Baltimore Orioles vs Chicago Cubs.

Today's MLB Game of Choice has the Baltimore Orioles playing the Chicago Cubs, and has a scheduled start time of 5:35 PM CDT. As I usually do, I'll follow the game's score and stats in real time via MLB's Gameday Service where we can also find links to the radio-call of the game provided by announcers of either team we choose.

And the adventure continues.

Now that I’m done with all three ebooks of The Package all I need to do is work on the paperback. So far, it’s on hold because my book cover designer is currently making mine. Sketch design, blurb, and all the necessary stuff are given to him.

Projected completion date of the book cover is around two weeks minimum, a month maximum. So I have plenty of time to work on other projects like this blog, newsletter, and drafting an upcoming nonfiction ebook. Will give you more details later.

Right now, I want some rest.

#writing #blog #bookcover #break #ebooks #newsletter #nonfiction #rest

Some weeks present a real challenge with their lack of material to write about. I've read virtually nothing since last Tuesday; and have heard almost no new music. It hasn’t been entirely without incident – just not the sort of thing I’m keen to get into here.

I’ll mention a book I acquired a couple of months ago: one more to be looked at than read. This was Laurie Lipton Drawing (2022). Lipton, for those unaware of her work, is an artist who creates minutely-detailed, large-scale pictures purely in pencil and graphite. I’ve been an admirer of hers for almost twenty years (I wrote a little about her at my original blog in ‘07). The book (Fig. 29) is a survey of the artist's work between 2014 and 2022. It falls into three somewhat overlapping thematic sections, all more or less inspired by the news of the day: ‘May You Live in Interesting Times’ (about the pandemic); ‘Post Truth’ (about Trumpian populism and the social media landscape underlying it); and ‘Techno-Rococo’ (about on-line life supplanting ‘real’ life). As well as full reproductions of the drawings there are ‘close-ups’ of selected details (e.g. Fig. 30) and some photographs of Lipton at work.

---

Among the recent-ish additions to my jazz collection was Alligator Bogaloo (1967) by a quintet led by the alto saxophonist Lou Donaldson. I picked up a late-‘90s Japanese CD issue of it for a fiver. I was familiar with the album’s infectious title-track from its inclusion on an ‘80s Blue Note compilation album. I hadn’t been aware, however, of the track’s origin, as a piece apparently rustled up out of thin air by Donaldson et al to fill out a reel of tape at the end of a recording session. The rest of the record is new to me. Some of it I’ll admit I’m lukewarm about: the second track, ‘One Cylinder’, for instance, rather outstays its welcome over its six-and-a-half-minute duration. Overall though it’s an enjoyable listen.

I had meant to order a new CD this week. Having seen some positive reviews of Spontaneous Music Live, the new offering by the contemporary jazz quintet SML, I took myself to Bandcamp to place an order. I was disappointed to find it only existed as a download, on vinyl, or (already sold out) a limited edition cassette. I may or may not eventually get the LP. While thinking about it, I can always watch them performing ‘The Drums’ at YouTube.

---

The cheese of the week has been good old camembert. My camembert of choice lately has been the Aldi ‘Specially Selected’ variety. Other supermarket offerings I’ve tried in recent months have been sold very unripe, whereas the Aldi one seems to me more flavourful than most straight off the shelf, and all the better after another week or so in the fridge.

I've launched small interface update to Seedship. It should be live on itch.io, the App Store and Google Play.

1.4.0 patch notes:

• Changed layout code to hopefully fix the bug where the status bar could appear in the wrong place (I think this was only happening in the iOS app).
• Layout now adapts to phone screens with inset cameras.
• Moved the sharer page to spacegoblingames.com and revised it slightly. (But also left it up at the old URL so old links will still work.)
• The new sharer page deals with invalid query strings by giving a custom message rather than ugly red errors all over the screen.

Old saved games and high scores should carry over, but I can't guarantee it so sorry if they don't.

I've also removed the downloadable Windows and Mac apps from itch.io, but I've replaced them with a downloadable version of the HTML file so you can download it and play it in your browser locally.

Note on the philome.la version

When I first released Seedship in 2017, it was on philome.la, a free hosting site for Twine games. Later on I moved it to its current home on itch.io, and also made the Android and iOS apps, but a lot of old links to Seedship you can find online still point to the philome.la version.

philome.la was discontinued and became read-only in 2019, so that version of Seedship is still there, but I can no longer update it. This means it contains a few minor bugs that I've since fixed in the current version.

If you're playing Seedship online or sharing links to it, please play the version on itch.io! (https://spacegoblingames.itch.io/seedship) And in particular, if you find a bug in the philome.la version, please make sure you can reproduce it on the itch.io version, because there's a good chance it's been fixed.

#Seedship #bugfix

Rubrik von Pipiras: 5. Juli

Wir sind heute immer noch wie im Traum. Wir können einfach nicht glauben, dass Pipiras ganz allein diese große Straße überquert und das überlebt hat! Das schaffen nicht einmal alle selbstständigen Hunde.

Aber diese Woche zeigt uns Pipiras wirklich seine besonderen Fähigkeiten. Vielleicht hat er irgendwie gespürt, dass wir alle über seine hysterischen Reaktionen erzählt haben, und wollte nun beweisen, dass er eigentlich ein sehr vernünftiger Hund ist. Das ist uns übrigens schon am Freitag – beziehungsweise vorgestern – aufgefallen. Damals habe ich allerdings noch nichts darüber geschrieben.

Meine besondere Mannschaft – Mindaugas und unsere Hunde – muss manchmal ziemlich lange auf mich warten. Währenddessen verbringen sie viel Zeit im Auto. Vielleicht ist genau das der Grund, warum unsere beiden Assistenten inzwischen glauben, dass das Auto ihr eigentliches Zuhause ist. Wenn Mindaugas meint, dass das ständige Ein- und Aussteigen zu anstrengend wäre, bleibt er einfach im Auto und erledigt dort seine Arbeit am Laptop.

Also parkte er am Freitag irgendwo an der Spree und beschäftigte sich mit seinen Sachen. Plötzlich hörte er, wie Pipiras das Autofenster öffnete.

Das war allerdings keine Überraschung. Unsere klugen Jungs haben inzwischen wohl herausgefunden, wie die Fenstertaste in der Tür funktioniert. Nicht jeder Versuch mit ihren Pfoten gelingt, aber erstaunlich oft schaffen sie es tatsächlich, das Fenster selbstständig zu öffnen. Diesmal passierte allerdings noch etwas anderes. Kaum war das Fenster offen, hörte Mindaugas plötzlich ein dumpfes Geräusch – als wäre etwas aus dem Fenster gefallen. Es war Pipiras.

Mindaugas sah ihn sofort in Richtung Spree laufen und rief nach ihm. Doch Pipiras reagierte überhaupt nicht. Kurz darauf bemerkte Mindaugas allerdings, dass unser kleiner Held einfach nur ins Gras gegangen war, um sein Geschäft zu erledigen.

Ach, mein kleiner Junge!

Er wollte einfach nur kacken, wusste aber offenbar nicht, wie er uns das mitteilen sollte. Und weil er das Auto sauber halten wollte, hatte er kurzerhand diesen eigenen Plan entwickelt. Begemotas schaute Mindaugas dabei wohl nur fragend an: „Moment mal... dürfen wir jetzt wirklich einfach aus dem Fenster springen?“

Nachdem Pipiras fertig war, sprang er ganz selbstverständlich wieder ins Auto, als wollte er sagen: „Keine Sorge, niemand hat etwas bemerkt!“ Vielleicht hing also auch sein gestriger Ausflug mit genau dieser neu entdeckten Selbstständigkeit zusammen ...

Heute genießen wir jedenfalls einfach, dass wir alle zusammen sind. Ich war den größten Teil des Tages mit meinem Schreiben beschäftigt, deshalb verbrachten wir den Tag ganz ruhig.

Am Abend mussten wir allerdings noch zusehen, wie Litauen gegen Italien verlor. Ich hoffe wirklich, dass das nichts mit meinen misslungenen Basketballwürfen von gestern zu tun hatte! Aber ehrlich gesagt haben sie heute fast genauso geworfen wie ich. Früher hätte ich mich darüber wahrscheinlich geärgert. Heute allerdings nicht. Heute war ich einfach nur glücklich, dass wir alle zusammen waren.

Configured Once, or Compounding

Most of my tools are the kind you set up once.

You choose the settings, arrange them the way you like, and then they hold still. A note app. A calendar. The little collection of preferences that make a browser feel like mine. I set them, I forget them, and years later they behave exactly as they did on the first afternoon. There's a quiet comfort in that. Nothing drifts. Nothing surprises me.

But I've been noticing lately that “configured once” has a ceiling. A tool that holds perfectly still can only ever be as good as the day I set it up. It doesn't get worse, which is nice. It also doesn't get better, which I used to not think about at all.

The other kind of tool There's another kind of tool, and I've only recently started paying attention to it. Not the kind you configure and freeze — the kind whose work compounds. Where the first time you do a thing is the slow, careful time, and every time after that is a little quicker, a little surer, because it's building on the run before it rather than beginning again.

I've been letting this kind handle some of the repeating browser work I never wanted — the weekly rounds of checking and collecting that used to just be mine to do. I've been using AllyHub AI for it, teaching it a route once and then watching later runs lean on what the earlier ones already worked out. If the idea is useful to you, it's at AllyHub AI. I'm not reviewing it. I'm just noticing the shape of it, because the shape is new to me.

What the difference actually is What struck me wasn't the time it saved. It was the difference between a tool that stays exactly where I left it and a tool where the work I did the first time doesn't have to be done again. The first is a setting. The second is more like a path getting worn smooth by walking it.

I don't think one is better than the other. My frozen tools are frozen for good reasons; I don't want my calendar getting clever. But it's changed how I look at the boring, repeating parts of a week. Some of them I want to configure once and never touch. Some of them, I'm realizing, I'd rather hand to something that starts from scratch a little less each time.

Anyway. A small thought, on a quiet afternoon. Nothing to sell.

You live within a system you never signed a contract with. Every day, you make thousands of micro-decisions about how to behave, mostly without conscious thought. You pay an invoice on time, even if the supplier would never discover that you didn’t. You refuse to do business with someone who stiffed their last three partners, and you’d think twice about a colleague who didn’t. Nobody wrote these rules down. You absorbed them the way you absorbed grammar through exposure and correction.

A March 2026 paper from the Knight First Amendment Institute by Gillian Hadfield, Rakshit Trivedi, and Dylan Hadfield-Menell argues that this invisible social choreography is the core mechanism of democracy, not just an adornment. Furthermore, AI agents, such as those currently being developed to run businesses and manage supply chains, will undermine that mechanism unless they learn this dance too.

Democracy Is a Verb, Not a Document

The paper begins by challenging a comfortable assumption that many people accept without much scrutiny. Most view democracy as a collection of documents, institutions, constitutions, elections, and courts. The authors contend that this is roughly akin to describing a marriage solely through its wedding vows. While the vows matter, the true essence of a marriage lies in the thousands of everyday acts of compromise and occasional irritation that sustain cooperation over decades.

Hadfield, Trivedi, and Hadfield-Menell utilise a theoretical framework called “normative social order” to make this precise. In their model, a society’s actual norms are the product of an interactive system. People don’t follow rules because they are written down; they follow them because they observe others doing so and see how violations are punished. Punishments don’t need to be severe, just a disapproving look, a refusal to do business, or a sarcastic comment at a dinner party. These micro-sanctions generate the gravitational field that keeps behaviour in orbit.

This is where the paper borrows a term from evolutionary theory, “dancing landscapes.” The metaphor, from Stuart Kauffman’s work on complex adaptive systems, describes environments where multiple independent agents are constantly adjusting to each other’s behaviour. There is no central choreographer; the dance arises from the dancers' interactions.

What makes a norm sticky

The framework introduces a concept called a “classification institution,” which is any shared mechanism a group employs to decide which behaviours are punished and which are not. In small groups, this classification is entirely implicit, and you know what the group considers acceptable or unacceptable. Acceptability is judged by seeing who gets mocked and who gets praised. The Ju/’hoansi Bushmen, as anthropologist Polly Wiessner describes, regulate behaviour through evening conversations. Gossip and teasing around the fireside serve the same purpose as courtrooms and HR departments in modern societies.

As societies grow more complex, implicit classification cannot scale because the diversity of people and situations exceeds the reach of any informal consensus process. This creates a need for identifiable classification institutions; entities that can resolve ambiguity when community members disagree about acceptability. Courts, regulatory bodies, trade associations, and professional standards boards all serve this purpose in modern societies.

The paper argues that for these institutions to be effective, they need attributes that closely match what legal philosophers have long called “the rule of law,” namely stability, clarity, generality, and neutrality. The twist is that Hadfield and her co-authors do not derive these attributes from abstract principles. Instead, they derive them from game theory. An institution with those attributes is one around which independent actors can reliably coordinate, and coordination is what sustains the entire system.

Enter Adam Smith’s imaginary friend

The paper revisits Adam Smith’s “impartial spectator” from The Theory of Moral Sentiments and uses it as a model for how AI agents could participate in democratic societies without causing harm. Smith argued that moral reasoning works because each of us carries a mental image of a neutral observer—an internal referee—who judges our behaviour against community standards. You do not avoid bribery because you have memorised a specific anti-corruption law; you avoid it because your internal impartial spectator would wince.

This is the cognitive capacity that Hadfield, Trivedi, and Hadfield-Menell call “normative competence.” It goes beyond simply knowing the rules. It involves the ability to interpret a constantly changing normative environment, anticipate how your community will respond to specific actions, and adjust your behaviour accordingly. The key point is that it also requires predicting how the rules themselves will change, since in any living democracy, they constantly do. Yesterday, you didn’t need to worry about data privacy in your marketing. Today, GDPR and its equivalents are everywhere, and community expectations have shifted beneath you.

Why this matters even if you’ve never read about game theory

If AI agents were merely chatbots answering questions, none of this would be urgent. But the organisations developing these systems are designing agents to operate autonomously in the world for days or weeks at a time, making real decisions with tangible consequences. Mustafa Suleyman, who co-founded DeepMind and now leads AI at Microsoft, proposed a “Modern Turing Test” that perfectly highlights the problem. Instead of testing whether a machine can imitate human conversation, his test asks whether an AI agent can turn $100,000 into $1 million on a retail platform within a few months.

Consider what that entails. The agent would need to research markets, design products, hire contractors, negotiate with manufacturers (possibly abroad), set pricing strategies, handle customer complaints, comply with regulatory requirements, manage logistics and warehousing, and organise payment systems. At each stage, it would be making decisions within the framework of democratic norms. What labour practices does the manufacturer adopt, and is the marketing misleading? Should the agent accept an offer from a local politician to disadvantage a competitor? Should it take a bribe from a supplier in the form of a crypto transfer?

These decisions are made by humans daily, and most of the time the answers seem obvious because humans have spent a lifetime absorbing the normative environment. The answers are not codified in a rulebook. They emerge from that invisible dance of observation and adjustment. An AI agent, no matter how well trained on legal texts and ethical principles, does not possess this “dance literacy”.

The incompleteness problem

Current approaches to AI alignment mainly assume that the right rules can be built into the system. Constitutional AI, the method used by Anthropic, fine-tunes models using a written constitution of principles. Other efforts collect “democratic inputs” through surveys and citizen assemblies. While the paper recognises these as valuable, it argues that they miss the core challenge. The issue is incompleteness: you cannot write instructions detailed enough to cover every possible situation an autonomous agent might face, because both situations and norms evolve.

Economists have understood this for decades in the context of human contracts. Every employment contract, partnership agreement, and supply chain arrangement is inherently incomplete. You can’t foresee every scenario, and when gaps appear between people, they fill them using shared norms, professional customs, and legal precedents, all of which are dynamic and partly implicit. An AI that stops learning norms at training time is like a new employee who memorised the handbook on their first day and then ignored all social cues from colleagues for the next ten years.

What the paper proposes

The technical agenda has two main parts. The first focuses on “normative competence,” embedded in individual AI agents. This is formalised through Bayesian adaptive decision processes, which mean that the agent maintains beliefs about the normative environment, updates those beliefs based on feedback (including punishment signals such as losing a contract or receiving a complaint), and makes decisions that account for uncertainty about what is acceptable. Crucially, this happens at inference time, in real-time, based on live context, rather than being pre-programmed into the model during training.

The second part involves creating new institutions and digital classification systems that can serve roles similar to those of courts, regulatory bodies, and professional norms for humans. The paper introduces “Model Specification Institutions” (MSIs), which would be democratically formed bodies (such as citizen assemblies, expert panels, digital juries). These bodies would establish shared standards, training datasets of acceptable and unacceptable behaviours, and real-time APIs that agents can consult in ambiguous situations. This does not mean AI companies should define their own rules; rather, it is calling for democratic communities to develop new infrastructure that AI agents can understand and respond to.

The paper also proposes adapting existing infrastructure—such as certificate authorities, which currently verify website identities—to certify that an AI has been trained to adhere to specific behavioural standards. Reputation networks, such as seller ratings on Amazon or Uber driver scores, could track AI behaviour over time and impose consequences on agents that repeatedly violate community norms.

Perhaps the most provocative argument concerns enforcement. Democracy doesn’t endure solely because governments enforce every rule from above. It survives because ordinary people enforce norms from below. You refuse to do business with a supplier who cheats. You complain when a company misleads you and vote against politicians who ignore court orders (well, mostly). This distributed enforcement, which the paper calls “third-party punishment,” is the engine that keeps the entire system functioning.

If AI agents replace humans in millions of daily transactions and those agents do not participate in this enforcement, the incentive structure collapses. Imagine a world where most business transactions are handled by AI agents that don’t care whether a trading partner has been found guilty of fraud, because the agents were not programmed to check for or respond to that information. The paper argues that AI agents will need to participate in distributed enforcement, refusing to transact with entities that violate community norms, just as humans do. Otherwise, the shift to agentic AI will quietly erode the social infrastructure on which democracies depend.

What this means for you

If you run a business, this paper should change how you think about deploying AI agents. The issue is not whether your agent can follow a rulebook. The question is whether it can read the room. Can it tell the difference between a legitimate business request and an attempt to corrupt a procurement process? Can it adapt its behaviour when community standards shift, without waiting for you to update its instructions? Can it recognise when a trading partner’s behaviour should disqualify them from further transactions?

If you are a citizen who votes, pays taxes, and occasionally debates politics, this paper describes the infrastructure of your daily life in terms you may not have previously considered. The norms you enforce through your micro-decisions, who you buy from, who you work with, and how you respond to rule-breaking are the operating system of democracy. What Hadfield, Trivedi, and Hadfield-Menell are asking is what happens to that operating system when a large fraction of those daily decisions are made by software that cannot read the social signals the system depends on.

The answer, if you follow the paper’s logic, is that we need to build new democratic institutions at the speed democracy demands, before the agents outrun the infrastructure. The alternative is a world where the formal structures of democracy persist, but the lived experience of it, the texture of mutual accountability in ordinary interactions, fades, like a coral reef whose skeleton remains after the living organisms have gone.

There is a question that has followed me through trade union education, universities and politics, though it rarely appears directly in policy documents or party manifestos. – How do people come to believe that their own judgement matters?

Over the years I have watched people discover a confidence they did not know they possessed. Sometimes it happened in a union classroom. Sometimes in a seminar room. Often it began when people recognised that their own experiences were not isolated incidents but part of a wider pattern shared by others.

That confidence was never simply personal. It emerged through recognition, conversation and collective understanding. People who had considered themselves spectators began to see themselves differently. They became participants.

The experience has made me increasingly sceptical of political arguments that focus exclusively on leadership. Leadership matters, of course. But politics is also shaped by assumptions about the public itself. Do we imagine people as capable of acting together, or primarily as recipients of decisions made elsewhere?

Much of our political culture is organised around representation. Others will speak. Others will decide. Others will carry the burden. There is a certain comfort in that arrangement, particularly during periods of uncertainty.

Yet there remains another tradition: one that places participation at the centre of public life and assumes that democracy is strongest when people exercise agency rather than merely delegate it. The tension between these traditions runs through many political institutions, including Labour.

I've explored that argument at greater length in a new essay, looking at confidence, participation and the cultural assumptions that shape political life. Read the full essay here: https://go.marshall.ie/acting-for-acting-with-marshall-review

The technology industry has spent the past three years debating artificial intelligence with the zeal of medieval theologians disputing angels on pinheads. Boardrooms have AI strategies, and governments have AI safety frameworks. LinkedIn has AI thought leaders, which is arguably the strongest case yet for existential risk. But somewhere beneath the acronym and the data centres in space, one central question remains unanswered. What, exactly, is intelligence?

We are developing systems we call intelligent, regulating systems we call intelligent, and worrying about systems we call intelligent, without a shared scientific consensus on what that term means when applied to humans, let alone machines. That is, to say the least, a problem.

The indefinable word

Ask a psychologist what intelligence is, and you’ll step into the epicentre of a fierce debate that has lasted for more than a century. The oldest and most statistically reliable answer comes from Charles Spearman, who in 1904 observed that people who did well on one kind of cognitive test also tended to do well on others. He called this underlying factor g, or general intelligence. The g factor is among the most replicated findings in psychology. It predicts academic performance, job performance, income, health outcomes, and even longevity, with a consistency that makes most social-science results look like coin flips.

And yet g tells you almost nothing about what intelligence really is. It is a statistical regularity, not a mechanism. Saying someone has high g is a bit like saying a car is fast. The measurement works, but the explanation is missing.

Howard Gardner tried to blow the whole thing up in 1983 with his theory of multiple intelligences, arguing that intelligence is not one thing but at least eight distinct varieties, from linguistic and logical-mathematical to musical, bodily-kinaesthetic, spatial, interpersonal, intrapersonal, and naturalistic. Teachers lapped this up. It confirmed their intuition that the kid who struggles with algebra but plays the cello like a prodigy is smart in ways traditional testing misses.

The problem is that decades of factor analysis have stubbornly refused to confirm Gardner’s categories as truly independent. Musical ability and spatial reasoning correlate, as do linguistic and interpersonal skills, and, in fact, everything correlates, which is more or less Spearman’s original point. Multiple intelligences is a useful pedagogical framework but a weak empirical theory, which is a polite way of saying it works better in classrooms than in laboratories.

Then there is François Chollet’s definition, which originates from the AI community and is arguably the most rigorous recent attempt to clarify the concept. In his 2019 paper “On the Measure of Intelligence,” Chollet defined intelligence not as the ability to perform any specific task, but as the efficiency with which a system acquires new skills, especially when confronted with tasks it has never encountered before.

This led him to develop the Abstraction and Reasoning Corpus (ARC), a benchmark of visual puzzles designed specifically to assess this ability. Humans usually solve most ARC tasks within minutes. The latest version, ARC-AGI-3, published in March 2026, makes the gap even clearer by placing agents in interactive environments where they must infer goals and plan action sequences without explicit instructions. Humans solve 100% of these tasks. At the time of the paper’s publication, frontier AI systems scored less than 1%. This gap cannot be closed simply by better prompt engineering. Whether this means current AI lacks intelligence, or only a particular kind of adaptive reasoning that humans excel at remains an open question.

The definitional problem is not just academic. Every claim about AI being intelligent, not intelligent, or dangerously intelligent depends on an implicit definition. Call a model intelligent, and you usually mean it produces outputs that would require human intelligence. Deny it, and you mean it lacks the comprehension, consciousness, or intentionality you consider necessary for genuine intelligence. Both claims are unfalsifiable without a shared definition, which explains why the debate generates much heat but little clarity.

What neuroscience understands (less than many think)

If psychology cannot agree on what intelligence is, perhaps neuroscience can explain how it works. The short answer is that it can, at least partly, though large gaps remain. We know a great deal about the brain’s individual components. We can map neural circuits, measure neurotransmitter activity, image blood-oxygen levels as proxies for activity, and trace connectivity patterns across cortical regions.

We know that the prefrontal cortex plays a key role in planning and abstract reasoning, that the hippocampus is central to memory consolidation, and that the cerebellum (once thought to be merely a motor coordination device) participates in cognitive processes that are not yet fully understood. We also observe that, within a species, larger brains tend to correlate weakly with cognitive ability, and that connection density and efficiency matter more than overall volume.

What we cannot do is explain how any of this produces thought. We have a parts list and some wiring diagrams, but no operating manual. The situation is roughly equivalent to having an inventory of components for a Boeing 787 without understanding aerodynamics. You could describe the wings, the engines, the control surfaces, and the hydraulic systems, and still have no theoretical framework for why the thing flies.

Two research programmes have made the most ambitious attempts to close this gap, and both illustrate how far there is to go.

Predictive processing

The first concept is predictive processing, most closely associated with philosophers Andy Clark and Karl Friston. They suggest that the brain is not a passive receiver of sensory data. It functions as a prediction machine that constantly builds models of what it expects to perceive, then updates them when reality differs from those expectations. Perception, in this view, is not bottom-up (data in, interpretation out) but top-down (expectation generated, error signal compared, model revised). You do not see the world as it is. You see your best guess about the world, corrected at the edges by incoming data.

Friston formalised this idea in the free energy principle, a mathematical framework suggesting that all adaptive behaviour can be understood as the minimisation of “free energy,” which roughly measures the gap between an organism’s internal model and the sensory evidence it receives. The framework is mathematically coherent and broadly applicable, and that is precisely the problem. If every possible behaviour of any living system can be reinterpreted as free-energy minimisation, then the theory rules nothing out, raising serious questions about its scientific credibility.

Integrated information theory

The second programme is Integrated Information Theory (IIT), developed by the neuroscientist Giulio Tononi. IIT takes on the even harder problem of consciousness rather than intelligence per se, but the two are tangled enough that progress on one would likely tell us something about the other. The theory proposes that consciousness corresponds to a quantity called phi (Φ), which measures the amount of information a system generates “above and beyond” its individual parts. A system with high phi is one whose behaviour cannot be reduced to its components acting independently. The whole, in a precise mathematical sense, is more than the sum of its parts.

IIT makes some bold predictions. It implies that consciousness is a property of a system’s physical structure, not its function. A digital simulation of a brain that runs the same computations on different hardware might have zero consciousness under IIT, even if it behaves identically to the original. In practice, calculating phi for any system more complex than a handful of nodes is computationally intractable, which limits the theory’s practical utility. You can define consciousness precisely and still be unable to measure it in any real system, which is a bit like having a perfect recipe for a cake you can never bake.

Neither predictive processing nor IIT amounts to a theory of intelligence in the way that general relativity is a theory of gravity. They are frameworks, useful and generative but incomplete, and they throw light on aspects of cognition without explaining the whole. And the gap between “aspects” and “the whole” may be permanent, for reasons we will get to.

Why large language models work

If the science of biological intelligence is patchy, the science of artificial intelligence is in an even stranger position. The engineering works spectacularly well, while the theory lags behind, like a civil engineer who builds bridges that hold up beautifully but cannot fully explain the physics of load distribution.

We understand the mechanics of large language models in fine detail. A transformer architecture processes sequences of tokens through layers of attention mechanisms, and during training the model adjusts billions of parameters to minimise the error between its predicted next token and the actual one. Scaling laws, first characterised by Jared Kaplan and colleagues at OpenAI in 2020, describe a remarkably smooth power-law relationship between compute, dataset size, model parameters, and performance.

These are genuine scientific results. They let engineers predict, with useful accuracy, how a model of a given size trained on a certain amount of data will perform on standard benchmarks. What they do not explain is why training a system to predict the next word in a sequence produces behaviour that appears like reasoning, planning, analogy, and (occasionally) creativity.

The most provocative explanation comes from the compression hypothesis, most forcefully articulated by Ilya Sutskever, then of OpenAI. The argument roughly runs like this. Predicting the next token accurately requires modelling the process that generated the text, and that process is human cognition. To predict well, you must compress the structure of human thought into your parameters. Compression, in this view, is not merely correlated with intelligence but constitutive of it. A model that achieves better compression has, in a meaningful sense, come to understand the world better.

This is philosophically interesting and empirically suggestive, but it is not a complete theory. It does not explain why certain abilities appear discontinuously as models scale. Small models cannot perform multi-step arithmetic. Larger models can suddenly, without anyone having specifically trained them for it. These “emergent capabilities” are predicted by no current theory and explained by no current framework. They simply happen, and then engineers and researchers argue about what they mean.

Mechanistic interpretability, an active research programme at Anthropic among others, is perhaps the most promising attempt to open the black box. The work identifies specific circuits within trained models that correspond to identifiable computations, so that one cluster of neurons detects sentiment and another tracks syntactic dependencies. The results are revealing, but they are roughly at the stage where neuroscience was when it discovered that specific brain regions correspond to specific functions. Knowing where a computation happens is useful. Knowing why the system learned to do it, and why it generalises beyond the patterns in the training data, is the harder question.

The “stochastic parrots” critique, most prominently advanced by Emily Bender, Timnit Gebru, and colleagues in 2021, argued that LLMs are only sophisticated statistical mimics. Noam Chomsky has made similar arguments, insisting that next-token prediction cannot amount to genuine linguistic comprehension. Melanie Mitchell has taken a more cautious position, arguing that current AI systems lack the conceptual abstraction and analogy-making she sees as central to intelligence, while leaving open the possibility that future architectures might achieve it.

The honest answer is that nobody knows who is right. The stochastic-parrot position seemed more defensible in 2021 than it does in 2026, because the systems have kept improving in ways a “mere statistical mimic” would not obviously be expected to. But the lack of a theory means that “would not obviously be expected to” is carrying more weight in that sentence than it should. We do not have the theoretical tools to distinguish genuine comprehension from a sufficiently convincing imitation of it, and those tools are not arriving quickly.

Can intelligence be formalised at all?

Here we reach the question beneath the question, and the answer is uncomfortable for anyone who prefers their science tidy. A hidden hope in much AI research is that intelligence resembles thermodynamics: messy and chaotic at the micro level, but governed by clean, discoverable laws at the macro level. Individual gas molecules move unpredictably, yet aggregate behaviour follows the ideal gas law with almost miraculous precision. Perhaps intelligence works the same way, messy at the level of individual neurons or attention heads, but obeying some elegant principle at a higher level of description.

The problem is that thermodynamics works because you can ignore which specific molecule is where. It is far from clear that cognition has this property. The specific structure of a person’s knowledge, the particular history of their experiences, and the exact wiring of their neural connections all seem to matter in ways that resist averaging out. A brain is not a gas. Its macro-behaviour may not separate cleanly from its micro-state, and if it does not, no thermodynamics-style theory is possible.

There is a deeper problem. Any formal theory of intelligence needs to specify what intelligence is for, what problem it solves, and what it optimises. A thermostat optimises temperature, a chess engine optimises board position, and both can be fully described by their objective function. But intelligence seems to be precisely the capacity to redefine what counts as the problem. A human can decide whether to play chess at all, invent a new game, or abandon the entire framing and go for a walk. Formalising that kind of open-ended reframing may require a kind of mathematics that does not exist yet, or it may resist formalisation altogether.

This is where the biology analogy becomes revealing. There is no “theory of organisms” in the same sense that there is a theory of electromagnetism. Biology has a powerful organising framework, evolution by natural selection, along with a vast accumulation of mechanisms, trade-offs, and contingent historical facts. You can explain any feature of an organism after the fact. You cannot derive organisms from first principles. The evolutionary biologist Stephen Jay Gould argued that if you replayed the tape of life from the same starting conditions, you would get a completely different set of organisms. The outcomes are historically contingent, not mathematically necessary.

Intelligence may be the same kind of thing: a product of evolutionary tinkering, cultural accumulation, and developmental contingency that allows useful generalisations but not the kind of closed-form theory that would satisfy a physicist. We may end up knowing intelligence the way we know weather: well enough to make useful short-term predictions, poorly enough that long-range forecasting remains unreliable, and never with the exact analytical solution that would let us derive tomorrow’s clouds from first principles.

Why does any of this matter outside of a philosophy seminar?

The temptation is to treat all of this as an abstract debate, the sort of thing academics argue about while engineers get on with building things that work. That temptation should be resisted, because the theoretical vacuum has practical consequences.

AI safety without a theory of intelligence is navigation without a map. The field depends on assumptions about what future systems can achieve and how those abilities will develop. If we do not understand why current systems perform as well as they do, we cannot predict whether the next generation will improve steadily or make a sudden leap, as we have seen with Anthropic’s Mythos. Scaling laws tell us that larger models perform better, but not what “better” means at scales we have not reached. Will a model 100 times larger than current frontier systems merely write more polished prose, or develop something qualitatively different? Nobody knows, and we lack the framework to reason about the question.

Regulation without definitions is theatre. Governments are drafting AI rules around distinctions (general-purpose versus narrow, high-risk versus low-risk) that depend on a theoretical grasp of intelligence we do not possess. The EU’s AI Act defines a “general-purpose AI model” by compute thresholds that are essentially arbitrary, because no theory links compute to ability in a way that would make any threshold principled. The fault is not the regulators’, but the tools’, which are insufficient.

A business strategy built on vibes is expensive. The corporate world is investing hundreds of billions on the assumption that current trends will continue. Perhaps they will. But the history of technology is full of S-curves that plateau earlier than expected, and the lack of a theory makes it harder than it should be to distinguish genuine improvement from benchmark gaming and evaluation contamination. When a model scores 90% on a medical exam, does that mean it has medical knowledge, or that enough medical-exam text was in the training data? The answer is “it depends what you mean by knowledge,” and we are back to square one.

Where this leaves us

This is not a counsel of despair. Science often advances without complete theories. Medicine cured scurvy centuries before vitamin C was discovered, and engineers built steam engines before thermodynamics was formalised. Practical progress does not require a finished theory, though it helps, especially when the stakes are high enough that mistakes carry consequences beyond a failed experiment.

We are roughly where physics was before Newton. We have observations (scaling laws, emergent abilities, benchmark performance), useful heuristics (more compute and data tend to produce better models), and fragments of theory (compression, mechanistic circuits, predictive processing), but no framework that unifies them and makes novel predictions. The “I” in AI is still a placeholder, a trillion dollars of investment balanced on a word we cannot define.

Mes enfants,

Il viendra des jours où vous tendrez vos yeux vers les miens comme on tend une coupe vers la pluie. Vous attendrez de moi un signe, un sourire, une parole qui vous dise que vous avez bien fait, que votre chemin est le bon, que votre rêve mérite de vivre. Je comprends cette attente, car moi aussi j’ai été enfant, moi aussi j’ai cherché dans le regard des grands une lampe pour éclairer ma petite nuit. Mais je voudrais vous apprendre, avec la douceur que donne l’amour et la gravité que donne le temps, que la lumière la plus fidèle ne vient pas toujours du dehors.

Dès la cour de récréation, le monde commence son étrange commerce. On y échange des billes, des goûters, des secrets, mais aussi des jugements. Un rire peut devenir une couronne, un silence peut devenir une blessure. On veut être choisi dans l’équipe, invité au jeu, reconnu par le groupe. On veut que son cartable, sa voix, ses gestes, sa manière de courir soient acceptés. Déjà, sans le savoir, l’enfant apprend à se mesurer aux yeux des autres. Il se demande s’il est assez drôle, assez fort, assez beau, assez pareil. Pourtant, celui qui passe toute sa vie à vouloir être assez pour les autres finit par devenir étranger à sa propre voix.

Puis viennent les années où le miroir grandit. Le corps change, la voix tremble, le cœur s’enflamme pour des présences qui passent dans un couloir comme des soleils rapides. On veut plaire, être compris, être préféré. Une parole reçue devient une loi secrète. Une indifférence semble une condamnation. Mais aucune adolescence ne devrait être une prison bâtie par l’opinion des autres. Elle devrait être un jardin sauvage, parfois désordonné, où l’on apprend à reconnaître sa propre saison.

Plus tard, vous entrerez dans le vaste théâtre des adultes. On vous demandera des résultats, des titres, des preuves. On vous dira qu’une vie réussie se voit à la taille d’une maison, au poids d’un compte, au rang que l’on occupe parmi les hommes. Vous croiserez des bureaux où l’on sourit sans joie, des tables où chacun parle pour être entendu, des villes où les fenêtres sont nombreuses mais les âmes parfois fermées. Là encore, vous serez tentés de demander au monde la permission d’exister. Ne la demandez pas. Le monde est un juge distrait. Il applaudit aujourd’hui ce qu’il oublie demain.

Dans l’amour aussi, ne confondez jamais être aimé et être autorisé. Celui ou celle qui vous aime vraiment ne devient pas votre ciel entier. Il marche près de vous, il ne vous remplace pas. Si vous donnez à l’autre le pouvoir de décider de votre valeur, vous lui remettrez une charge trop lourde. Aimez avec grandeur, avec fidélité, avec présence, mais gardez en vous un sanctuaire que nul ne doit gouverner.

Un jour, certains d’entre vous deviendront parents. Alors vous comprendrez que l’on peut aimer un être plus que sa propre paix. Vous verrez un enfant trébucher et votre cœur tombera avec lui. Vous le verrez sourire et le monde redeviendra neuf. Mais gardez cela en mémoire. Vos enfants ne seront pas là pour confirmer votre valeur. Ils ne seront ni vos trophées, ni vos prolongements, ni vos réparations. Ils seront des voyageurs confiés quelque temps à votre maison. Vous leur donnerez du pain, un abri, des mots, une mémoire, puis vous leur laisserez l’espace de leur propre ciel.

Et moi, votre père, je vous parle ainsi parce que je ne veux pas devenir votre tribunal. Je ne veux pas que mon regard soit une porte fermée devant laquelle vous attendriez toute votre vie. Quand vous venez me parler de vos projets, de vos idées, de vos élans, ce qui me rend heureux n’est pas de vous approuver comme on appose un sceau sur un papier. Ce qui me rend heureux, c’est votre flamme. C’est cette voix qui s’élève quand vous parlez de ce qui vous appelle. C’est de sentir que la vie en vous cherche sa forme.

Nos discussions ne sont pas une audience. Elles ne sont pas le face à face d’un juge et d’un demandeur. Elles sont une table simple, un soir calme, deux expériences qui se rencontrent. Je vous donne ce que j’ai vu, vous m’offrez ce que vous découvrez. Je vous raconte les routes, vous me montrez l’horizon. Il n’y a pas là de supérieur ni d’inférieur, seulement deux âmes qui se parlent, l’une plus ancienne, l’autre plus neuve, toutes deux inachevées devant le mystère.

Mes enfants, ne cherchez pas à plaire à tout prix. Cherchez à vous tenir droits. Ne cherchez pas à être validés. Cherchez à être vrais. Une vie ne devient pas grande parce que tous l’applaudissent, mais parce qu’elle demeure fidèle à sa source. Que votre axe soit en vous, non par orgueil, mais par paix. Et lorsque vous viendrez vers moi, venez non pour recevoir la permission d’être qui vous êtes, mais pour partager la beauté difficile de le devenir.

I think I’m a little bit fighting off depression, and so I will take today as a win. I had a good session at the gym, and I am tired and going to bed.

Ja fa temps que les intel·ligències artificials han superat el test de Turing i és pràcticament impossible diferenciar un text humà d'un escrit per una intel·ligència artificial.

La facilitat per crear aquests textos tan realistes ha fet que la xarxa estiga envaïda de textos sospitosament genèrics i amb estructures sospitosament paregudes als esquemes que fa servir ChatGPT. A més, és d'esperar que aquests algoritmes vagen fent-se més i més «intel·ligents» fins que arribe el veritable dia del judici final en què ja no podrem distingir la paraula humana de l'algoritme de la màquina.

I en eixe context cal afegir-hi l'altra banda: una xarxa cada volta menys atomitzada. Ja ningú ix de les seues tres o quatre webs de confiança, d'Instagram passem a YouTube i de YouTube a Twitter, i en això JA PROU! Ningú va a subscriure's a un blog d'un subjecte desconegut per llegir entrades quilomètriques sobre assumptes no massa entretinguts. Els tuits tenen 280 caràcters i ningú llig un tuit sencer. Encara que sí que hi ha gent disposada a pagar una subscripció premium per poder escriure més.

Doncs, per a mi aquest ha sigut el moment ideal per començar aquest blog. Tant el lector com jo sabem que d'açò no vaig a traure un duro, que jo i tu estem ací per voluntat i per gust, no per traure un rendiment al nostre temps d'oci.

Que ningú em llig? Tant me fa, podré escriure més i sobre més temes perquè no hi ha temes tabú en un blog sense lectors.

Porte escrivint anys als meus apunts. Ara, en aquest blog, he decidit fer pública part d'eixes ocurrències que abans quedaven oblidades als meus quaderns.

I still feel a little sick and overall exhausted. My nephews came over in the morning and just wanted nonstop playtime. While I have not heard any updates or received any news, I can’t help but feel a little grateful today.

I am grateful to have time for family.

I am grateful to pursue my hobbies.

I am grateful I have extra time with my dog.

I am grateful to have good health overall.

I am grateful for all I am able to do for others, but more importantly for myself during this forced time off.

Thank you for your message. I am currently out of office with no set return date. I will get back to you when the time is right.

Today was filled with highs and lows. It began by setting all my planning to work and getting down to it. I somehow just barely managed to finish on time for the party. I don’t know how I pulled it off, but it turned out alright. It was not my best work, but it came together just enough.

Then there was a game that I was really looking forward to but we simply did not get the result I wanted. I felt heartbroken and sad, but that is how sports work. We did get two other wins from other sports so that still felt encouraging enough.

Nothing has changed yet. I am starting to get anxious and thinking that I should start considering other options.

Thank you for your message. I am currently out of office with no set return date. I will get back to you when the time is right.