Critical thinking infrastructure: an essay series

This is a series of essays on building infrastructure for better decision-making. The end-goal is to allow decision makers dealing with high-pressure and time-sensitive problems, to ask and get defensible answers in minutes, instead of weeks and months.

The immediate context is logistics and supply chain planning, though the ideas reach further. The argument is that the bottleneck in most high-stakes decisions is not the absence of data or models but the absence of infrastructure for asking good questions quickly. The essays cover shortcomings in existing decision models, the technical architecture of a solution, AI, and the human layer.

Mapping the venture case: locating the moat, the open-source targets, and the build-buy line

A technical accompaniment to the Deliberation Meets Reality series, in which we analyse and position the platform as a commercially viable and venture backed initiative.

Abstract

The same architecture mapped from a venture-backed company perspective rather than a public infrastructure one. The exercise identifies where value is evolving on the stack, why the substrate layer (the graph engine, the domain ontology framework, the rehydration contracts) is underfunded relative to where capital will flow next, and which components should be open-sourced to accelerate the commodity tier rather than defended as proprietary. It names the moat, the build-buy-consume decisions that fall out of the geometry rather than out of preference, and the specific events that would force a revision of the map. Written as a technical piece in which the map does most of the argumentative work.

Mapping the UK public good case: components, evolution, and the public data commons

A technical accompaniment to the Deliberation Meets Reality series, in which we analyse and position the platform as a UK-public good initiative.

Abstract

A Wardley mapping exercise that positions the fifteen components of the platform on the evolution axis from genesis to commodity, and reads the resulting landscape for a UK public infrastructure play. The map’s load-bearing commodity anchor is the UK public data commons (Ordnance Survey, ONS, Land Registry, Companies House, NUAR, and the FAIR data work of the Geospatial Commission), and its deliberate inertia sits on the Rust graph engine, which is held in place because the architectural principles it enforces are antibodies against failure modes observed repeatedly in production logistics systems. The post names the three climatic forces acting on the landscape, the seven strategic plays available in response, and the events that would force a revision of the map. The argument: grounded, inspectable decision support for physical operations is an under-provided public good, and the UK has the public data and academic open-source anchors to build it without importing a commercial stack.

Hypothesis register: collective flourishing under real-world constraints

A technical companion to the Critical thinking infrastructure essay series

Abstract

A formalisation of the arguments made across the essay series into fourteen testable hypotheses, each written as a null hypothesis (the perceived status quo), an ARIA alternative, and the position this work takes, with a critical test that would falsify the claim. Seven open self-critical questions are listed alongside the fourteen hypotheses, covering the places where the platform’s thesis is weakest and the evidence thinnest. The register exists so that the argument can be attacked on specific, named claims rather than on vibes, and so that the programme has a public ledger of what it is betting on and what would make it change its mind.

The social layer: modelling how decision-makers think about problems

The fourth post in the When deliberation meets reality essay series. The previous post argued that AI value comes from specialised models at specific points in a stable infrastructure, orchestrated by a general-purpose facilitator. This post examines the layer that has been hardest to formalise: how the facilitator learns the framings of the people a decision will reach, whether they are the few who make it or the many it will land on.

No single model will save us

A continuation of When Deliberation Meets Reality and Building Critical Thinking Infrastructure, which together argued that the missing precondition for collective flourishing is critical thinking infrastructure fast enough to survive contact with reality. This post, Part 3 of the Critical thinking infrastructure essay series, addresses where AI sits in that architecture, why no single model can take its place, and where the architecture goes next.

Abstract

The instinct in 2026 is to reach for one large language model and ask it to do everything. This post argues, from Wolpert and Macready’s no free lunch theorems and from seventy years of overpromise and correction in computational optimisation, that no single model outperforms all others across all problem instances. The sensible response is to invest in a stable substrate with pluggable specialised models rather than to chase a god-model. The post then works through where AI sits in the four-layer architecture: extraction and entity resolution at the spatial layer, graph neural networks and causal models at the domain layer, classical mathematical solvers at the compute layer, and a general-purpose LLM as the facilitator that translates questions into compositions of the above. The platform is the infrastructure; the models are replaceable.

Building critical thinking infrastructure

A companion to When Deliberation Meets Reality, which argued that the missing precondition for collective flourishing is critical thinking infrastructure fast enough to survive contact with reality. This post, Part 2 of the Critical thinking infrastructure essay series, describes what that infrastructure looks like.

Abstract

This post proposes what decision-support infrastructure must look like to compress time-to-question and time-to-answer enough to fit rigorous analysis inside the decision window. It describes a four-layer architecture (spatial knowledge graph, domain graph, solver integration, micro-tools) governed by nine principles. Three matter most here: separating stable spatial knowledge from volatile project data, keeping logic out of data objects, and treating materialisation as an economic decision rather than a structural one. The architecture draws on six years of commercial deployment in logistics planning. The post is a ledger: what has been tested, what has been designed, what remains an open research question.

When deliberation meets reality

A response to ARIA’s Collective Flourishing opportunity space and Part 1 of the Critical thinking infrastructure essay series.

Abstract

Researchers have studied collective decision-making methods for decades, from multi-criteria analysis to deliberative democracy platforms. Under real operational pressure people almost always drop them in favour of gut feel and spreadsheets. The field has plenty of methods. What it lacks is an explanation for why analytically superior methods fail when the stakes are real. Speed binds decision quality more tightly than rigour does: the constraint is the time it takes to ask a question and get a defensible answer. The missing precondition for collective flourishing is critical thinking infrastructure that runs fast enough to survive contact with reality and leaves a trace the collective can inspect.

Short-blogs

A bunch of short blogs and thoughts, in no particular order. Insults are mostly self directed.

80-20 rule in small vs big companies

In a small company, 80% of what you do doesn’t really matter. In a large company, there is an 80% chance that nothing you do matters.