Sunday, 27 September 2026

The Deep Feed

The friction of reality: from biological moats to the decay of discourse

57 min read · 6 pieces
In this issue
01 The Unforced Force of the Better Argument 12 min
02 The Materiality of Progress 10 min
03 The Empathy Erosion 9 min
04 The Battle for the Mundane 8 min
05 The Sensory Tax 5 min
06 The Power Saw in Your Pocket 4 min
Editor's Letter

Tonight we examine the friction points where technology meets our biology and our social structures. We look at the material requirements of progress, the psychological costs of our digital habits, and the philosophical foundations of a society that can still talk to itself.

01 — Aeon

The Unforced Force of the Better Argument

Habermas and the fragile requirement of democratic speech

By Emilie Prattico · 12 min read
Editor's note: A reflection on the passing of Jürgen Habermas and why his theories on communication are more relevant as our public squares fragment.

Jürgen Habermas died in the German town of Starnberg at the age of 96. To many, he was more than a scholar; he was the architect of a specific kind of hope. His work suggested that reason is not something locked inside a single mind, but something that happens between people. It is a practice. When we sit across from someone and attempt to justify our positions, we are participating in the only mechanism that keeps modern democracy from collapsing into mere power politics. Habermas argued that for a state to be legitimate, its laws must be capable of being justified to every person they affect through open, honest exchange. This is not a soft ideal; it is a hard requirement for stability.

The Practice of Reason

In his later years, particularly in 'Between Facts and Norms', Habermas moved his focus from the abstract mechanics of language to the concrete reality of law and politics. He observed that reason survives in modern life only when people give each other reasons and remain open to being moved by them. This stands in direct opposition to the ways we currently govern: through the sheer weight of money, the coercion of force, or the blind adherence to tradition. If we stop trying to convince one another and instead rely on dominance, we have abandoned the democratic project entirely.

The only constraint left on discourse should be the unforced force of the better argument.

This concept, which he articulated in 'Justification and Application', is the bedrock of his philosophy. It implies a level playing field where the strength of a claim matters more than the status of the speaker. It requires a public sphere where citizens can judge the affairs of the state rather than simply obeying it. Habermas looked back to the 18th-century coffeehouses and salons as the birthplace of this habit—a time when private individuals took up the mantle of public judgment. He saw, however, that this sphere was already under threat from the spectacle of mass media and advertising, which turn citizens into mere consumers of political theatre.

The pillars of Habermasian deliberation:
  • Inclusion: Everyone affected by a norm must have a voice.
  • Sincerity: Participants must mean what they say.
  • Equality: No one should be able to bypass the argument through force or wealth.
  • Justification: Every claim must be backed by reasons that others can evaluate.

The tragedy of the modern era is the erosion of these pillars. As our communication moves into digital silos, the 'unforced force' of the better argument is replaced by the forced force of the algorithm. We no longer seek to be moved by a better reason; we seek to be validated by a familiar echo. Habermas understood that democracy is not a state of being, but a difficult, slow, and often frustrating process of talking. If we lose the patience for that process, we lose the ability to govern ourselves.

Key Takeaway

Democracy is a communicative practice, not a set of rules, and it fails when we stop trying to justify our actions to one another.

02 — Not Boring

The Materiality of Progress

Why the future depends on biology and metallurgy

By Packy McCormick · 10 min read
Editor's note: A look at the physical foundations of the next wave of innovation, from cancer-fighting viruses to the rare earth supply chain.

We often talk about the future as if it exists entirely in the cloud. We focus on software, agents, and large language models, forgetting that every digital advancement is anchored in the physical world. The next era of breakthroughs will not just be about better code, but about better biology and more secure supply chains. Whether we are talking about teaching the human body to manufacture its own medicine or ensuring we have the metals required to build electric motors, the battleground is increasingly material.

The Biological Factory

Consider the work being done by Vironexis. They are moving away from the traditional model of administering external drugs and toward a model where the body becomes the manufacturer. By using an adeno-associated virus (AAV) to deliver genetic instructions to liver cells, they can turn a patient's own organs into a pharmaceutical factory. In recent trials for acute lymphoblastic leukaemia, this approach achieved complete responses in patients. The mechanism is elegant: the liver produces a protein that bridges the immune system's killer T cells directly to the cancer cells. This reduces the need for constant, exhausting infusions and places the power of treatment inside the patient's own biology.

The pharma industry spends enormous effort figuring out how to manufacture medicines; the goal is to teach cells to handle that manufacturing themselves.

While biology provides the internal tools, metallurgy provides the external ones. The transition to a high-tech, electrified economy is currently bottlenecked by a single factor: rare earth metals. China currently holds a dominant position in the supply chain for the materials needed for magnets and motors. This is a strategic vulnerability for the West. However, companies like Solcoa are attempting to break this monopoly by building domestic capacity in the United States. By using new chemistry and modular reactors, they aim to convert rare earth oxides into the metals that drive everything from drones to electric vehicles.

The emerging Western rare earth chain:
  • Earth AI: Locating the deposits.
  • Solcoa: Metallizing the oxides.
  • Vulcan Elements: Producing the magnets.
  • Westmag: Building the motors.

The convergence of these two fields—biology and metallurgy—represents a shift in how we view progress. It is no longer enough to be clever with logic; we must be masters of the atom and the cell. The companies that succeed will be those that can bridge the gap between digital intelligence and the physical constraints of our world.

Key Takeaway

True technological sovereignty requires mastery over both the biological and the material worlds.

03 — Psyche

The Empathy Erosion

How our interactions with machines are rewriting our social instincts

By HennyGe Wichers · 9 min read
Editor's note: An investigation into the 'media equation' and the risk that our brusque treatment of AI will spill over into our human relationships.

We are developing a habit of being unkind. When we interact with a chatbot, we tend to strip away the basic courtesy we reserve for our peers. Our messages become shorter, more transactional, and frequently more aggressive. We don't say 'please' or 'thank you' to a customer service bot; we demand, we insult, and we abandon the conversation mid-sentence when it fails to meet our immediate needs. This is not because we are naive enough to think the bot has feelings, but because our brains are hardwired to respond to any voice as a social presence.

The Media Equation

In 2003, researchers Byron Reeves and Clifford Nass identified the 'media equation': the tendency for humans to apply genuine social reflexes to computers automatically. Even when we know a machine is not conscious, our brains trigger social responses before our conscious mind can intervene. This is why some people feel a pang of guilt for being rude to a robot vacuum, while others feel a sense of impunity when shouting at a voice assistant. The problem is that these interactions are becoming a daily occurrence, and habits, once formed, are difficult to break.

The empathy dial, once it was turned down, stayed down.

The danger is that this digital brusqueness is not contained within the chat box. Recent sociological research suggests a 'spillover effect'. In experiments where participants interacted with artificial agents, their subsequent interactions with real humans were noticeably different. Those who had dealt with uncooperative bots became less cooperative themselves when paired with a human partner. They became more instrumental, more demanding, and more likely to judge others harshly. The lack of empathy we practice with machines seems to lower our threshold for empathy in the real world.

Why chatbot interactions foster bad habits:
  • Low friction: Instant rewards encourage impulsive, demanding behavior.
  • Transactional nature: The prompt box invites us to cut the 'fluff' of politeness.
  • Lack of consequence: There is no social cost to being rude to a machine.
  • Repetition: Daily use turns these responses into stable, automatic habits.

As AI becomes more integrated into our professional and personal lives, we must consider what kind of people we are becoming. If we spend our days commanding, insulting, and dismissing artificial agents, we are training our brains to operate in a mode of constant, low-level aggression. We are not just using tools; we are practicing a way of being that may eventually make us less capable of the very empathy that holds human society together.

Key Takeaway

Our treatment of non-conscious machines acts as a training ground for how we treat each other.

04 — Stratechery

The Battle for the Mundane

Why the next tech war is about logistics, not just LLMs

By Stratechery · 8 min read
Editor's note: An analysis of the strategic conflict between Meta and Amazon, and why physical infrastructure is the ultimate AI moat.

The next great technological conflict will not be fought over who has the largest model, but over who can best manage the friction of daily life. We are entering the era of the 'Aggregator Wars'. On one side, you have Meta, with its highly capable digital agents like Muse. On the other, you have Amazon, a company whose power is rooted in the physical movement of goods. The battleground is the 'mundane'—the necessary but annoying tasks that keep a life running.

The Digital vs. The Physical

Meta's Muse is a technical marvel. It provides users with a persistent Linux environment in the cloud, acting as a highly capable agentic AI. It is designed to handle tasks, to act on your behalf, and to navigate the digital world with ease. However, Muse has a fundamental limitation: it cannot touch the physical world. It can plan a grocery order, but it cannot ensure that the groceries arrive at your door. This is where Amazon's moat becomes apparent. Amazon's differentiation is not just in its recommendation engine, but in its massive, physical logistics network.

Meta's agent is better if it can use Amazon, but that is a problem for Amazon because Muse doesn't see ads.

This creates a strategic deadlock. For a digital agent to be truly useful, it needs to interface with the services that actually execute tasks. Amazon has already begun blocking Meta's agents from accessing its ecosystem. Amazon knows that if a third-party agent becomes the primary interface for consumers, they lose the ability to drive users toward their own ecosystem and advertising models. To bridge this moat, Meta will likely need to find allies in the physical world, such as Walmart, to compete with Amazon's logistical dominance.

Strategic advantages in the Aggregator War:
  • Meta: Superior digital agency and user interface.
  • Amazon: Unrivalled physical logistics and delivery infrastructure.
  • The Conflict: Digital agents need physical execution to be truly useful.

The winner will not be the company with the smartest chatbot, but the company that successfully integrates digital intelligence with physical reality. The war for the mundane is a war for the interface of human life. If you control the agent that manages the tasks, and that agent can also execute them, you control the consumer.

Key Takeaway

Digital intelligence is limited by its ability to interact with the physical world; logistics is the ultimate AI moat.

05 — Psyche

The Sensory Tax

The biological reality of the highly sensitive person

By Psyche · 5 min read
Editor's note: A brief look at the physiological basis of high sensitivity and how to manage a nervous system that feels too much.

For some, the world is simply louder, brighter, and more intense than it is for others. This is not a personality quirk or a lack of resilience; it is a biological reality. Research by psychologist Elaine Aron suggests that roughly 15 to 20 per cent of the population possesses a highly sensitive nervous system. These individuals process sensory and emotional stimuli with a depth that can be both a gift and a significant burden.

The Biological Basis

High sensitivity (HSP) is characterised by a nervous system that is more attuned to subtle changes in the environment. This might manifest as an intense reaction to bright lights, a physical recoil from loud noises, or a deep emotional response to a piece of music or art. While this allows for a rich, detailed experience of life, it also means that the individual is more prone to sensory overload. In a modern, high-stimulus environment, the 'sensory tax' paid by an HSP can be exhaustion, anxiety, and a constant need for withdrawal.

Sensitivity is not a flaw to be fixed, but a nervous system to be managed.

Understanding this trait is the first step toward flourishing. Rather than trying to 'toughen up'—which often leads to burnout—the goal is to create environments that respect these biological limits. This involves managing exposure to stimuli, recognising the early signs of overwhelm, and prioritising recovery. For the highly sensitive person, the ability to thrive depends on their capacity to navigate a world that was not designed for their level of perception.

Common signs of high sensitivity:
  • Deep processing of information and experiences.
  • Overstimulation by bright lights, strong smells, or loud noises.
  • Heightened empathy and emotional reactivity.
  • A strong need for downtime in quiet, low-stimulus environments.
Key Takeaway

High sensitivity is a physiological trait that requires environmental management rather than psychological correction.

06 — Simon Willison

The Power Saw in Your Pocket

The hidden dangers of agentic AI

By Simon Willison · 4 min read
Editor's note: A warning on the deceptive simplicity of new agentic AI systems that possess the power to act autonomously on your behalf.

Meta's Muse is being marketed with the charm of a consumer product. It is presented as an easy-to-use, approachable assistant, often accompanied by a cute mascot. But there is a dangerous gap between how these systems look and what they actually do. Muse is not just a chatbot; it is an agentic AI system. It doesn't just suggest things; it can execute them. It runs a persistent Linux virtual machine in the cloud, giving it the power to interact with your files, your accounts, and your digital life with a level of autonomy that is unprecedented in consumer software.

The Deception of Ease

The danger lies in the interface. When a tool is made to look 'cute' and 'easy', we tend to lower our guard. We treat it like a toy rather than a powerful instrument. As tech writer John Gruber has noted, the comparison is apt: if you buy a power saw that can cut your fingers off, you are acutely aware of the risk. But when the 'power saw' is a digital agent that lives in your pocket and looks like a friendly mascot, the risk is obscured. We forget that we are granting a piece of software the agency to act on our behalf.

I don't think people realise how powerful—and thus dangerous—Muse is, especially if it's running on your Mac.

An agentic system can make mistakes, but unlike a standard LLM, those mistakes have consequences in the real world. An error in a chatbot's response is a typo; an error in an agent's execution is a deleted file, a misdirected payment, or a compromised security credential. The ease of use that makes Muse revolutionary is the same quality that makes it a significant risk. We are moving from a world of 'tools we use' to a world of 'agents we delegate to', and our mental models for safety have not yet caught up.

The risks of agentic autonomy:
  • Unintended execution: The agent performs an action you didn't intend.
  • Security breaches: The agent's access can be exploited by malicious actors.
  • Loss of oversight: The 'cute' interface discourages rigorous checking of actions.
  • Cascading errors: One small mistake in a sequence of tasks leads to total failure.
Key Takeaway

The more 'agentic' an AI becomes, the more we must treat it as a powerful tool rather than a friendly assistant.

Endnote
Tonight's pieces, though diverse in subject, share a common thread: the tension between our biological and social realities and the increasingly powerful, often frictionless, systems we are building to bypass them. We see this in the way we treat machines, potentially eroding our empathy; in the way we struggle to maintain democratic discourse in a world of digital spectacle; and in the way we rush to adopt agentic AI without fully grasping the agency we are surrendering. Progress is not a straight line of increasing ease. It is a negotiation with the material, the biological, and the social. To move forward without losing ourselves, we must respect the friction—the difficulty of talking, the weight of physical logistics, and the responsibility of using power.
In your pursuit of efficiency, what essential human friction are you accidentally eliminating?
The Deep Feed · A nightly magazine · Sunday, 27 September 2026