My Philosophy

I don’t worry very much. Some of that is temperament, and some of it comes from my parents: enjoy life as it is, make plans, and get on with things. I’m happy with my life and optimistic about what people can do.

When I first outlined this page in 2020, I left a lot of headings waiting to be filled in. Reading it six years later, I still recognise most of the thinking. The ideas that matter most to me are learning by finding mistakes, caring about people beyond those closest to me, and taking human creativity seriously when thinking about the future.

Learning by finding mistakes

Karl Popper, David Deutsch and Brett Hall have influenced how I think about knowledge. The idea I keep coming back to is that our explanations can be wrong, and that we improve them through criticism and finding errors. My notes on Popper’s Conjectures and Refutations and Deutsch’s The Fabric of Reality record some of that reading.

That is what I take from critical rationalism. I want to understand why an explanation works, what problems it solves, and where it fails. A famous name or a confidently stated opinion doesn’t settle a question, including when it comes from someone whose work I admire.

I’m interested in getting things right, while accepting that my current understanding is incomplete. That applies to my own experiences and intuitions too. It is easy to see the holes in someone else’s argument and much harder to notice the assumptions in your own.

I don’t spend my days practising philosophy. I listen to Hall’s podcast, read, and sometimes have an explicit discussion with a friend or my wife. Most of the time, these ideas sit in the background while I work and live my life.

The future includes what we invent

One idea from Deutsch and Hall has become especially important to me: people create knowledge, and that changes what happens next. A prediction about our future needs to leave room for solutions nobody has thought of yet.

An asteroid makes this easier to picture. If one were heading towards Earth, its path alone wouldn’t tell us how the story ends. We could try to change that path. We’ve already tested part of that idea: NASA’s DART mission changed the orbit of the asteroid moonlet Dimorphos.¹ The difficult questions are what we could detect in time, what we could build, and whether it would work.

The same thought makes me cautious about declaring what humanity will or won’t be able to do in the far future. Could we eventually live elsewhere, or find solutions that now sound absurd? I don’t know. I also don’t know what knowledge people will have created by then.

This is why I’m optimistic about our ability to solve problems. We can change the conditions we are looking at, sometimes in ways that are hard to imagine beforehand. I find that a much more useful starting point than assuming we have already seen the limits of what people can do.

AI, uncertainty and impressive-looking numbers

AI has made this feel much less abstract. When I wrote the first outline, it wasn’t part of my working day in the way it is now. Today I use it to write, build things and work through problems, including this page. Agents and scheduled tasks have become useful parts of how I get work done.

I’ve become more sceptical of “p-doom”: a precise percentage for the chance that AI will kill us all. Dressing up a guess about humanity’s future as a measured probability looks like bullshit to me.

Human creativity is what I think those estimates miss. People notice problems, invent solutions and change what happens next. We don’t yet know what knowledge we’ll create or what options it will open up. A precise p-doom figure makes the future look more settled than I think it can be.

I’m interested in the concrete problems with a technology and what we can do about them. I don’t feel well described by either doom or acceleration at any cost.

Where probabilities help me

My old outline gave Bayesian thinking its own heading. I still find it useful for questions such as how to interpret a medical test when you know something about the test and how common the condition is. Updating a probability as evidence arrives can clarify a well-defined question.

I’m less convinced by treating that as a complete account of how people learn or make choices. The part that interests me most is how we come up with a new explanation or an option we hadn’t considered. Hall’s criticism of Bayesian epistemology helped me articulate that distinction.²

I don’t think putting a percentage on a belief automatically makes it more rational. I’d rather understand the explanation and its weaknesses than feel reassured by a number whose assumptions I can’t examine.

Choices have a history

I feel that I’m the one making my choices. I decide what to work on, change my mind and sometimes regret something I’ve said. I’m not enlightened; that is still how it feels to me. But I don’t think those choices come from nowhere. What I want and what occurs to me depend on everything that came before.

Reflection and experience can change what I want. I can hear an argument, think about it and come to want something different. Yet I didn’t choose all the experiences that led me there, or give myself the disposition to reflect in that particular way. I don’t see where an ultimate author, independent of those causes, would enter the picture.

This is where I find useful ideas in Sam Harris as well as Deutsch and Hall, although they disagree about what deserves the name “free will”. Harris’s argument against ultimate authorship makes sense to me. I also find the emphasis on people creating explanations and choosing between options useful.³

Hall’s example of a copper atom in a bronze statue helps me think about this. Explaining why it is there involves the person being commemorated, the people who wanted a monument and the decisions involved in making it. Describing the particles alone would miss what I wanted to understand. Likewise, saying that someone thought of an option and chose it explains something about what happened. I don’t need that choice to be uncaused for it to matter.

Consequences with compassion

That affects how I think about responsibility. We need to protect people from harm, prevent someone from hurting others again and give people reasons not to do something harmful. Consequences can become part of what shapes a future choice. I don’t think any of that requires believing someone ultimately made themselves who they are and therefore deserves to suffer.

A brain tumour changing someone’s behaviour makes this easier to see. A published case describes a man whose harmful sexual behaviour resolved after the tumour was removed.⁴ That doesn’t settle the free will debate. It does make me question how readily we blame someone for being the person they are, especially when we can’t see the causes so clearly.

“Punishment with compassion” is how I’d put it. I still want consequences that protect people and deter harm. I want us to consider what those consequences achieve, rather than treating suffering as something we owe the offender.

Doing good, and having a life

I became interested in effective altruism after graduating. It gave me a way to think more carefully about helping: where could my money or time make a useful difference? That question still matters to me, even though I’m less actively involved in the community than when I lived in Rotterdam.

Reducing suffering is a strong motivation. I also want people to have good lives. I don’t want my concern for someone to depend only on whether they live nearby, whether I know them, or how easily I can picture their situation. The distinction between compassion and empathy still helps me here: I can care about someone without feeling exactly what they feel.

These ideas have affected choices I’ve made. Donating a kidney was one of them. Continuing to give money is another. They also help explain why I spend my working time on psychedelic research, policy and access: I want that work to contribute to people being better off.

I’m doing that while enjoying my own life. I live in the countryside, work on projects I care about, make quarterly plans and spend time on my goals. Having fun and being content matter to me too. I don’t experience doing good as something that requires me to worry about the world all day.

Progress takes work

My optimism doesn’t depend on believing the world gets better automatically. Things can get worse, and progress can be lost. COVID was a reminder of how much suffering a new problem can bring. Climate change is another problem that calls for sustained effort.

Medicine gives me reasons to be hopeful about what that effort can achieve. Effective HIV treatment can enable long, healthy lives, and treatments for multiple sclerosis can help manage the condition. Neither currently has a general cure.⁵

We need people to do the research, find the mistakes, test alternatives and make useful treatments available. My work in psychedelics keeps me interested in that last part as well: how do we turn promising research into something that can help people?

Ideas I’m still living with

I’m interested in questions about how the universe works, including Deutsch’s arguments about the multiverse. They make sense to me as ideas to explore. I wouldn’t present that interpretation as settled physics, and I’m still experiencing one life from one perspective. Those questions don’t all need to become rules for how I live.

A friend who is a doctor once said that we’re all trying to figure out what is going on. That stuck with me. Even in fields where we know a great deal, there are things we don’t understand.

One day I’d like to pass some of this on to my children. I don’t imagine starting by handing them Karl Popper; he isn’t the easiest read. I’d rather show it through how I act: taking questions seriously, being willing to admit a mistake, and looking for something we can try when a problem comes up.

Footnotes

1. What DART demonstrated.

   NASA’s account of Didymos and Dimorphos describes the 2022 DART impact and the change to Dimorphos’s orbit around Didymos. Neither asteroid was threatening Earth. This was a deflection test, rather than an asteroid intercepted on its way to hitting us.

   Back to text ↑
2. Probability and creating explanations.

   These paragraphs describe my view of how probability fits into thinking. In Bayesian “Epistemology”, Hall distinguishes Bayes’ theorem from broader claims about how knowledge grows. His Decision Making essay develops his argument about creating new options. These are philosophical arguments, not a demonstration that Bayesian methods cannot be used to study people or uncertain futures.

   Back to text ↑
3. Harris, Hall and different meanings of free will.

   In Life Without Free Will, Harris rejects ultimate authorship while arguing that deliberate effort and practical responsibility still matter. Hall’s Free Will defends the term for deliberation and choice. Hall explicitly draws on Deutsch’s account of emergent explanations and uses a copper atom in the nose of a bronze Churchill statue as his example. These are different philosophical positions; I’m drawing on parts of each, rather than claiming they agree.

   Back to text ↑
4. The brain-tumour case.

   Burns and Swerdlow’s 2003 case report describes a man with a right orbitofrontal tumour, acquired paedophilic behaviour and impaired impulse control despite retained moral knowledge. The authors report that the behavioural symptoms resolved following surgery. This is evidence about one patient, not proof of a general philosophical position or a basis for assuming every harmful action has this cause.

   Back to text ↑
5. Treatment is different from a cure.

   The WHO’s HIV and AIDS fact sheet explains that access to effective treatment and care can enable long, healthy lives, while HIV has no general cure. The NHS’s MS guidance says treatment can help manage the condition, but it cannot currently be cured. Neither supports a blanket promise of normal life expectancy or freedom from disability for everyone.

   Back to text ↑