Articles
The Wrong Unit
Why systemic AI risk lives in deployment, not FLOP counts.
Three ways through my thoughts
Dossiers
Three ways through my thoughts.
The Landlord Economy
Why Sweden’s AI “leadership” is just cheap rent
The purest signal is the one no one profits from sending
Why some recognition weighs more than others.
The Lottery That Looks Like Genius
Why AI solving maths isn't AI understanding maths.
The Silicon Valley Central Banks: Welcome to the Compute Standard
How Big Tech builds sovereign reserves, and why states become battery packs.
The Vatican Wrote the Policy Document We Refused to Write
How Magnifica Humanitas reclassifies AI infrastructure, and why Anthropic went there to avoid talking about it.
Vad jag ser från insidan
En replik till Karin Pettersson om AI-debatten. Varför offentlig europeisk AI-infrastruktur är vägen framåt.
Spörv Is Not an Argument Against AI. Spörv Is a Business Model.
A response to Anna Andersson's AI critique in Aftonbladet Kultur. The problem isn't the technology, it's the distribution model.
Spörv är inte ett argument mot AI. Spörv är en affärsmodell.
Ett svar på Anna Anderssons AI-kritik i Aftonbladet Kultur. Problemet är inte tekniken, det är distributionsmodellen.
From Cornholio to 6 7: What "Bad Influence" Media Actually Looks Like
A generational autopsy of the things adults were terrified of, and the things they probably should have been.
Från Cornholio till 6 7: Hur "Dåligt Inflytande"-medier faktiskt ser ut
En generationsobduktion av det vuxna fruktade, och det vuxna förmodligen borde ha fruktat. Från Beavis och Butt-Head ...
You Can't Own the Accountability If You Don't Own the Stack
Sweden signed Pax Silica without debate. Accountability ends where control ends.
Du kan inte äga ansvaret om du inte äger stacken
Sverige undertecknade Pax Silica utan debatt. Du kan inte ha granskningsbart AI-beslutsfattande om infrastrukturen ko...
The Accountability Gap Is the Function, Not the Bug
Gothenburg's school algorithm placed 450 children wrong. The problem isn't design, it's incentives.
Ansvarsgapet är funktionen, inte felet
Göteborgs skolalgoritm placerade 450 barn fel. Problemet är inte designen, det är incitamenten. Ett svar på Jacob Dex...
The Wizard of Silicon Valley: AI's Invisible Human Underclass
Behind the AI industry's magic curtain sits an army of invisible workers paying for our cognitive luxury with their mental health.
When Safety Becomes the Danger
One hundred robotaxis stopped simultaneously on a Wuhan expressway. The safety mechanism designed to protect passengers became the danger itself.
The CLI Never Died. You Just Couldn't Type.
Natural language didn't kill the command line. It democratized it. Why forty years of GUI was a necessary detour.
The 20-Watt Challenge: Why AI Is Drowning in Its Own Energy
Microsoft signed a nuclear power deal for its data centers. The brain runs on 20 watts. The difference isn't efficiency, it's architecture.
The Faster Idiot Paradox: Why More Data Won't Save AI
Why scaling data and compute won't produce genuine intelligence. From Epic's failed sepsis model to Tesla's crashes — the case for causal AI architecture.
Why Context Isn’t the Moat Everyone Thinks It Is
There is a narrative gaining traction in the C-suite: as models commoditize, context becomes the moat.
The Timestamp Problem
Exploring cryptographic timestamping using Bitcoin's blockchain to solve priority disputes over digital work. A Boxing Day craft that solved a fundamental problem.
AI Reality Check Oct.2025: Beyond the Agent Hype
Where we actually stand in the AI landscape
Cost per Correct Completed Task (C/CCT)
A deployment-grade framework for robust evaluation of AI systems
The Art of Building and the Art of Undoing
An architect's reflection on creation and deconstruction
And then… nothing
A story about post-breakthrough paralysis
Why Are We Building Economic Systems That Never Grow?
How five economic realities converge to make self-reinforcing capital structures possible before 2030.
Varför bygger vi ekonomiska system som aldrig växer?
En kritisk analys av ekonomiska system och tillväxt. Varför bygger vi system som är designade att stagnera? En djupdy...
AI Is Hitting an Energy Wall
The solution isn't more power — it's a new kind of intelligence
The Alignment Horizon
On AI alignment and its challenges
The Transformer Epoch is dying
Architectural inflections and the future of artificial intelligence
Limitations of Large Language Models
Context, truth, and scaling challenges — 2025 and beyond
The Great Convergence
How AI is igniting a new industrial revolution
The fundamental limitations of AI agent frameworks expose a stark reality gap
The promise of autonomous AI agents has collided with harsh technical realities
AI Agents Are Broken by Design
Think your AI employee is autonomous? Wait until you see the pancake recipe bill
Cloud Capital
Infrastructure, innovation, and influence in the post-physical economy
Governance as Science
How we learn to rule ourselves
Why Don't We Build Machines That Grow?
How five engineering domains are converging to make living-silicon systems viable
Tax as Architecture
Toward a post-industrial civilizational contract
Skatt som arkitektur: Mot ett post-industriellt civilisationskontrakt
Skatt är mer än en funktionell mekanism för resursutjämning. Den är en ontologisk markör – ett uttryck för vad en civ...
Social Media Was Meant to Connect Us
But it's fueling mental health issues, stress, and identity struggles among young people
Uppkopplade och nedstämda: AI, sociala medier och ungas välmående
Sociala medier skulle skapa kontakt – istället ser vi en ökning av psykisk ohälsa, stress och självbildsproblem hos u...
CivOS: The Instruction Set for a New Civilizational Operating System
Redesigning the foundational logic of societal coordination
CivOS: Installationsguide för ett nytt Civilisatoriskt Operativsystem
Ett ramverk för att förstå och omdesigna den grundläggande logik genom vilken samhällen tolkar världen, koordinerar h...
Peak LLM: Why Language Models Reach Their Limit
And the path toward machine cognition
Peak LLM: Varför språkmodeller når sin gräns
Utforskar de fundamentala begränsningarna hos stora språkmodeller och vägen mot maskinell kognition genom hybridarkit...
Everything Is Already One
A worldview for a post-mystical humanity
Allt Är Redan Ett
En världsbild för en post-mystisk mänsklighet. Ett försök att sammanfoga det vi vet med det vi intuitivt anar, till e...
The Anatomy of the Modern Human
A deconstruction of the post-biological self
Den moderna människans anatomi
En dekonstruktion av det post-biologiska jaget. Vad har vi blivit? Och vad har vi förlorat på vägen?
OTRUST: Building a Protocol for Distributed Truth
Because internet was missing this layer
Neural Networks and AGI: Limitations, Progress, and Testable Paths Forward
Mapping documented limitations and empirically grounded possibilities
Neurala nätverk och AGI: Begränsningar, framsteg och testbara vägar framåt
Kartläggning av dokumenterade begränsningar hos neurala nätverk och empiriskt grundade möjligheter för framtida utvec...
Distributed AI Systems: Architectures, Challenges, and Future Prospects
If traditional AI resembles a brain, distributed AI represents a digital nervous system
Distribuerade AI-system: Arkitekturer, utmaningar och framtidsutsikter
Om traditionell AI liknar en hjärna, representerar distribuerade AI-system ett digitalt nervsystem: utspritt, adaptiv...
Kris scribbles...on basic edu
Revamping the education system with recursive metacognitive integration
The Blind Spot of Industrial Collapse
Why sustainability directives fail without systemic transformation
The Present Future of Collaboration
Remote work, team collaboration tools, and the evolution of how we work together
The "Collaboration Paradox" and What We Can Do About It
More apps and tools to help us collaborate better, but with less effectiveness in the execution
How to Ensure Secure File Transfer
Safeguarding personal and proprietary data using secure protocols
5 Ways to Adapt Your Software to Remote Workers
Helping your users get used to the new norm — adapt and overcome