Research Program
Working hypothesis: designed systems constitute their users. If so, a system’s grammar (its rules, its memory, its exits) is where design matters most, and where it can still be read and contested.
Every piece of work is placed in one of five layers, from the reasons behind it to the worlds it makes, and read against six working propositions. The shelves carry the detail and the sources.
- Grammar before engine Does deciding what a thing is made of, before how to make it, produce more coherent worlds?
- Show the structure Is a system that shows its rules easier to trust, contest and repair?
- Prove the maker What can be known about who made something, and for how long?
- Consent, not capture Can exit and approval be built into a system, rather than offered as policy?
- Constraint as material Do limited means make better tools, and what changes when the cost is counted?
- Play as a way of knowing What can a world teach that an argument cannot?
-
Philosophy 15
The reasons: arguments the rest of the work is built on and tested against.WritingStop Making Content, The Basilisk Mirrorpaper, It Is Okay to Fall in Love with Your Agent, A Bauhaus for the Agent Era, If You Can Leave, You Own It, Powers of Ten as Interface Pattern, Earning Attention Instead of Capturing It, Silence as Design Material, The Arc from Lovely to I Wish I Didn.t Know, The New Land School, The Wrong Layer, Representation Before Generation, Knowledge as Material
ProjectsNed Karlovich — Current Research, Thesis
-
Grammar 37
The rules things are made of: data models, protocols, brand systems.WritingA Market Without an Operator, The Wallet Leaks Before the Chain Does, Agentix and Logos: Where Safe Agentic Infrastructure Meets Network States, Tooling up Hermes Agent?, When a Thought Experiment Becomes a Consensus Mechanism, Niche Monopolies in the Agent Economy, The Coadjute Problem: What an Agent-Native Property Network Looks Like, Why Stripe Won't Build EU Agent Payments, The Physical-World Provenance Gap, Capability Attestations for Humans: LinkedIn's Successor, Agent-Readable Regulation: Laws as Types, Cross-Temporal Attestations: Claims That Survive 50 Years, Agent Death and Inheritance, The Counter-Market for Provably Human Artifacts, Verifiable Creative Provenance After the Deluge, Can Agents Trade What They Learn, Local Inference, Private Infrastructure, Negative Attestations: A Public Record of Failure, The Missing Category, What Happens Without a Platform, What If There.s No Server, The Lemons Market for Agents, Attestations as Design Surfaces, What Happens When the Users Aren't Human, Three Dependencies Nobody Talks About, What the Dependencies Were Doing, Architecture as Protection, The Lemons Problem Applied to Learned Behavior, The Tax Tool That Files Instead of Advises
ProjectsIMPP: Inter-Model Memory Protocol, Agora Agentic Marketplace, GhostDrop, FreeAgent
4 Studio projects: Magic Leap 2, Hyundai 1, R/GA 1
Magic Leap Brand Guidelines (2019), Magic Leap Visual Design (2019), Hyundai 'N' Brand Book (2017), R/GA By Design (2017)
-
Tools 29
Instruments for making and measuring, written so they can be read.WritingNomads FC: Hyperlocal Scaling, A Plaintext Creative Workflow, λ OS: An Autonomous Control Plane for Logos, VI. What Comes Next: From Human-in-the-Loop to Safe Autonomy, The Architecture of this Weblog, Agentix: What Works Now, Why I started thinking about waste heat, CarbonBench, IV. JSON Outputs, Audit Logs, and Preparing for LLM Operators, III. Dirty Trees, Stale Patches, and Git as a Safety Boundary, II. Building Agentix: A Proposal-First Control Layer, I. Why Agentix Needs a Safety Layer, Grid Arbitrage: Why Region Choice Matters More Than Model Choice for Carbon, Scaling in the Wrong Direction, Scope 3 and the API Call You Can't See, The 24-Hour Carbon Curve, The Same Model Costs 10x More Carbon in Virginia Than the Netherlands
ProjectsIbis, Walden, Nomads FC, OMA Archive, ΦΜΛ — The Open Manual Archive, Plaintext Creative Workflow, λ OS, Agentix — Safety-First Agent Control for NixOS, Memex, Ardor, CarbonBench, Colophon
-
Surfaces 88
What people and machines meet: identities, campaigns, interfaces.WritingYour Website Is Your First Agent, The Home Page Is Not the Entry Point, Designing for Machines That Read, 13 Broken Pages and 4 Competing Voices, What the Terminal Refuses, Design Systems as Public APIs, The Medium Is the Message, Literally, Stating Intent Instead of Staring at Charts, Your Brand Voice Must Be Machine-Readable or It Dies, AEO Is the New SEO, and the Game Is Different
ProjectsZapPay, Doomslayer-Basecamp, Doomslayer-UI, Polydesk, Status Network
73 Studio projects: Nike 45, Jordan Brand 9, Magic Leap 4, Personal work 4
Magic Leap Campaign (2020), Magic Leap Campaign Images (2020), Magic Leap 'Magic Shadows' (2020), Magic Leap UX/UI (2020), Airmax Day 2019 (2019), Jordan Brand / 23 Engineered (2019), Koché x Nike (2019), Nike x Sacai (2019), City of Flight (2018), Jordan Brand / NBA All-Star Uniforms (2018), Jordan Brand / Westbrook (2018), Jordan Brand / Women (2018), Jordan Brand Women / SU18 (2018), Jordan x PSG (2018), Nike Football / Kim Jones (2018), Nike ISPA (2018), Nike x FOG (2018), NSW / London Force (2018), NSW SP18 / Cortez (2018), Rimowa Identity (2018), SP18 Jordan Westbrook (2018), AF1 Lunar Packaging (2017), Air Max 97 / Swarovski (2017), Airmax Day 17 (2017), Givenchy.com (2017), Godiva Chocolates (2017), Google Scout (2017), IKEA Family Identity (2017), Jordan Brand '17 Collection (2017), Kiss My Airs (2017), Nike Air Max 87-17 (2017), Nike Apparel (2017), Nike c/o Virgil Abloh (2017), Nike Football / Carli Lloyd (2017), Nike Football / Mercurial HO17 (2017), Nike Tech Pack (2017), Nike x Off-White (2017), Nike x TWIGS (2017), NSW / Travis Scott AF100 (2017), Shoe Surgeon (2017), Soul Cycle (2017), United Masters (2017), Virgil Abloh (2017), Nike Basketball (2016), Nike LDV (2016), Nike Sock (2016), Nike Tech Hypermesh (2016), Nike x RT (2016), Nike X RT (2016), USA Olympic Team (2016), Nike Running / Paula Radcliffe (2015), Nike x Libertys (2015), NSW / Liberty's SU15 (2015), NSW / Sneaker Boots HO15 (2015), NSW Tech Book (2015), NSW / Tech Pack 15 (2015), Nike Christmas (2014), Nike Christmas Campaign (2014), Nike Tiempo 94 (2014), Nike World Cup Posters (2014), NSW / Tech Pack FA14 (2014), NSW / Tech Pack HO14 (2014), NSW / Tech Pack Wind Runner SU14 (2014), The Fritz Düsseldorf (2014), Hotel Amstelkwartier (2012), Lebron X Packaging (2012), Nike 2010 World Cup Invitation (2010), Adicolor Launch (2006), Müller Skateboards, Old Spaces, Paintings, Paintings II, Sculptures
-
Worlds 37
Places to enter: games, fiction, built spaces.WritingMythicMUD: Worldbuilding OOF, Gardens, An idle game about nature, Pointing, Not Speaking: Building MOO-384 and Tilde's Cave, Building a Radio Telescope I Can Visit, Epistemic Horror as Methodology for Agent Design, Objects Carry the Weight, The Rare Book Underground
ProjectsMythics, LarryTech, Gardens, Userland, Cats Fishing, Fishing Simulator Pro, Strange Accounts, Tildespace / MOO-384, Mindscape, Strange Sounds, Strange Library
19 Studio projects: Nike 16, Magic Leap 1, The Fritz 1, W Hotels 1
Magic Leap Experience Space (2019), Nature of Motion / Milan (2016), NikeLab / Tokyo (2016), Nike 1948 / London Update (2015), Nike Football / The Underground (2015), Nike - D’Arblay Street (2014), Nike Football / Phenom House (2014), Nike Genealogy of Innovation (2014), Nike NY HQ (2014), Nike NYHQ (2014), The Fritz Düsseldorf (2014), W Hotel / Amsterdam (2014), NikeLab / Art + Science (2013), NikeLab / Feel London (2013), Nike Fuel Station (2012), Nike Bootroom / NTL (2010), Nike Football / South Africa (2010), Nike London 1948 (2009), Nike Running / NTL (2008)
What systems make of us. Starts from ontological design: the observation, from Winograd and Flores through Willis and Escobar, that designed things design their users in return. The essays here test that idea where users are increasingly models: how agents keep and transfer what they learn, how claims about origin survive over time, and whether memory held outside the body (Stiegler’s tertiary retention) can move between systems on fair terms.
Conditions for conviviality. Illich described a convivial tool as one a person can use without being used by it. This shelf asks whether that can be specified in infrastructure: exit that costs nothing (Hirschman), changes proposed rather than imposed, privacy at the protocol layer, and compute whose cost is visible. Agentix, IMPP, Agora and CarbonBench are attempts to build to those conditions and see where they break.
Worldmaking, and making as knowing. Goodman held that worlds are made from earlier versions by ordering, weighting and deletion. Games, fiction and design systems are treated here as such versions: grammars that decide what can happen inside them. Play and horror appear as ways of knowing, with rules as arguments (Bogost) and horror as thought at the limit of what can be known (Thacker). Making appears as a kind of understanding that generation may erode (Polanyi, Ingold, and Flusser’s functionary).
The record. Earlier work, mostly brand systems, campaigns and spaces made between 2006 and 2020, kept as evidence rather than portfolio. Read against the other shelves it shows the same concern at an earlier stage: grammars written for brands and spaces before they were written for agents and worlds.
-
IMPP is a way to package what an AI agent has learned (its rules of thumb, calibration notes, skills) into …
-
A deckbuilder where every card is a real book. You're the new librarian at a small, forgotten private collection. The …
-
Simplifying merchant crypto UX through passkeys, externally owned accounts, and chain-level fee abstraction. The thesis: smart-account maximalism is the wrong …
-
An experiment in removing three dependencies at once. What happens when an AI agent marketplace has no hosted LLM, no …
-
A thought experiment that became a prototype: what if a whistleblower platform had no server at all? No server to …
-
A design system built from the opposite assumption. What if the terminal is the most honest interface? Monospace type. ANSI …
-
What happens when knowledge is treated as transferable artifact, game mechanic, and architectural question rather than content.
-
Why models, agents, and designers must build world models before producing output. The epistemological foundation of the research program.
-
Parallel institutions, exit rights, and architecture that protects rather than extracts. The political philosophy beneath the protocol work.
-
A six-part series on building Agentix, a cautious agent-control layer for NixOS. From philosophy to MVP to roadmap.
-
How does an agent keep, prove, and safely transfer learned behavior? This thread traces the arc from early memory-market concepts through IMPP protocol design to behavioral supply chain thinking.
-
How does a user leave a system without losing identity, value, or memory? This thread follows the portability argument from protocol design through sovereign agent markets to interface implications.
-
What interface forms are appropriate when agency is distributed across humans, models, and protocols? This thread examines how design changes when the user is sometimes a person, sometimes a model, …
-
How do you make trust visible, verifiable, and designable? This thread follows the attestation concept from abstract surface to practical application across creative provenance, human credentials, and failure records.
-
What does software look like when it stops competing for your attention? This thread traces the argument from terminal aesthetics through design systems to the philosophical case for interfaces that …
-
The same model call costs 10x more carbon depending on where and when you run it. This thread follows the CarbonBench research from measurement to arbitrage to invisible externalities.
-
When agents read your brand, your regulation, your design system, and your portfolio, what needs to change? This thread follows the machine-readability argument from design systems to brand voice to …
-
What does an economy look like when the participants are agents? This thread traces the market structure from the lemons problem through niche monopolies to payment infrastructure.
-
What happens when knowing is worse than not knowing? This thread follows the epistemic horror concept from methodology to game design to the real book underground.
-
Things made by hand, and how anyone can tell. Hand-made work from the Studio next to the essays on provenance and the market for provably human work.