Quotes

Eating too is a fine thing, though it makes difficulties (do not laugh; economy and original sin may in fact be inseparable): and there is the pleasure at the conclusion of effort, the best of all delights, as the swimmer returned to the sunlight, his being glowing in the warmth responsive to the shocking chill of the waters (surrounded by them, he understood his body). Or the pleasure of the idle who, prone, full-length, made almost unknowingly a few exact perceptions, especially of those who hurry. And the satisfaction of the guilty (thus to have an identity not dissipated by their weakness); the delight of the famous, their self-regard coming from the outside; the joy of narration, thus to invent and, inventing, understand; the sweetness of the musician, from thunder and whisper tone’s moving constellations; and also the pleasure of small pains, the sweetness of anger, as Homer observes; the delight of the game (from out of the scrimmage came the tall and plunging figure and ran to a touchdown!); the pleasure of the task, the pleasure of the opus (the span, the parts, the detail, the conclusion); the delight, clear as fresh water, of theory and knowing (O lucid mathematics!). To each age and each stage a special quality of satisfaction, enough for everyone, and enough for all time, no need to compete. States of being suffice. Let the handsome be familiar with the looking-glass, and let the ugly be gourmets (since so many cannot be beautiful, let eating be socially superior to portraits). Let this unwarranted sadness come to an end, sound and fury signify a multitude of enjoyments, the pleasure of pain, the pleasure indeed of pleasure. Pleasure believes in friends, pleasure creates communities, pleasure crumbles faces into smiles, pleasure links hand in hand, pleasure restores, pain is the most selfish thing. And yet, I know, all this is nothing, nothing consoles one, and our problem and pain are still before us. Let us continue to gaze upon it. Let us, I say, make a few sharp clear definite observations before we die. Let us judge all things according to the measure of our hearts (otherwise we cannot live). Let us require of ourselves the strength and power to view our selves and the heart of man with disgust.

If the individual lived five hundred or one thousand years, this clash (between his interests and those of society) might not exist or at least might be considerably reduced. He then might live and harvest with joy what he sowed in sorrow; the suffering of one historical period which will bear fruit in the next one could bear fruit for him too.

A Flea settled upon the bare foot of a Wrestler and bit him, causing the man to call loudly upon Hercules for help. When the Flea a second time hopped upon his foot, he groaned and said, "O Hercules! if you will not help me against a Flea, how can I hope for your assistance against greater antagonists?"

With AI now, we are able to write new programs that we could never hope to write by hand before. We do it by specifying objectives (e.g. classification accuracy, reward functions), and we search the program space via gradient descent to find neural networks that work well against that objective. This is my Software 2.0 blog post from a while ago. In this new programming paradigm then, the new most predictive feature to look at is verifiability. If a task/job is verifiable, then it is optimizable directly or via reinforcement learning, and a neural net can be trained to work extremely well. It's about to what extent an AI can "practice" something. The environment has to be resettable (you can start a new attempt), efficient (a lot attempts can be made), and rewardable (there is some automated process to reward any specific attempt that was made).

Jevons paradox is coming to knowledge work. By making it far cheaper to take on any type of task that we can possibly imagine, we’re ultimately going to be doing far more. The vast majority of AI tokens in the future will be used on things we don't even do today as workers: they will be used on the software projects that wouldn't have been started, the contracts that wouldn't have been reviewed, the medical research that wouldn't have been discovered, and the marketing campaign that wouldn't have been launched otherwise.

With fitting exaltation, William Paton Ker wrote: "The great achievement of the older world in its final days was in the prose histories of Iceland, which had virtue enough in them to change the whole world, if they had only been known and understood" (English Literature, Medieval, 1912), and on another page of another book he recalled "the great Icelandic school, the school that died without an heir until all its methods were reinvented, independently, by the great novelists, after centuries of floundering and uncertainty" (Epic and Romance, 1896). These facts suffice, in my understanding, to define the strange and futile destiny of the Scandinavian people. In universal history, the wars and books of Scandinavia are as if they had never existed; everything remains isolated and without a trace, as if it had come to pass in a dream or in the crystal balls where clairvoyants gaze. In the twelfth century, the Icelanders discovered the novel—the art of Flaubert, the Norman—and this discovery is as secret and sterile, for the economy of the world, as their discovery of America.

I've been using Claude Code for a couple of days, and it has been absolutely ruthless in chewing through legacy bugs in my gnarly old code base. It's like a wood chipper fueled by dollars. It can power through shockingly impressive tasks, using nothing but chat. [...] Claude Code's form factor is clunky as hell, it has no multimodal support, and it's hard to juggle with other tools. But it doesn't matter. It might look antiquated but it makes Cursor, Windsurf, Augment and the rest of the lot (yeah, ours too, and Copilot, let's be honest) FEEL antiquated.

The key to understanding the pace of today’s infrastructure buildout is to recognize that while AI optimism is certainly a driver of AI CapEx, it is not the only one. The cloud players exist in a ruthless oligopoly with intense competition. [...] Every time Microsoft escalates, Amazon is motivated to escalate to keep up. And vice versa. We are now in a cycle of competitive escalation between three of the biggest companies in the history of the world, collectively worth more than $7T. At each cycle of the escalation, there is an easy justification—we have plenty of money to afford this. With more commitment comes more confidence, and this loop becomes self-reinforcing. Supply constraints turbocharge this dynamic: If you don’t acquire land, power and labor now, someone else will.

Originally, however, speech recognition was going to lead to artificial intelligence. Computing pioneer Alan Turing suggested in 1950 that we “provide the machine with the best sense organs that money can buy, and then teach it to understand and speak English.” Over half a century later, artificial intelligence has become prerequisite to understanding speech. We have neither the chicken nor the egg.

Before events took this bad turn, the contract represented by a link was simple: “Here’s a string, send it off to a server and the server will figure out what it identifies and send you back a representation.” Now it’s along the lines of: “Here’s a string, save the hashbang, send the rest to the server, and rely on being able to run the code the server sends you to use the hashbang to generate the representation.” Do I need to explain why this is less robust and flexible? This is what we call “tight coupling” and I thought that anyone with a Computer Science degree ought to have been taught to avoid it.

… there is a special place of torment reserved for those have been neutral in life. Their sin is regarded so grave that they are not even allowed into hell, only its vestibule, separated from hell by the river Archeron. For their sin of indecision and vacillation, Dante devised an appropriate and awful torment: they were condemned to rush for ever behind a banner “which whirls with aimless speed as though it would never take a stand, while also being stung by swarms of persuing hornets”.

I saw yet another “CSS is a massively bloated mess” whine and I’m like. My dude. My brother in Chromium. It is trying as hard as it can to express the totality of visual presentation and layout design and typography and animation and digital interactivity and a few other things in a human-readable text format. It’s not bloated, it’s fantastically ambitious. Its reach is greater than most of us can hope to grasp. Put some respect on its name.

My first games involved hand assembling machine code and turning graph paper characters into hex digits. Software progress has made that work as irrelevant as chariot wheel maintenance. [...] AI tools will allow the best to reach even greater heights, while enabling smaller teams to accomplish more, and bring in some completely new creator demographics. Yes, we will get to a world where you can get an interactive game (or novel, or movie) out of a prompt, but there will be far better exemplars of the medium still created by dedicated teams of passionate developers. The world will be vastly wealthier in terms of the content available at any given cost. Will there be more or less game developer jobs? That is an open question. It could go the way of farming, where labor saving technology allow a tiny fraction of the previous workforce to satisfy everyone, or it could be like social media, where creative entrepreneurship has flourished at many different scales. Regardless, “don’t use power tools because they take people’s jobs” is not a winning strategy.

The person who says, as almost everyone does say, that human life is of infinite value, not to be measured in mere material terms, is talking palpable, if popular, nonsense. If he believed that of his own life, he would never cross the street, save to visit his doctor or to earn money for things necessary to physical survival. He would eat the cheapest, most nutritious food he could find and live in one small room, saving his income for frequent visits to the best possible doctors. He would take no risks, consume no luxuries, and live a long life. If you call it living. If a man really believed that other people's lives were infinitely valuable, he would live like an ascetic, earn as much money as possible, and spend everything not absolutely necessary for survival on CARE packets, research into presently incurable diseases, and similar charities. In fact, people who talk about the infinite value of human life do not live in either of these ways. They consume far more than they need to support life. They may well have cigarettes in their drawer and a sports car in the garage. They recognize in their actions, if not in their words, that physical survival is only one value, albeit a very important one, among many.

The problem I have with [pipenv shell] is that the act of manipulating the shell environment is crappy and can never be good. What all these "X shell" things do is just an abomination we should not promote IMO. Tools should be written so that you do not need to reconfigure shells. That we normalized this over the last 10 years was a mistake and we are not forced to continue walking down that path :)

Our stockpile of antiquity grows constantly, it's overflowing, reckless squatters jostle for a place in history, hordes of sword fodder, Hector's nameless extras, no less brave than he, thousands upon thousands of singular faces, each the first and last for all time, in each a pair of inimitable eyes. How easy it was to live not knowing this, so sentimental, so spacious.

Tomorrow, in the fields of my kingdom, may you have a happy battle. May your kingly hands be terrible in weaving the sword stuff. May those opposing your sword become meat for the red swan. May your many gods glut you with glory, may they glut you with blood. Victorious may you be in the dawn, king who treads on Ireland. Of your many days may none shine bright as tomorrow. Because that day will be the last. I swear it to you, King Magnus. For before its light is blotted, I shall vanquish you and blot you out, Magnus Barfod.

Whenever you build a security system that relies on detection and identification, you invite the bad guys to subvert the system so it detects and identifies someone else. [...] Build a detection system, and the bad guys try to frame someone else. Build a detection system to detect framing, and the bad guys try to frame someone else framing someone else. Build a detection system to detect framing of framing, and well, there's no end, really.

The reason current models are so large is because we're still being very wasteful during training - we're asking them to memorize the internet and, remarkably, they do and can e.g. recite SHA hashes of common numbers, or recall really esoteric facts. (Actually LLMs are really good at memorization, qualitatively a lot better than humans, sometimes needing just a single update to remember a lot of detail for a long time). But imagine if you were going to be tested, closed book, on reciting arbitrary passages of the internet given the first few words. This is the standard (pre)training objective for models today. The reason doing better is hard is because demonstrations of thinking are "entangled" with knowledge, in the training data. Therefore, the models have to first get larger before they can get smaller, because we need their (automated) help to refactor and mold the training data into ideal, synthetic formats. It's a staircase of improvement - of one model helping to generate the training data for next, until we're left with "perfect training set". When you train GPT-2 on it, it will be a really strong / smart model by today's standards. Maybe the MMLU will be a bit lower because it won't remember all of its chemistry perfectly.

The idea that one needs powerful philosophical theories to settle [whether there are gods, witches or ghosts] might be called the "philosophy fallacy." People are particularly prey to it in religious discussions, both theist and atheist alike; indeed, atheists often get trapped into doing far more, far riskier philosophy than they need.

Before we scramble to deeply integrate LLMs everywhere in the economy, can we pause and think whether it is wise to do so? This is quite immature technology and we don't understand how it works. If we're not careful we're setting ourselves up for a lot of correlated failures.

The Steering Council (SC) approves PEP 779 [Criteria for supported status for free-threaded Python], with the effect of removing the “experimental” tag from the free-threaded build of Python 3.14 [...] With these recommendations and the acceptance of this PEP, we as the Python developer community should broadly advertise that free-threading is a supported Python build option now and into the future, and that it will not be removed without following a proper deprecation schedule. [...] Keep in mind that any decision to transition to Phase III, with free-threading as the default or sole build of Python is still undecided, and dependent on many factors both within CPython itself and the community. We leave that decision for the future.