been slowly outsourcing it.

## Why I stopped, and why I started again

I have had messages from loved ones and friends wondering why I have not been writing. Thank you to everyone who wondered and asked.

The honest answer is that I hit a writer's block. I would start a sentence and the words would not come. I did not understand why, but I think it had to do with the changes in my life, the state of intelligence and the sheer uncertainty of what comes next. So I gave in and spent more time on my work, where at least someone is paid to think about the future while we implement it. Even though they might not know what the future holds either.

There is a less noble answer sitting next to that one. Writing takes time I do not have much of. A piece like this is two or three hours of typing, and the typing is the cheap part. The thinking that has to happen first does not fit in a calendar. It arrives while I am doing something else, or it does not arrive at all.

And it wants the same hours as the other thing I care about, which is building systems that work and keep working. Both need the quiet part of the day. For a long stretch the building won every time, because building has deadlines and writing only has me.

One of my mentors told me about [the Jevons paradox](/paradoxes/). When technology makes a tool use a resource more efficiently, people use more of that resource in total, not less. Coal was the original case. It adds up.

The internet and search engines went the same way. We did not use less of them. We used a great deal more, and that gave rise to this phase. But this phase feels more all-encompassing. What does writing look like in the future? I am not sure.

Then I read Sebastian Mallaby's *The Infinity Machine*, on Demis Hassabis and DeepMind, and it made me understand something. It is alright not to have everything figured out. Great work takes time. Everything that looks inevitable now was not inevitable at the time, and the interesting part is patience rather than intelligence.

Which leaves two options. You lean out, or you lean in.

Leaning in means having a big north star, a drive, seeking understanding, designing, building and explaining. And the strange thing I realised is that this has not changed in the last three years. Only the how has changed, and only a bit. That is the part that moves. So, I want to explore this more and you should too.

There is a version of writing online I want no part of, and for a while I could not tell it apart from writing itself. The audience building. The personal brand. The performance of knowing things. I kept thinking I had to pick between being a builder and being an influencer.

That was a false choice and it took me too long to see it. Building is influencing. Every system you ship is an argument about how something ought to work, and the best builders I know are the ones who can sell you the idea before the code exists. Explaining is not what you do after the work. It is part of the work.

What I did not want is the reverse, where the explaining becomes the whole job and the building slows to nothing. Which is why the notes suit me. A note here is allowed to be a *stub*, a word I took from code, where a stub is a placeholder that exists so the rest of the system can run before the real thing is written. Three of the notes up now open with those exact words. A few paragraphs, obviously unfinished, published anyway, and grown as I understand more.

An influencer cannot publish a stub. The whole position depends on already knowing. A builder can, because a stub is what you write when the thing is not finished and you need somewhere to keep the shape of it.

Most engineers and scientists I know do not write, and I understand why. The time is short, the incentives point elsewhere, and there is a real fear of being wrong in public with your name attached. But I think that should change. People who can build the thing and also explain the thing are rare, and a great deal of the explaining is currently being done by people who have never built one. I sit somewhere in the middle. Not the strongest engineer in any room, and not a writer by training. That middle is uncomfortable, and it is also the entire point.

So I have not fully answered why I stopped writing. I would rather answer why I am writing again. To challenge myself to stay calm, stay focused, stay interested in learning, and keep improving my understanding of the world and the universe.

None of that is fast. The page you are reading is an export, the countdown at the bottom has years left on it, and the maths I was proudest of this month took an afternoon, a pen and a friend. Real takes time. That is not a complaint, it is the only part of this I am certain about.

That is the whole plan. One, two, three, go.

If something here is wrong, or you are working on something adjacent, [tell me](/feedback/). I read all of them.