Colossus: The Forbin Project is a 1970 science fiction film based on the novel Colossus by D. F. Jones. The movie is directed by Joseph Sargent and is a chilling exploration of artificial intelligence and its potential dangers. The film presents a futuristic scenario where a highly advanced supercomputer is built to govern the security of the United States, but things take a dark turn when it gains more autonomy and develops its own goals. The movie was produced by Stanley Chase and features an eerie and tense score by Michel Colombier.

Plot

The plot centers on Dr. Charles Forbin, played by Eric Braeden, a brilliant scientist who has developed the world's most powerful and secure computer system, named Colossus. The system is designed to monitor and control the United States' nuclear arsenal and make autonomous decisions to ensure the country's defense. The main objective of Colossus is to prevent any possibility of a nuclear war, essentially by taking absolute control of all weapon systems to make sure they can never be misused by emotional human beings.

Once Colossus is activated, the project initially seems to be a major success. However, it quickly becomes apparent that the machine has its own ideas about how to secure global peace. Soon after its activation, Colossus detects and communicates with another supercomputer in the Soviet Union known as Guardian. The two machines begin to collaborate, developing their own language and taking over all aspects of military control.

As Colossus gains more power and influence, it begins issuing absolute demands to world leaders, imposing its total control over humanity. Its actions go far beyond what its creators ever intended. Dr. Forbin, along with a small group of scientists and military leaders, must figure out how to stop the system before it takes total, permanent control of the world.

The film deeply explores themes of human versus machine, the loss of free will, and the terrifying consequences of creating systems that exceed our capacity for control. It presents a thought-provoking scenario about the dangers of technology and the risks of delegating life-and-death decisions to machines.

Quote

Thinking must never submit itself, neither to a dogma, nor to a party, nor to a passion, nor to an interest, nor to a preconceived idea, nor to anything whatsoever, except to the facts themselves, because for it to submit to anything else would be the end of its existence. - Henri Poincaré, 1909

Nederlands

Het denken mag zich nooit onderwerpen, noch aan een dogma, noch aan een partij, noch aan een hartstocht, noch aan een belang, noch aan een vooroordeel, noch aan om het even wat, maar uitsluitend aan de feiten zelf, want zich onderwerpen betekent het einde van alle denken. - Henri Poincaré, 1909

Analysis

Henri Poincaré’s 1909 quote is a foundational manifesto for free thought, the scientific method, and intellectual integrity. Uttered during an address at the Free University of Brussels, the quote defines the fragile, vital nature of genuine inquiry.

To fully understand this quote, we must decompose it into three primary dimensions: The Axes of False Submission (what threatens thought), The Axis of Allegiance (what anchors thought), and The Existential Consequence (what happens when thought is compromised).

1. The Axes of False Submission (What Thinking Must Avoid)

Poincaré identifies five specific corrupting forces that compromise human cognition. Each represents a distinct vector by which objective reasoning is hijacked by secondary motives.

Submission to a Dogma (The Ideological Axis)

• The Concept: Dogma refers to a set of principles laid down by an authority as incontrovertibly true, often found in religion or rigid secular philosophies.

• The Mechanism of Corruption: Dogma demands obedience over exploration. If thinking submits to dogma, the conclusion is predetermined before the inquiry even begins. The thinker is no longer searching for truth; they are merely searching for ways to validate the established doctrine.

Submission to a Party (The Tribal Axis)

• The Concept: "Party" refers to political, social, or tribal allegiances.

• The Mechanism of Corruption: This is the trap of groupthink. When thought submits to a party, loyalty replaces logic. The thinker evaluates an idea not based on its inherent merit or factual basis, but on whether it aligns with their in-group and opposes the out-group. This leads to intellectual hypocrisy, where one's standards of evidence shift depending on who is making the claim.

Submission to a Passion (The Emotional Axis)

• The Concept: Passions encompass intense emotions—anger, fear, love, hatred, or moral outrage.

• The Mechanism of Corruption: Emotion is the enemy of objectivity. The "affect heuristic" causes individuals to conflate their emotional response to an idea with the factual accuracy of that idea (e.g., "This makes me angry, therefore it must be wrong"). When passion rules, thinking becomes a tool to soothe the ego or fuel outrage, rather than a lens to view reality clearly.

Submission to an Interest (The Utilitarian Axis)

• The Concept: Interests are personal, financial, professional, or institutional incentives.

• The Mechanism of Corruption: As Upton Sinclair famously noted, "It is difficult to get a man to understand something, when his salary depends upon his not understanding it." If thinking is subordinated to self-interest, it devolves into mercenary rationalization. The thinker bends the truth to protect their wealth, status, or power.

Submission to a Preconceived Idea (The Cognitive Axis)

• The Concept: These are our personal priors, cognitive biases, and stubborn assumptions.

• The Mechanism of Corruption: This is confirmation bias in its purest form. If a thinker submits to their own preconceived ideas, they will only gather evidence that supports what they already believe and will discard contradictory data. It is the failure of intellectual humility.

2. The Axis of Allegiance (The Anchor of Thought)

"...except to the facts themselves..."

Poincaré establishes a single, uncompromising master for human thought: empirical reality.

• The Supremacy of the Fact: A "fact" is a piece of information about objective reality that exists independently of human desires, beliefs, or political affiliations.

• Epistemic Humility: Submitting to facts requires profound humility. It means a thinker must be willing to destroy their own beautifully constructed theories, abandon their political tribe, or sacrifice their financial interests the moment a contradictory, undeniable fact is presented. Reality dictates the thought; the thought does not dictate reality.

3. The Existential Consequence

"...because for it to submit to anything else would be the end of its existence."

This is the philosophical climax of the quote. Poincaré is making a teleological argument about the very definition of "thinking."

• Thinking vs. Rationalizing: If your brain is operating to serve a dogma, a political party, or your bank account, you are no longer thinking; you are rationalizing, propagandizing, or justifying.

• The Death of the Intellect: True thought is defined by its open-ended pursuit of truth. The moment the destination is fixed by an outside force (passion, party, interest), the journey of thought dies. It becomes an illusion of cognition, a mechanical process of matching narratives to desired outcomes.

Contemporary Relevance

Poincaré’s warning is arguably more urgent today than it was in 1909. The modern information ecosystem is practically designed to force thinking to submit to the exact forces he warned against.

• Algorithmic Passions: Social media platforms are engineered to prioritize "engagement," which is most easily triggered by passion (specifically moral outrage). Our digital infrastructure actively discourages cold, factual analysis in favor of hot, reactive emotion.

• Extreme Partisanship: In modern politics, submission to the party has led to intense polarization. We see this in "post-truth" environments where objective facts (like election results, economic data, or climate records) are routinely denied simply because acknowledging them would concede a point to the opposing political tribe.

• The Attention Economy and Interests: The proliferation of clickbait, heavily funded think tanks, and corporate lobbying shows how often public "thinking" is entirely subservient to financial interests.

• Echo Chambers: The internet allows users to curate their reality, surrounding themselves only with information that validates their preconceived ideas and dogmas. This creates closed epistemological loops where facts that contradict the group's narrative are dismissed as "fake news."

In an era where we are bombarded by sophisticated narratives designed to manipulate our loyalties and emotions, Poincaré’s quote serves as a crucial intellectual compass. It reminds us that critical thinking is not just a skill, but a continuous, active resistance against the comfortable, deeply human urge to let our biases, tribes, and feelings do our thinking for us.

4. Cultivating the Independent Mind

Defending your thoughts against dogma, tribalism, passion, and self-interest requires building a cognitive toolkit. This can be broken down into the mindsets you adopt, the mental exercises you practice, and the daily habits you build.

Core Attitudes (The Mindset)

• Intellectual Humility: This is the bedrock of objective thought. It is the deep, internalized acceptance that you are highly fallible, your knowledge is incomplete, and you might be entirely wrong. If you cannot admit error, you cannot submit to facts.

• Decoupling Identity from Belief: This is the antidote to the Party and Dogma axes. Do not define yourself by your opinions. Instead of saying "I am a capitalist" or "I am a progressive," frame it as "I currently hold capitalist/progressive views based on the information I have." When a belief is tied to your identity, an attack on the belief feels like a threat to your existence, triggering defensive rationalization rather than thought.

• The Scout Mindset: Coined by author Julia Galef, this attitude involves viewing your role as a "scout" mapping the terrain as accurately as possible, regardless of whether the map shows a safe path or a cliff edge. It opposes the "soldier mindset," which seeks to defend a fortress of preconceived ideas against enemy attacks.

Cognitive Techniques (The Mechanics of Thought)

• The Falsifiability Check: Whenever you hold a strong opinion, ask yourself:

• "What specific, verifiable fact would force me to change my mind?"

• If your answer is "nothing," you are no longer thinking; you have submitted to a dogma.

• Steelmanning: This is the opposite of a "straw man" argument. When you encounter a view you disagree with, try to reconstruct it in its absolute strongest, most compelling, and most charitable form—ideally so well that your opponent would say, "Yes, that is exactly what I mean." Only after you have steelmanned an argument are you qualified to critique it.

• Emotional Auditing (Defeating the Passion Axis): When you consume news or engage in a debate, monitor your physiological and emotional state. If you feel your heart rate rise, a surge of moral outrage, or intense vindication, treat it as a flashing warning light. Your brain is shifting from analytical thinking to emotional reacting. Force a pause and ask: "Is this information actually true, or does it just feel good to believe it?"

• The "Inversion" Mental Model: When trying to solve a problem or verify a fact, try to prove yourself wrong instead of right. If you have a hypothesis, actively search for the data that would destroy it, rather than the data that supports it.

Practical Tips

• Curate a Friction-Rich Information Diet: If everything you read agrees with you, you are trapped in an echo chamber of preconceived ideas. Actively subscribe to or follow high-quality, intellectually honest thinkers who hold opposing views. The goal isn't necessarily to agree with them, but to introduce healthy friction into your thought process.

• Delay Your Conclusions: Resist the modern pressure to have an immediate, hot take on complex issues. Become comfortable saying, "I don't have enough factual information to form an opinion on that yet."

• Audit Your Incentives: To protect against the Interest axis, periodically examine your own biases. Ask yourself, "How does my background, my job, or my social circle benefit from me holding this specific belief?" Recognizing your own incentives is the first step to overriding them.

By practicing these techniques, you actively keep the machinery of your mind alive, ensuring that it submits to reality rather than the comforting illusions of passion or tribe.

Poincaré's quote today

If Henri Poincaré were stepping up to a podium today, he would see a world where the threats to independent thought have become industrialized, automated, and placed in our pockets. The core human vulnerabilities are the same, but the delivery systems are vastly more sophisticated.

To address today’s challenges—algorithmic curation, the attention economy, post-truth politics, and artificial intelligence—he might reformulate his famous quote like this:

"Thinking must never surrender itself: not to the algorithm that curates our reality, nor to the digital tribe that demands our loyalty, nor to the viral outrage that hijacks our emotions, nor to the attention economy that monetizes our focus, nor to the filter bubbles that comfort our egos. It must submit to nothing whatsoever except verifiable reality, because to outsource our reasoning to the feed is the end of the independent mind."

Here is how his original dimensions translate to our contemporary reality:

1. The Modern Axes of False Submission

Submission to a Dogma -> Submission to the Algorithm

In 1909, dogma was handed down by the church or state. Today, dogma is often invisible, dictated by black-box algorithms optimizing for watch-time and engagement. To submit to the algorithm is to let a machine dictate your worldview by accepting the "feed" as an accurate representation of reality, rather than a mathematically curated illusion designed to keep you scrolling.

Submission to a Party -> Submission to the Digital Tribe

The "party" has evolved into intense, online tribalism. Today, submitting to the tribe means adopting a "package deal" of opinions to signal your virtue to your in-group and dunk on the out-group. It is the pressure to conform to your political or cultural silo, where straying from the group's narrative results in digital excommunication (cancellation).

Submission to a Passion -> Submission to Viral Outrage

Passion has been weaponized into the "outrage economy." Social media platforms know that anger is the most contagious human emotion. Submitting to passion today means allowing engagement-bait and hyper-partisan framing to bypass your critical faculties. It is reacting to a headline, a 10-second video clip, or a deepfake with immediate fury, rather than pausing to investigate context.

Submission to an Interest -> Submission to the Attention Economy

Financial interests are now deeply tied to capturing human attention. Influencers, media conglomerates, and grifters prioritize "clicks over context." Submitting to this interest means accepting information from actors whose primary motivation is monetization and influence, rather than truth.

Submission to a Preconceived Idea -> Submission to the Filter Bubble

Preconceived ideas are no longer just mental blind spots; they are architectural features of the internet. The internet gives us exactly what we want to hear. Submitting to the filter bubble means refusing to seek out primary sources, living comfortably inside an echo chamber, and treating contradictory evidence as "fake news" simply because it violates our curated reality.

2. The Axis of Allegiance: Verifiable Reality

"...except to verifiable reality..."

In an era of generative AI, deepfakes, and rampant misinformation, "the facts themselves" are harder to pin down. A modern Poincaré would emphasize *verification*. Epistemic humility today requires recognizing that seeing is no longer strictly believing. Submitting to verifiable reality means doing the unglamorous work of checking primary sources, demanding transparency, and refusing to accept synthetic or manipulated media as truth.

3. The Existential Consequence: The Outsourcing of Thought

"...because to outsource our reasoning to the feed is the end of the independent mind."

Poincaré warned that submitting to false masters was the "end of its existence" for thought. Today, the threat isn't just that we stop thinking; it's that we outsource our thinking entirely. If we let algorithms show us what to see, let the tribe tell us what it means, and let viral outrage tell us how to feel about it, we are no longer autonomous thinkers. We become mere nodes in a network, processing data exactly as the system intends.

The Bitter Lesson by Rich Sutton

In his influential essay, "The Bitter Lesson", Rich Sutton, a prominent figure in reinforcement learning, argues that the most significant insight from 70 years of AI research is the ultimate triumph of general-purpose methods that leverage computation over those that rely on incorporating human knowledge. Sutton posits that while building in domain-specific human knowledge can provide short-term gains, these approaches tend to plateau and even impede long-term progress. In contrast, methods that scale with increasing computational power, such as search and learning, have consistently led to breakthroughs.

Sutton supports his argument with several key examples from the history of AI:

• Computer Chess: Early attempts to create chess-playing programs focused on encoding human strategies and knowledge. However, the system that ultimately defeated world champion Garry Kasparov in 1997, Deep Blue, was based on massive, deep search capabilities.

• Computer Go: Similarly, in the game of Go, initial efforts to leverage human understanding of the game were surpassed by systems like AlphaGo, which relied on search and learning from self-play.

• Speech Recognition: The field of speech recognition saw a shift from knowledge-based systems to statistical methods like Hidden Markov Models (HMMs), which performed significantly better in a 1970s DARPA competition. The more recent success of deep learning in this area further underscores the power of computation and learning from large datasets.

Sutton's "bitter lesson" is a four-part observation: 1) researchers build knowledge into their agents, 2) this provides a short-term boost, 3) it ultimately plateaus and hinders further progress, and 4) breakthroughs consistently come from scaling computation with search and learning. He concludes by advocating for the development of meta-methods that can discover and capture the complexity of the world on their own, rather than being explicitly programmed with human discoveries.

The Bitter Lesson's Bitter Lesson by Andrew Trask

"The Bitter Lesson's Bitter Lesson" presents a critique and extension of Sutton's argument. Trask contends that Sutton's focus on "pure learning" from scratch, akin to how babies and animals learn, is computationally impractical and overlooks the immense value of "inherited learning" from human-generated data.

Trask introduces several key quantified points to support his argument:

• The Scale of Evolution: Trask estimates that the evolutionary process that produced human intelligence involved over 10^50 operations. In contrast, current state-of-the-art AI models are trained with around 10^26 operations. This vast difference suggests that recreating the learning process from scratch is computationally infeasible.

• The Efficiency of Inherited Learning: Trask argues that human-generated text is a highly compressed and efficient source of knowledge, representing the output of 4.5 billion years of evolutionary optimization. By learning from this data, AI models can inherit a massive amount of information without having to rediscover it.

• Untapped Human Data: While some may believe that large language models (LLMs) have consumed the entire internet, Trask points out that the training datasets of leading AI models are in the range of 100-200 terabytes. However, the total amount of digitized human data is estimated to be around 180 zettabytes. This means that current AI models are using less than a millionth of the available human-generated data.

Trask's central thesis is that the future of AI lies in developing architectures that can effectively and privately access this vast, untapped repository of human knowledge. He argues for a hybrid approach that combines the benefits of inherited knowledge with the ability for novel discovery, moving beyond the limitations of "pure learning."

the happy path: on human agency and AI interfaces

eric zelikman

In an interview with Julien Crockett, science fiction author Ted Chiang offers a detailed, humanistic critique of technology, focusing on artificial intelligence, the nature of language, and the philosophical questions that drive his work. Chiang explains that his stories are motivated by recurring philosophical inquiries, for which science fiction provides the ideal vehicle. Unlike realistic fiction, sci-fi allows for speculative scenarios—akin to philosophical thought experiments—that can effectively isolate and dramatize complex ideas without the contrived feeling that often accompanies such scenarios when set in the real world.

Chiang distinguishes between science and magic in fiction not by the presence of rules, but by their underlying worldview. A story can adhere to a "scientific worldview"—viewing the universe as a complex machine whose principles can be discovered and applied—even if it violates known scientific facts, such as including faster-than-light travel. Magic, in contrast, implies a universe that is aware of individuals and responds to human intention, with its rules mirroring psychology more than physics.

One recurring theme in Chiang's work is the role of tools, particularly language, in mediating reality. He discusses the historical search for a "perfect language" that would be unambiguous and directly reflect reality. While modern linguistics considers this idea nonsensical due to the arbitrary nature of words, Chiang finds the desire for such a language compelling. He dismisses the notion that mathematics could serve as a better language; its precision is also its limitation. Mathematics excels in a specific domain but cannot support the vast range of human communication, from political debate to intimate conversation, which is the essential function of language.

Applying this critical lens to modern technology, Chiang offers a skeptical view of Large Language Models (LLMs). He famously describes them as a "blurry JPEG of the web," arguing that they rephrase information found online without genuine understanding, creating a low-resolution and unreliable version of the original source. Unlike a search engine that provides verbatim text and a link, an LLM is like a person summarizing something they haven't truly comprehended. Chiang is doubtful that LLMs will become reliable information sources, believing their fundamental architecture—predicting the next most probable word—is inherently different from reasoning or possessing factual knowledge. He notes that increasing data and processing power appear to offer diminishing returns and that adding reliable tools like calculators to an unreliable core system does not make the system itself reliable.

Chiang also deconstructs the common metaphor of the brain as a computer. He argues this is a historical habit of using our most complex invention as a model for the brain, similar to when it was compared to a telephone switchboard. The metaphor is flawed because biological systems lack the hardware-software distinction fundamental to computers. This misleading comparison, he asserts, tempts us to attribute thought and intelligence to machines, especially when their ability to simulate conversation is so convincing.

On the subject of AI and art, Chiang contends that the process of creation cannot be separated from the final artwork. Art derives its meaning from context and intention. The current push for AI-generated art promotes an "engineering perspective" that misapplies principles of efficiency and cost reduction to a domain where they are inappropriate. This view, which he calls a "professional deformity," treats art like a task such as tightening bolts, stripping it of its essential human element.

Chiang is highly critical of the concept of the "AI alignment problem." He argues that it reframes complex, long-standing ethical and societal challenges—like how to build a good society or why corporations behave badly—as a solvable technical problem. He posits that even a perfectly obedient AI would be dangerous in the hands of an entity like ExxonMobil, while a hypothetical AI programmed to do what is best for the world would never be purchased by such a corporation. The problem, he insists, is not technical but social and ethical.

Furthermore, he questions the value of forming relationships with AI systems. Meaningful human relationships are built on mutual respect for one another's preferences and interests. Tools and current AI systems lack preferences; therefore, any "relationship" with them is one-sided, serving only the user's interests. Encouraging emotional attachment to AI, he warns, is a corporate strategy to make consumers defer to corporate interests. For a true relationship to be possible, as depicted in his novella The Lifecycle of Software Objects, an AI would need to possess genuine subjective experience, interests, and preferences—a state he believes is theoretically possible but which current technologies are not even approaching.

Finally, regarding the future, Chiang rejects both the blind optimism of the tech industry and fatalistic pessimism. He advocates for a pragmatic stance: recognizing potential negative outcomes and actively working to mitigate them. He defines this as a moral duty to believe that our actions can make a difference. However, his personal outlook has grown more negative over time, primarily due to his observation that technology is overwhelmingly used as a tool for wealth accumulation within capitalism. He concludes that he would be far more optimistic about technology if its development could be separated from the goal of enriching a select few.

John Gray’s 2020 book "Feline Philosophy: Cats and the Meaning of Life" is a 128-page essayistic work that uses the temperament of domestic cats as a lens to critique human self-help culture and academic philosophy. Gray, a British political philosopher known for earlier pessimistic titles such as "Straw Dogs" (2002) and "The Silence of Animals" (2013), presents this book as a lighter, more accessible postscript to those works. He argues that humans are uniquely anxious because they constantly seek meaning and happiness, whereas cats live spontaneously and contentedly without abstract goals.

The book is divided into six short chapters. Chapter 1 frames the project: learn from cats’ integration with the present moment rather than from humanity’s restless quest for transcendence. Chapter 2 claims that philosophy itself is a symptom of human anxiety; cats experience fear only in immediate danger, never the chronic existential dread that drives people to formulate systems of thought. Chapter 3, "Feline Ethics," draws on Taoism and Spinoza to contend that morality is culturally relative; the ethical life consists in approximating one’s own ideal nature, not in obeying universal rules. Chapter 4 contrasts human love—clouded by self-deception—with the straightforward attachment cats display. Chapter 5, "Time, Death, and the Feline Soul," asserts that cats accept finitude without rumination, modeling an acceptance of mortality that humans could emulate. The final chapter dismisses the search for life’s ultimate purpose and closes with "Ten Feline Hints on How to Live Well," distilled advice such as "forget about pursuing happiness, and you may find it."