---
date: Sunday, February 1^st^, 2026
subtitle: 2019--2025
title: Generalised readings
---

Once upon a time, at an end-of-year teaching team dinner, I was talking
with a friend of mine about New Year's Resolutions. Every year, she
said, she set herself the goal of reading 52 books the following year.
What an amazing achievement! To read one book a week! I realised I
wanted to spend substantially more time reading and, one day, achieve
this myself.

I haven't made it yet, but ever since the beginning of 2019, I've kept a
list of all of the books I have read. Actually, I'm counting more than
just books, for example there are a few web comics in my log, and I'm
counting more than just reading, for example there are a couple of
audiobooks and lecture series. That's why I called this post
'*generalised* reading.'

This weekend, I set myself the task of reviewing that the first six
years of my log. I've grouped by topic and in roughly chronological
order within topics all of the books from the list, and with some brief
notes about why I picked up each book or what I remember about reading
it (sometimes there is not much).

As a disclaimer, there's not much general insight to be found here for
anyone other than me. I wrote this for myself in the process of
reflecting on my reading habits. I am only posting it to encourage
myself to complete that reflection thoroughly.

For some deeper reflections, wait for my upcoming post *fear of missing
out.* Going forward, I'll try to make more in-depth book summaries and
reviews at the time of reading, so see [2026](readings-2026) for that
(as I note below, I might re-read a few of the books from this list at
some point).

## *Gödel, Escher, Bach*

It may be a slight exaggeration to say that there's life before one
reads ***GEB,*** and there's life afterwards. In my case, it's at least
literally true that this list covers my life from the point when I read
this classic by Douglas Hofstadter.

I picked up a copy on a recommendation in a CS lecture from my colleague
and mentor Justin Zobel. I was a TA for the course at the time, having
already graduated from a degree in computer science. In this sense, I
think I was a bit late to the party.

Even though I had a few years of experience teaching students about
recursion and the limits of computation, I've never dived that deeply
into music or art, and I hadn't actually studied Gödel's incompleteness
theorem, so the book had plenty of interesting and new material for me.

I am sure this book has since implicitly shaped my thinking, and even
the trajectory of my career, in countless ways. I should consider a
re-read. After all, RICERCAR!

## The Sherlock Holmes canon

I read the entire ***Sherlock Holmes*** canon from Sir Arthur Conan
Doyle, including the four novels and the five short story collections. I
found the characters and mysteries very entertaining. I have also
watched some of the film and TV adaptations, and I reliably enjoy them
too.

One of the most memorable things about the writing is the archaic
expressions that came up from time to time. One notable example is the
many uses of the word 'singular,' which I hadn't encountered many times
before in standard usage. As it happens, soon after reading the canon, I
would become a student in the Melbourne Deep Learning Group, spending
some of my time studying singular learning theory, and becoming familiar
with the mathematical version of the term. In fact I ultimately used a
quote from one of the first few short stories as the epigraph for my
Master's thesis:

> *I felt that he must have some solid grounds for the assured and easy
> demeanour with which he treated the singular mystery which he had been
> called upon to fathom.*\
> ---Dr. John H. Watson, "A Case of Identity," *The Adventures of
> Sherlock Holmes*

## The Rosie Trilogy

At one point in his life, Graeme Simsion was a data modelling consultant
who earned a PhD from and taught at my alma mater and wrote the textbook
used in my least favourite undergraduate subject. But my distaste for
*INFO20003 Database Systems* wasn't exactly Graeme's fault. In fact, by
the time I enrolled, he had left industry, taken some writing courses at
RMIT, and published his debut novel ***The Rosie Project*** to national
acclaim.

This I learned when I took INFO20003 and the lecturer explained who had
written the slides and why someone else was now teaching the course.
Several years later, in 2019, I found a copy of the book and its sequel
(***The Rosie Effect***) in an op shop, and I decided to give them a
shot. I found them totally hilarious and delightful, and soon after
ordered the third installment (***The Rosie Result***) to complete the
trilogy.

My only complaint is that my lecturer had said that the main character
had been loosely based on someone from the University, but, with no
further clues, I couldn't figure out who it was.

## Some science fiction

I read ***Encounter with Tiber*** by actual astronaut Buzz Aldrin and
actual writer John Barnes. I could not remember a thing about this book,
even after re-reading the synopsis on the Wikipedia page just now, so
that shows you how much of an impression it left.

I liked Shelley's ***Frankenstein.*** In particular, I thought her take
on what it's like to become conscious for the first time was pretty
interesting.

I read Douglas Adams's five-part ***Hitchhiker's Trilogy.*** Apparently
fans of the books don't like the 2005 movie, but I watched the movie a
while before and liked it, and it helped me follow the shenanigans in
the first book. I found the later books increasingly harder to follow,
but still sometimes pretty funny. I liked the ending of book 4, and
probably should have stopped there, or given the sequel by Eoin Colfer a
shot.

I found a really nicely bound edition of Asimov's ***Foundation
trilogy*** in an op shop for a few dollars! I found the premise of
psychohistory and the Seldon plan interesting, as well as some of the
political dynamics that play out over the years. However, I found the
frequent and long time-skips and the later sections involving plots
based on extensive psychological manipulation quite jarring.

My nice edition of the Foundation trilogy also contained ***The Stars,
Like Dust.*** I thought the ending was a bit cheap, but maybe it's
because I'm not an American. I haven't read the rest of the Galactic
Empire series.

I later found a second-hand copy of a short story collection called
***Machines That Think,*** edited by Asimov and others, which contained
some more Asimov and other works by various other classical sci-fi
authors.

In 2021, I would return to Asimov for the ***Robots series*** (*The
Caves of Steel;* *The Naked Sun;* *The Robots of Dawn;* and *Robots and
Empire*). I also read the short story anthology ***The Complete
Robot.*** I liked this series much more than the Foundation series. It
had cooler big ideas about cultural clashes between different branches
of space-faring civilisation and actual recurring characters like Susan
Calvin and R. Daneel Olivaw. Again, not a big fan of powerful
psychological manipulation as a plot device, but somehow it was not as
overdone here. At some point, I plan to read the Foundation prequels and
sequels, mainly for the tie-ins.

I read ***Snow Crash*** by Neal Stephenson, the book that invented the
metaverse, popularised the term avatar, and inspired the invention of
Google Earth. Very cool to see where these ideas came from before they
migrated from science fiction to reality.

I continued to look out for classic sci-fi novels in my local
second-hand book shop. Eventually I got a copy of ***Ringworld*** by
Larry Niven. It was interesting, but not enough to get me to look into
the sequels or his other work.

In contrast, I quite enjoyed William Gibson's ***Neuromancer,*** the
cyberpunk classic. I eventually found copies of the sequels ***Count
Zero*** and ***Mona Lisa Overdrive,*** though I didn't find them as
strong as the original.

More recently, I found the webcomic *Dresden Codak* by Senna Diaz. I
read ***Hob*** and ***Dark Science*** (to date). It is amazingly
illustrated, rich with philosophical texture, and every page takes an
unexpected turn. I don't know how she does it. I am still reading
whenever new pages come out.

## A little fantasy

I haven't read that much fantasy as an adult. I have noticed that I find
some fantasy writing a bit hard to follow, since the rules of the world
can be quite detached from the reality I am expecting. (Maybe a similar
problem came up with softer science fiction like *Hitchhiker's.*)

One example is Ursula K. Le Guin's ***A Wizard of Earthsea,*** a seminal
fantasy classic that has been famously underserved in adaptation. I
found the world really interesting, though a little hard to follow in
the places where it was more fantastical. I would like to return to
Earthsea (and some of Le Guin's other works) when I get the chance.

I find it much easier to follow fantasy in graphic novel form. For
example, I somehow found the work by the Finnish artist Minna Sundberg
and read through ***Stand Still. Stay Silent.*** (both series). These
were beautiful and gripping post-apocalyptic survival stories. I was
particularly impressed by how Sundberg visually communicated the dynamic
of characters speaking in a bunch of different languages. I also enjoyed
her shorter web comics ***Lovely People*** and ***A Redtail's Dream.***

I later read ***Guilded Age*** by T Campbell, Erica Henderson, and Phil
Kahn, a webcomic about a fantasy RPG-style adventure, with some sci-fi
twists. It was quite good.

It occurs to me now that a modern fantasy story that inherits all of the
tropes and touchstones pioneered by earlier fantasy writers is generally
going to be something I find easier to follow. I'm already familiar with
the rough shape of the fantastical 'logos'---the laws of the fantastical
reality therein. Whereas, someone like Le Guin who is writing outside
(before) that framework has been established, and who is genuinely
creating new fantastical realities, is going to be harder to follow, at
least without me putting in more effort or reading more work that adopts
a similar logos.

## Other comics and manga

I have already mentioned a couple of webcomics above, let me collect the
rest of my graphical media here.

I enjoyed ***Darths and Droids*** by the Comic Irregulars---a comic made
entirely from screenshots from the Star Wars movies, edited with dialog
as if the stories are being played out by a D&D party. It's hilarious. I
caught up at the time I read it in 2021, but haven't kept up since.

I worked through the entire archives of ***xkcd*** by Randall Munroe.
Maybe this should be listed under 'classics.'

As for manga, I was up to date with ***One Piece*** before this list
began, and I kept up throughout the period. It's a steep buy-in but once
you are in, there is no more epic or romantic adventure tale.

I also read the physical 12 volume set of ***Death Note*** books, which
I bought as a child after really liking the anime. It took me a while,
but I finally got around to it in this period.

Finally, at some point I decided that reading *One Piece* was not
enough, and I made the questionable decision to binge another extremely
long best-selling shōnen serial I had watched the first few episodes of
as a kid: ***Naruto.*** It was pretty fun and nostalgic to see out the
stories of the characters I had met as a child. I particularly
appreciated the themes of family and tradition that it explored in the
later sections.

## On history

I acquired an old book titled ***Sarton on the History of Science,***
collecting some of George Sarton's essays on various periods and figures
from the history of science and medicine. I remember finding this quite
interesting, though at this point the only thing I can recall is
Sarton's evocative claim that science alone of mankind's endeavours is
truly cumulative. I would definitely like to read this one again.

I also read Yuval Noah Harari's ***Sapiens.*** I was disappointed
because I had hoped that this would be a comprehensive introduction to
the history of the human species that could help me fill in gaps in my
own model of world history. In reality, it was more like an impassioned
reply that assumed massive amounts of existing context. This is not
necessarily a knock on the writer, who is not responsible for my
expectations. I do think the book had some interesting takes, e.g., it
points out that the agricultural revolution was actually great from the
perspective of the crops themselves, who Harari casts as domesticating
*us* in a sense. But it was also quite long, and I have since held off
reading Harari's other works.

I later found and started reading ***The Outline of History*** by H. G.
Wells, which was much more like what I was looking for, though it's a
very long book and I didn't make it very far. I should try again, or
maybe start with *A Short History of the World,* which is obviously
shorter.

## On the future

When I started getting into the field of AI safety I read some of the
popular books on the topic.

I started with the slightly less well-known ***The AI Does Not Hate
You*** by Tom Chivers. This is a journalist's exploration of the AI
safety / EA / rationalist ecosystem, and it contained lots of
interesting cultural insights. After being in this space for a few
years, I hope to revisit it and see if some of the community references
make more sense now.

Of course I also read Nick Bostrom's ***Superintelligence.*** As for
many in the field, the taxonomies and definitions developed in this book
have framed a lot of my thinking since then. My physical copy ended up
with a LOT of margin notes as my intuitions clashed with Bostrom's
claims, and particularly his suggestions for various cryptographic
alignment schemes. There were also cases where he opened my eyes to
flaws in my own position, such as how I previously thought that merging
with AIs was a promising path forward, however he pointed out that the
human--computer interface would always be a bottleneck. I would like to
revisit this book as well some day, in physical form so that I can also
review my old reactions in the margins.

Finally, I read Stuart Russell's ***Human Compatible.*** The first parts
of the book, where Russell is laying out the challenge of the alignment
problem, were really compelling to me, whereas the later chapters, where
he lays out his lab's at-the-time solution of beneficial AI and
assistance games, were deeply flawed. I would therefore recommend only
the early parts of the book, or to take the later parts with a lot of
salt.

## Existentialist literature

I watched JBP's lecture series ***Maps of Meaning*** (along with
***Personality and its Transformations***), which I found pretty
insightful (without endorsing his more recent move into the culture
war). Throughout his lectures, he makes a lot of references to
existential philosophy and literature, and I followed some of these up.
(In a similar vein, I listened to about two thirds of John Vervaeke's
***Awakening from the Meaning Crisis*** lecture series. This was some
pretty profound stuff and I want to re-watch and finish it some day.)

I have wanted to try Nietzsche for a while. I haven't made much
progress, but I did manage to read his autobiography ***Ecce Homo*** and
a short audiobook called ***On the Future of our Educational
Institutions.*** I really want to read more, but I haven't had the time,
mental space, or the courage so far.

I also read Victor Frankl's ***Man's Search for Meaning,*** plus a small
collection of short stories from other holocaust survivors. These were
quite emotionally challenging reads, but I think ultimately valuable.
It's important to know what the stakes are when thinking about the
future of AI.

I tried a little Dostoyevsky. While I haven't yet found the time or
mental space for his longer works, I did find some audiobooks of
***Notes from Underground*** and a collection of nine of his short
stories, including, memorably, *A Novel in Nine Letters,* *The Heavenly
Christmas Tree,* and *The Crocodile.*

## Some more philosophy

A long time ago, I bought myself a nice edition of some of the works of
Plato. I eventually got around to trying them. The collection included
the short dialogues ***Euthyphro***, ***Apology***, ***Crito***, and
***Phaedo,*** as well as the much longer dialogue ***The Republic.*** I
am pleased that I managed to follow the dialogues for the most part,
with the exception maybe of some of the stuff concerning the afterlife
towards the end.

I read some essays by stoic philosopher Seneca (**"On the shortness of
life;"** **"Consolation to Helvia;"** and **"On tranquility of mind"**).
I generally find some value in some parts of stoicism, but I wouldn't
say that reading these essays was that enlightening. I do remember
finding "On the shortness of life" fairly compelling, though I don't
think the impression lasted, I probably need to read it again.

I found a nice old collection of some of the works of John Stuart Mill.
I read some of the shorter works (***Thoughts on Poetry and its
Varieties;*** ***Bentham;*** and ***Inaugural Address at
St. Andrews***), mainly because this was more achievable than reading
the longer and deeper ones; I don't know that I got much from them. When
I did try *Utilitarianism* or *On Liberty* (can't remember which one), I
got stuck and put it aside, with the intention to return, some day.

I read Peter Singer's ***Practical Ethics.*** I'm a fan of Singer's
based on some of his impact, and I had been self-studying a little moral
philosophy at the time, so I was pleased to find it second hand.
Unfortunately, while a lot of the conclusions made sense, I recall being
somewhat disappointed by some of the arguments and their presentation in
the book.

As a more casual entry on this list, I read the entire backlog of the
excellent ***Existential Comics***. This webcomic is fantastic, and I
totally recommend learning about existential philosophy through satire.

## Some other classics

One of the best things that happened to my reading habit was moving to
Geelong, where there was an excellent second-hand book shop. Many of the
above readings had come from the philosophy and literature shelves
there. Some more finds, mostly from that book shop are as follows.

I read ***The Art of War*** by Sun Tzu, which I think I mostly followed,
and ***The Prince*** by Machiavelli, which I think I mostly did not.

I read the short essay **"Civil disobedience"** by Thoreau, but didn't
manage to make much of a dent in *Walden.*

I read a couple of societal parodies. As a budding AI safety researcher,
I read Samuel Butler's ***Erewhon.*** I read through Voltaire's
***Candide,*** though I think most of it was lost on me. A bit later, I
read ***Utopia*** by Thomas More, which didn't leave much of an
impression on me.

I finally managed to read Orwell's ***1984.*** Of course, everyone knows
a lot about the contents of this book without even having read it. I
read this immediately before reading *Neuromancer,* and the contrast in
writing styles was extremely stark: Orwell spends a lot of time
explicating the dynamics of and functions of his dystopia (which is the
real subject of the book I suppose), whereas Gibson throws you into his
future and lets you piece together how the world works implicitly
through the references made by the narrator or the characters.

Not itself a classic but a window into some old mythology: I read Neil
Gaiman's ***Norse Mythology,*** containing his telling of the ancient
stories. I found this pretty disappointing, and I would like to find a
different perspective on these stories.

## On self-improvement

I am not above seeking easy solutions to perceived inadequacy in areas
of my life!

***Deep Work*** by Cal Newport was very uplifting and left a lasting
impression. I have re-read it once since the first time in 2019. At some
point I completely quit Facebook, and I don't exactly remember but this
book might have been instrumental in that decision. Definitely still
worth reading again and pushing harder to get more deep work into my
life!

In late 2019 I was preparing to learn German as part of my exchange
semester to ETH Zürich, and I read ***The Fast, Easy Way to Learn a
Language*** by Bill Handley, a Melbourne-based author of study guides,
as this book was recommended to me by a family member. I don't remember
finding it very insightful. I had much more luck with my own systematic
learning approach, though my motivation was cut short when the pandemic
hit a couple of weeks into my exchange.

I *do* remember reading ***Ultralearning*** by Scott Young, and finding
it very *un*insightful. There were some interesting anecdotes throughout
and I don't have objections to the specific recommendations, but the
taxonomy put forward in the book is embarrassingly, unforgivably lazy
and underdeveloped, and this made reading it perhaps more enraging than
inspiring. Another memorable thing about this book is that merely
reading the foreword by James Clear made me swear off ever reading
*Atomic Habits.*

I would go on to read ***Peak*** by K Anders Ericsson. That is a much
better book about how to learn and get better at things than
*Ultralearning.* Deliberate practice is a core concept in how I think
about learning, though I wouldn't say I have succeeded in implementing
it in my life as thoroughly as I would like.

I later read ***Grit*** by Angela Duckworth. I found this one inspiring
but also challenging to read. Whenever I take these kinds of
psychological surveys for measuring personality traits, I don't know how
to interpret the questions. I can see myself strongly agreeing or
strongly disagreeing depending on what kind of contexts in my life I am
thinking about. In this case, the results tell me I am either in the
bottom 10th percentile or the top 10th percentile for grit. So helpful!

At some point I read an old book called ***Rational Recovery*** by Jack
Trimpey. This is a book about mastering alcoholism but I interpreted it
more broadly as about self-control and discipline. Anyway, it really
resonated with me and I think has led to some actual improvements in my
life, though perhaps I should revisit it and see if I can eke out some
further improvements from this vein.

Straight after reading Trimpey I read ***The Scout Mindset*** by Julia
Galef. I know this is a popular book, but everyone hears about popular
things for the first time at some point, and I happen to remember that
in my case, I heard about this book when I was speculating to some
rationalist colleagues that maybe militaries have developed particularly
useful practical knowledge about gathering information and forming
accurate world models, and that I would be interested in learning from
these sources. Well, I was informed that my idea was not original, in
fact professional rationalist Galef had last year published this book on
a similar topic. Great, that saved me a lot of trouble!

I liked the book, but in the end I realised that I had already
internalised a lot of what it had to say. At least I had internalised it
on an intellectual level---not that I'm a master at putting it all into
practice. But no book is going to help me much with that part of the
process. Anyway, come to think of it, while the scout/soldier
terminology is inspired by a military analogy, I don't remember whether
the book itself drew *that* much on lessons from the military world.
Maybe something still to ponder.

## Books received as gifts

Giving books as gifts is pretty hard. I don't always read books that I
receive as gifts. At some point, I noted I did read ***In the Heart of
the Sea*** by Nathaniel Philbrick, and ***a biography of Albert
Einstein*** by Sri Lankan professor K. A. I. L. W. Gamalath. I wouldn't
have picked these books myself, but I remember enjoying them at the
time.

## Eliezer Yudkowsky

In 2024 and 2025, for reasons I will describe in an upcoming post, I
found almost no time for reading books at all as my "generalised
reading" habits got a bit stuck on listening to TTRPG podcasts. The one
author who was able to cut through this has been Eliezer Yudkowsky (a
somewhat divisive figure, I'm a fan overall).

In late 2024 when I arrived in Oxford, I resolved to finally read 'the
sequences' (a.k.a. ***Rationality: from AI to Zombies***). I made this
feasible by finding an audiobook version, and now many of my walking
paths around Oxford trigger memories of snippets of the book. I had
already internalised most of the ideas before reading, including since
there is some overlap with some of Yudkowsky's publications on
existential risk, but there were still a few new things that stood out
to me as valuable frames, and it's always good to get a reminder to put
this stuff into practice.

I read a bit of ***Inadequate Equilibria*** but didn't get the chance to
finish it at the time. I got the main idea from the first few sections,
which is a sign of a well-written book if you ask me.

I pre-ordered ***If Anyone Builds It, Everyone Dies*** (Yudkowsky and
Soares), and I read the whole thing the Saturday after it arrived. I
thought this was a solid read, and I have recommended it to a few family
members and close friends. It was particularly nice to see the modern
argument for AI risk all at once. Of course, this was highly relevant to
me, as I view my research aim as turning this argument from one driven
by intuition and analogy to one that has been rigorously empirically
tested.

## Books I haven't read

I bought a lot more books than I read over this period. Most of them are
back home in Australia at the moment. The year I read the most books
over this period was 2020. In that year, I set myself a rule that I
should read 12 books, and if I buy a new book, I have to increase the
target by one. Therefore, I limit the rate at which my book collection
can get away from me.

This worked well as an incentive, but I don't actually think collecting
unread books is a bad thing in itself. See the Japanese concept
[tsundoku](https://en.wikipedia.org/wiki/Tsundoku) or Umberto Eco's
concept of an [antilibrary](https://en.wikipedia.org/wiki/Antilibrary).
My ever-growing book collection can, in principle, serve as an important
reminder of the sublime depth and breadth of human civilisation as I aim
to extend our knowledge for the good of all of us. While I left my
collection at home, my favourite place in Oxford is the Norrington room
at Blackwell's, since that place instills in me the same feeling (and
because it has lots of levels).

However, there is a failure mode I am at risk of falling into. One thing
I would like to do now after having looked at the books I *have* gotten
around to reading, is to look over my bookshelf at all the books I have
*not* gotten around to reading. I suspect I will find a pattern where I
tend, not always but more often than not, to put off reading things that
seem too deep or demanding of my attention. The result is that I end up
wasting my time with shorter and more shallow reads, or books that I am
reading just for the sake of saying I have read them. Evidently, a few
years later, many of these books have left no lasting impression on me.
That is a real shame.

I have had some quite intense experiences of nostalgia, profundity, and
joy when reading good books throughout my life. It's not always possible
to tell in advance when I am going to feel these things, but sometimes I
have good reason to think a book is going to have this kind of effect,
and I actively put off engaging with it, rationalising that 'now is not
the time.' But there is never a 'time' for reading unless I make it.

Here's hoping I can be more bold in my choices of titles going forward.