Matt Burgess

It Doesn't Remember

Sun, 10 May 2026 20:30:29 +0100

Mark Cuban posted a question this week that’s been doing the rounds. Why can’t Enterprise AI guarantee the same answer to the same question, every time?

The standard answer: train a specialised domain model, add human-in-the-loop verification, log the audit trail. Not wrong. But it treats the model as the primary artifact. The inconsistency problem isn’t a model quality problem. It’s a memory architecture problem. To understand why, you have to go back to the hairball.

The promise was seductive and coherent. Enterprise knowledge is relational - things connect to other things in ways that flat tables can’t represent. A customer isn’t just a row; they’re a node connected to transactions, products, complaints, locations, household members, lifetime events. Google had demonstrated the Knowledge Graph in 2012. Neo4j, TigerGraph, Amazon Neptune, Azure Cosmos - a whole vendor ecosystem emerged. Gartner put it on the hype curve. Consulting firms built practices around it. Enterprise architects might have built roadmaps with knowledge graphs at the centre.

The vision was ‘connect all your enterprise data into a unified graph’. Traverse the connections. Discover the hidden relationships that siloed systems couldn’t see. Surface the insights that lived in the space between the data, not in the data itself.

It was right about the problem. That knowledge is relational. The connections do matter. The insight does live between the nodes, not in them.

The hairball

But what actually happened in the projects was this. You built the graph. You connected the nodes. You ran the queries. And you got a hairball - a visualisation so dense with connections that no human could read it, traversals so expensive that queries timed out, and insights so buried in the structure that extracting them required specialist skills almost nobody had.

The construction cost was brutal. Ontology design - deciding what the entities and relationships fundamentally are - was inherently contested. Every team has a different view of what a ‘customer’ is, what a ‘product’ is, whether a ‘transaction’ connects to a ‘customer’ or to an ‘account.’ Before you’d built the graph you had to resolve these questions, and resolving them required organisational alignment that most enterprises couldn’t produce. Projects ran 18 months before they got any value out. Most ran out of patience first.

Then the inference problem. That vision assumed that once you had the graph, you could traverse it and reason over it - the machine would find the connections that humans missed. In practice, semantic reasoning over large graphs was slow. The semantic web dream of machines reasoning over linked data never materialised at enterprise scale. What you got instead was graph traversal for well-defined queries - useful for fraud detection and recommendation engines, but not for the general ‘surface-for-me-some-hidden-insights-from-enterprise-knowledge’ promise.

The specific use cases that worked were narrow. The general knowledge management use case - ‘connect everything and discover what we know’ - produced expensive infrastructure, impressive B-of-the-Bang architecture diagrams, and maybe some dashboards nobody used.

By 2021 most enterprise graph projects had either narrowed to the specific use cases where graphs genuinely worked, or were quietly shelved. The hairball became shorthand for what happens when you try to make graphs do too much.

ChatGPT lands in November 2022. Within six months the enterprise technology conversation has shifted completely. LLMs can answer questions about connected knowledge without requiring explicit graph construction. Vector databases and RAG offer semantic retrieval without ontology engineering. The narrative becomes: you don’t need to build a graph to connect your enterprise knowledge, you just embed your documents and let the model find the connections.

Graph adoption plateaus. Projects that had been limping along get cancelled. The CTO who had championed the knowledge graph in 2019 is now championing the LLM platform in 2024, and nobody wants to hear about graphs.

The conclusion made complete sense. If graphs had been struggling to do the intelligence work for five years, and LLMs arrived and did it in an afternoon - why would you keep the graph? The reasoning was sound. The premise was wrong.

A retrospective

The 2019 knowledge graph was trying to be an intelligence layer. It was trying to do the reasoning, surface the insights, answer the questions. That’s hard, expensive, slow, and the intelligence was brittle. It was the wrong job.

The right job for a graph is not intelligence - it’s memory. Holding the structure that makes intelligence trustworthy. These are completely different design requirements.

An LLM does what a graph couldn’t. A graph holds what an LLM can’t.

Specifically, the graph holds things the LLM structurally cannot:

Typed relationships. Not ‘these two things are semantically similar’ - which is all a vector embedding (an LLM) gives you - but ‘A-funded-B-on-date-X-in-the-context-of-decision-Y’, and that-decision-was-made-by-person-Z-who-held-these-views-at-this-time.' The relationship type, its provenance, its temporal context. Semantic similarity collapses all of that into proximity in vector space. The graph preserves it as structure.

Chain of custody. The directed path from original signal to current artifact, with every transformation recorded as a named edge (a recorded connection). This signal came from this source, was interpreted by this person, was encoded in this system, and was modified at this crossing. An LLM has no memory of any of this once it’s processed the context. A graph holds it permanently.

Temporal history. What was true at time T; what changed at time T+1; what was lost in the transition. An LLM lives in a perpetual present - its context window is now, and what-came-before only exists if someone puts it in the window. A graph holds the history as structure, not as context.

Contradiction structure. Two signals pointing in different directions about the same thing, held explicitly in tension rather than collapsed into a single averaged representation. A financial frame says ‘the deal looks clean’ or an operational frame says ‘something is wrong’. In a vector space (an LLM) these produce a confused proximity. Whereas in a graph they’re two nodes with different relationship types to the same entity, and the tension between them is itself a named structure.

The missing signal. What was captured and then, wasn’t. What arrived and got lost in normalisation. The graph can hold an absence as a named thing - a node with no edges where edges should be, a relationship type present in one period and absent in another. An LLM cannot tell you about what isn’t there.

Cross-session continuity. A pattern confirmed in 2019 is explicitly connected to the contradicting signal emerging in 2026. An LLM session just has no access to this unless someone deliberately loads it into the context window. A graph makes it structurally accessible, because the 2019 confirmation and the 2026 signal are both nodes in the same persistent structure.

Wrong job. Right job

The hairball was a graph doing the wrong job. When folks (me included) tried to make the graph the intelligence layer - to force it to produce insight through its own traversal and reasoning - then we needed to connect everything to everything, because one doesn’t know in advance what connections will be meaningful. Voila. Hairball. Because comprehensiveness is the design requirement, and comprehensive connection at scale is visually and computationally intractable.

Change the design requirement from comprehensiveness to fidelity and you get a completely different graph - leaner, typed, provenance-rich. You connect what-you-actually-know-is-connected, with typed relationships, with provenance, with temporal context. The graph doesn’t need to be traversed by humans. It doesn’t need to produce visualisations. It just needs to be queryable by the LLM as a source of structured, grounded context. The LLM does the intelligence over that, doing the reasoning over something with actual provenance rather than over whatever-arrived-in-the-session.

A Seven-Year Itch

The LLM won the intelligence argument. Completely. It is the most powerful processing layer ever built for enterprise knowledge. But it doesn’t remember. The graph is the memory. It just spent a decade being asked to be the intelligence too. That failure obscured what it was actually good for. An LLM needs a memory layer to be trustworthy at the specific things that matter for decisions - provenance, chain of custody, contradiction structure, temporal history, the missing signal. Better intelligence makes reliable memory more important, because the intelligence is only as trustworthy as what it’s grounded in.

An LLM doesn’t make the graph redundant. It makes it more necessary.

So back to Cuban’s question. You can’t get consistent answers from a model that has nothing stable to read from. A domain-specific LLM answer still treats the model as primary. A graph-as-memory-layer answer treats the original signal as primary - and gives the model something it can actually be consistent about.

I built Stratum. If the LLM needs a memory layer to be trustworthy, the question is: what’s worth anchoring it to? My answer, after a long time looking at where signal gets lost, is the customer. Not because it’s the only domain - but because the gap between what a customer expresses and what survives to the decision meant to serve them is the widest, least visible, and most expensive gap in most organisations.

The expression is the immutable coupling. Customer-researcher interpretation, product response, commercial viability all coexist on the same graph in typed registers - multiple disciplines reading the same evidence without any one view replacing the others. Production teams stay tethered to the specific customer whose words shaped the work. Not because she’s the point - the firm builds for customers like her, across millions of interactions. And those decisions at scale are only as trustworthy as the specific, honest signal underneath them. Lose the verbatim, lose the ground truth.

The architecture doesn’t replace a researcher’s judgment. It holds it - across every team, every crossing, every decision downstream. The architecture is the guarantee, not the model.

Follow along: mattburgess.micro.blog/subscribe… · mattburgess.micro.blog/feed.xml · micro.blog/mattburge… · Mastodon @mattburgess@micro.blog

You're making decisions on bot data

Thu, 07 May 2026 10:38:33 +0100

Lumen CEO says AI bots are taking over the internet It’s a good line. It’s also a PR piece - Lumen sells the infrastructure that carries internet traffic, so they have a reason to make you anxious about what’s on it. But strip the self-interest out and the data underneath holds up.

Cloudflare, which processes a substantial fraction of global internet traffic, found that automated traffic grew at 23.5% year-on-year in 2025 - eight times faster than human traffic. Imperva, tracking bot activity for over a decade, recorded 2024 as the first year non-human traffic overtook human traffic outright.

Most of what is on the internet is now machine-made. This matters for customer insight - and not only in the ways you might expect.

Start with the quant. NPS scores. Customer satisfaction surveys. The numbers that entire teams use to set targets, justify investment, and claim progress. Those reach customers digitally - by email, by SMS, through apps. If half of internet activity is now automated, what proportion of your survey responses are genuine? No one knows with any certainty. If they’re contaminated, then the infrastructure built on those contaminated numbers is increasingly unreliable - and AI systems processing that data to generate summaries and recommendations are amplifying the problem, not diagnosing it. Your decisions are downstream of bots.

And where the quant is now contaminated at source, the qualitative is getting lost in AI-transit. Good research comes from genuinely engaged customers - people who have something to say because they care, not because they’re compensated to attend. The research challenge isn’t just finding those customers. It’s holding what they express in a way that survives the crossing into decision.

Every AI intervention from insight to action - the transcription, the auto-theming, the synthesis, the summary - is a processing step: a riff on a riff. Maintaining signal fidelity was always hard. AI just made it harder.

If what’s upstream has already been summarised, the AI might produce something coherent. It may well produce something with the texture of insight. It will not produce something tethered to what a customer actually experienced.

This is what I mean by customer coupling: the relationship between what a customer expressed and what eventually informs a decision. In a world where AI mediates more of that chain, the coupling degrades faster - each processing step is a translation, and translation loses something. The original signal, skillfully obtained, is anchored in something real. AI processing a summary-of-a-summary is not.

Something genuinely valuable was expressed. The hesitation, the aside, the thing that didn’t fit the question. Whether any of it survives is another matter.

As AI becomes the processing layer for everything, that original signal becomes simultaneously more valuable and more endangered. More valuable because it’s the only ground truth available. More endangered because the incentive to let AI handle it - summarise it, riff from it, approximate it - is overwhelming and accelerating.

What you do with that signal before AI touches it determines the quality of everything AI produces from it.

The architecture for holding customer signal matters more now than it did two years ago. Not less.

Follow along: mattburgess.micro.blog/subscribe… · mattburgess.micro.blog/feed.xml · micro.blog/mattburge… · Mastodon @mattburgess@micro.blog

It's not the AI. It's the coordinates.

Fri, 24 Apr 2026 11:13:47 +0100

I wrote about Dave Plummer and Tempest - a 1981 arcade game where Dave spent a year trying to improve his AI before realising the problem wasn’t the AI. It was the way he’d taught it to see the game.

Tempest (Video Arcade Game, 1981)

Tempest is a shooter played from the centre of a geometric web outward - enemies crawl in from the edges and you spin around the rim to meet them. Dave taught his AI to play it using Cartesian coordinates - X and Y positions on a grid. The AI got good. It kept getting better. And then it plateaued, hard, and stayed there. It wasn’t until Dave gave it a completely different way to be aware of the game - polar coordinates, angle and distance from the centre - that something shifted. The same AI, the same game, the same enemies. Suddenly it didn’t just improve. It leapt. Because the coordinate system it had been given finally matched the geometry of what it was actually doing.

There’s more in it than a gamer-AI anecdote.

It demonstrates something that sounds intimidating when you name it properly. The coordinate system isn’t neutral. It doesn’t just describe what’s there - it determines what questions you can ask. Cartesian coordinates generate Cartesian questions. Polar coordinates generate polar questions. And if the situation you’re navigating is polar in its nature, Cartesian questions won’t find the answer no matter how well you process them.

Philosophers call this ontology - the study of what is real and how it can be described. (I know, I know. I wrote a whole post about why that word doesn’t need to be as scary as it sounds.) The short version: an ontological register is just a coordinate system for reality. And the interesting thing (the liberating thing) is that the same reality can have multiple valid coordinate systems simultaneously, none of which is complete on its own.

A deal seen through a financial lens and a deal seen through an operational lens aren’t two descriptions of one underlying truth. They’re two coordinate systems, each generating different questions, each making different things visible. The financial register asks: does this stack up? The operational register asks: can we actually do this? Both questions are real. Neither exhausts the deal.

If you’re working with AI systems, low confidence scores aren’t just a verification problem.

When a model returns low confidence, the instinct is: something went wrong, add more, verify harder. But there’s another reading. The model processed everything it could in the register it was given - and then correctly reported that something remained outside it. It’s not failing. It’s pointing. The register ran out.

That’s not a gap to fill. That’s a gap to see differently.

The Tempest AI didn’t need more training. It needed a different map - one drawn for the terrain it was actually crossing. The question for any AI-assisted system isn’t just: did it get the answer right? It’s: did it have all the registers to ask the right questions - or just one register and a good vocabulary?

Change the coordinates. Different questions become available.

Follow along: mattburgess.micro.blog/subscribe… · mattburgess.micro.blog/feed.xml · micro.blog/mattburge… · Mastodon @mattburgess@micro.blog

Signal Debt: what would it mean to let good signal in?

Wed, 22 Apr 2026 14:36:39 +0100

Ward Cunningham coined the term technical debt in 1992 to describe the cost of choosing a simpler solution today that you’ll need to revisit tomorrow. It was a precise metaphor for a specific problem in software development. Then it escaped the engineering department and went mainstream - usually as a complaint. The system is slow, the feature is late, the codebase is a mess. Someone says technical debt and everyone nods as if that explains it.

Jabe Bloom’s framing of technical debt is one of the most clarifying. Technical debt isn’t a failure. It’s the natural cost of building under uncertainty. You make the best decision available with what you know, you ship, you learn, and you carry the cost forward. The debt isn’t the problem. Losing track of it is.

But every build is downstream of something. A decision. And every decision is downstream of something else - the quality of what your organisation thinks it knows about its customer.

So there is an equivalent debt. I’m just going to call it signal debt.

It doesn’t live in the codebase. It lives upstream of it - in the research, the synthesis, the insight decks, the quarterly themes. Everything that crossed from the outside world into your organisation before it ever reached a decision. Unlike technical debt, nobody decided to take it on as a debt. You don’t get anything for it. When a customer truth degrades in transit you don’t move faster or ship sooner. You just lose it. Nobody is accounting for it - nothing in the operating model is designed to.

Most organisations consult the customer at the start. Then build whatever they were going to build anyway. Not out of malice. Not out of laziness. The research happens. The interviews get done. The insight-driven strategy deck lands in inboxes across the organisation. And then gradually, locally, rationally, something else takes over. The transformation programme has momentum. The product roadmap has commitments. The data platform has a delivery date. The customer enters the process as a presence and exits as a percentage. By the time the decision gets made, the signal that was supposed to inform it has crossed so many boundaries - from researcher to analyst, from analyst to deck, from deck to meeting, from meeting to ticket - what arrives is the organisation talking to itself about what-someone-outside-once-said. It has forgotten what it has forgotten.

This has been happening for years. And AI is about to make it very expensive. Which might - finally - be the thing that forces a reckoning.

Every handoff or crossing destroys a little of the common ground (Klein via Bloom) - the shared understanding that lets people appear rational to each other even when nobody holds the full picture. Nobody is doing anything wrong. But by the time a signal reaches a decision, what-the-customer expressed is inaudible. Signal debt is what accumulates in the gap. And unlike technical debt, it can’t be refactored. When code is reclaimed, the logic comes with it. When a customer signal gets flattened, the thing that made it true is gone. There’s no undo function in the enterprise.

The customer estate - those teams organised around customer knowledge - is built to activate signal, not carry it. Look at the 2025 martech landscape for crying out loud. 15,000 products. 100x growth in fifteen years. Built for activation, not fidelity.

And because teams organise around activation they make locally rational decisions - what to capture, what to summarise, what to pass forward. Each step makes sense. None of them owns the crossing. (this is boundary objects again, but that’s for another post). The underlying architecture makes signal debt almost inevitable - systems that don’t speak to each other, data that doesn’t travel with its context, handoffs with no mechanism for checking what-arrives against what-was-sent.

And on top of all of that is something more dangerous still. “Customer”, the word, is doing too many jobs at once. It’s everything everywhere all at once — the strategy decks, the data platforms, the service models, the call centres, the slideware.

[Everything Everywhere All At Once]

And because it’s everywhere, nobody questions it. It becomes a flashing neon sign rather than a person or persons that once-upon-a-time had a story to tell. Calling something ‘customer-this’ or ‘customer-that’ justifies internal momentum rather than being tightly coupled to something real arriving from the outside. It’s a multi-stakeholder activity where the customer estate plays exquisite corpse. Each function draws their section, folds the paper, passes it on. Nobody sees the whole. The risk isn’t that we disagree about the customer. It’s that we stop knowing we do.

[Ted Joans, Exquisite Corpse (1976–2008). 132 contributors across 32 years. MoMA]

And the customer themselves? The personhood that sees and fears and feels your product service experience - distant. Structurally remote. Getting to them - the fieldwork, the research, the synthesis - takes weeks, months. Everyone knows this. It’s just what the work involves. So knowledge gets batched. Commissioned, completed, distributed. Each snapshot becomes a million-parts-diluted version of ‘the customer’. That the organisation then acts on until the next one. And everything that follows - the commercial modelling, the transformation programme, the product operating model, the data platform, - flows from it. If the signal was thin when it was captured, or thinner still by the time it reached the decision, none of that changes anything fundamental. Nobody noticed that the input got degraded.

Leaders feel this without being able to name it. Something is slow that should be fast. Something is uncertain that should be knowable. The organisation is busy and the customer is still somehow not quite present in the room.

That’s signal debt at scale.

And now, ladies and gents, we have AI.

So what-the-organisation-thinks-it-knows-about-its-customer isn’t what-the-customer-expressed, but, yep, that’s what AI is running on. AI is being bolted onto the customer estate. Onto the CRM. Onto the insight function. Onto the content pipeline. Onto the decisioning layer. And because the output is fluent - structured, confident, plausible - it feels like progress. Unfortunately AI doesn’t repair signal debt. It makes it worse. Whatever model or fragment of ‘the customer’ it runs on, it then runs for you, at speed and at scale. If that model is thin or batch-processed or is the product of that exquisite corpse assembled across disconnected functions - then the output will be wrong in ways that are very hard to see but extremely easy to distribute.

The problem, in this situation at least, isn’t AI - it’s what AI is fed.

A genuinely high fidelity customer truth - qualitative, longitudinal, persisting across the crossings rather than degraded by them - would change what’s possible entirely. That’s not where most organisations are. And while signal debt goes unacknowledged, the AI initiative is just the fastest way yet to scale a polished, high-confidence mistake.

The architecture for signal fidelity exists. It has been proven. In operational environments - supply chains, logistics, fulfilment - observability is engineered into the stack. There the signal stays live and checkable at each crossing, not just at the end. Nothing arrives as an autopsy. When a store needs milk, the signal is clear and the supply response matches it. Nobody is arguing about whether the data can be trusted. That rigour isn’t happening at the customer boundary. Upstream, where the customer signal should be at its most vital, the transductive architecture is absent. Same organisation. Different estate. The crossings are unowned, the handoffs unchecked, the debt accumulating silently.

It isn’t that the problem is unsolvable. It’s that it isn’t a process failure with a root cause. It’s a condition - distributed across every crossing, diffuse by design. That’s why the answer isn’t better monitoring. It’s better observability.

Signal debt won’t appear on a balance sheet. It won’t show up in a sprint retrospective or a quarterly business review. It accumulates quietly - and divergently. The same compromised signal fans out across the organisation simultaneously: into the commercial model, the product roadmap, the data platform, the content pipeline, the customer insight function. Each team working from their version of it. Each one now with an AI assistant, running on that same thin input, at speed, at scale, confident in what arrived, unaware of what was lost. Not a single-point-of-failure-compounding-in-one-direction but a structural liability replicating across every surface.

What’s needed isn’t a new methodology - it’s a different relationship to the signal. Keeping faith with what was meant, not what survived the crossing.

Next time: what it would mean to carry a signal you could actually trust.

Follow along: mattburgess.micro.blog/subscribe… · mattburgess.micro.blog/feed.xml · micro.blog/mattburge… · Mastodon @mattburgess@micro.blog

Transduction - to lead across

Wed, 15 Apr 2026 19:50:03 +0100

Fancy a bit of ‘vector logic’? No? Never mind - you’re already doing it.

You arrived here a few seconds ago and made a snap judgment: (“This looks worth clicking on”) That’s abduction - a bet on the best available explanation for why you clicked. Now, as you read this sentence, you’re inducting: scanning these words and lines to see if the pattern holds. (“Hmm, jury’s still out”) If you already know your Latin, you may have deduced where this is going before you even reached the end of this paragraph (“This is going to get philosophical. And yet, here you are”)

Three modes of reasoning. You’ve just used all of them.

Ducere is Latin for “to lead” - and the root is hiding in plain sight. Deduct. Induct. Abduct. Each prefix names the direction your mind is being led along.

Deduction leads down (de-). From a general rule to a specific conclusion. The strategy deck. The mandate from above. You already know this word: you de-duct on your tax return.

Induction leads in (in-). From many specific observations to a general truth. The survey, the NPS score, the quarterly themes report. Your kitchen hob works the same way.

Abduction leads away (ab-). From a lone observation to the best available explanation. The design sprint hypothesis: “We think the problem is X - let’s test it.” The everyday version is darker: to abduct is to kidnap - same root, only the consent differs. You’re essentially stealing a conclusion before you’ve earned it with proof.

Three directions. Same root. Different coordinates.

Then there is another ‘duction’ - one that doesn’t point down, in or away.

Transduction leads across - across registers, across disciplines, across the boundaries where meaning changes form in transit. It is the act of carrying structural information through the crossing without losing it. The Latin root is old. But the word’s modern life is largely a 20th century story. You’ll find it in the specialised lexicons of genetics, electrical engineering and signal processing; Piaget borrowed it for a specific kind of reasoning and it is the cornerstone of Gilbert Simondon’s philosophy. Manuel DeLanda followed, and is arguably the most useful on it — tracing how new structures emerge through intensity rather than being handed down from above.

And yet it never crossed into the everyday vocabulary of organisations — which is strange, because a hand-off is one of the most discussed failure points in any team, any process, any transformation programme. We know where things break. We just haven’t had the word for what the crossing actually involves. Nobody talks about transducing a brief into a build, or transducing a strategy into a team. Nobody says they’re off to transduct a ball around a five-a-side pitch.

Every time Jurgen Klopp described why his setup at Liverpool worked — and after this season, we’re allowed to be misty-eyed — he was pointing at transduction, even if he never named the logic. There was a name for the tactical application, of course: Gegenpressing. But Gegenpressing was a term for the grass; it described what the legs did. The underlying logic was transduction. It wasn’t about a homogeneous decree or reaching a consensus on what to do next. The shape was emergent, not prescribed. The movement of the first player didn’t dictate exactly what the second had to do; instead, it transduced the ‘state’ of the game to the rest of the team. One player shut down the ball-carrier, which changed the meaning of the space for the player behind him, who then cut the passing lane. They weren’t doing the same thing, but they were all maintaining a similarity of relations to the ball. You couldn’t coach that by mandate. You could only create the conditions and let the signal transduce.

Most organisations haven’t got there. Not because the argument is wrong, but because you can’t work on what hasn’t been named yet. Klopp, of course, proved you can build the reality without knowing the word. But he had the advantage of the pitch - a high-bandwidth environment where the signal is visible and the feedback is instantaneous. In the enterprise, the signal is invisible, buried in the professional registers of different departments and teams and architectures. He could point at the grass; you have to point at the logic.

The arguments for flatter structures, cross-functional teams, value streams - the aspiration is right, and the rationale still holds. Fewer layers between the person who knows what-the-customer-meant and the person deciding what-to-build means fewer membranes for the signal to cross. That’s a signal fidelity and transduction argument. It just hasn’t been named as one.

Take the current state of the art: Suzanne Kaiser’s architecture for flow (What is it about the Germans?) Her work, like Klopp’s, builds the conditions for this movement. Neither uses the word, yet the word explains why they work. If a system in an enterprise is actually functioning, it is probably transductive.

The enterprise tried the Klopp move: it installed the value stream and rebranded departments as tribes, then waited for the magic. It never came. The org chart remained the most legible thing in the room because while the components moved, the logic didn’t. The mandate won. You can deploy squads, or standing teams but if the signal is trapped in a siloed professional register, the teams are just ‘standing-around’ in new positions waiting for a ball that never arrives. Nobody named building-the-conditions-for-that-signal-to-transduce as the actual work - and by conditions, we don’t mean breakout spaces and free coffee. We mean the structural capacity for the signal to survive the crossing.

It’s not about better translation. Think of a record deck. The needle doesn’t ‘translate’ the grooves of the vinyl into music - translation assumes the meaning is already stable and just needs a new label. Instead, the stylus transduces the physical topography of the record into an electrical signal. It is a fundamental change of substrate. If the needle is blunt or the tonearm is poorly weighted, the meaning of the recording - the depth, the timbre, the nuance - is lost at the point of contact. If a signal even makes it to the speakers.

What is also rarely named: working across domains requires negotiation. When two people from different parts of an organisation meet in a value stream, they don’t automatically see the world the same way. That negotiation isn’t just about ‘getting along’ or winning. It is a specific, technical struggle: What is the signal that comes out of this exchange? When a researcher meets a product manager, they are negotiating which parts of the customer’s nuanced reality will survive the crossing into a roadmap. When an architect meets a developer, they are negotiating which structural principles will survive the crossing into code. Ideally they are trying to ensure that a core signal doesn’t evaporate as it moves from one professional language to another (which leads us toward boundary objects, but that’s for another post).

This is the transductive work: carrying signal across these boundaries while keeping the meaning of the information intact. But when leading down is more legible than leading across, the mandate wins by default. We default to the decree because we haven’t designed the crossing.

If you’ve built observability into your engineering stack, you’ve already built transduction architecture - the signal stays live at each crossing, nothing arrives as an autopsy. Same in a well-designed supply chain: each handoff is engineered for fidelity, not just for the convenience of the next stage. You know what it looks like when the crossing is owned. The question is why that same rigour doesn’t extend to customer signals. Every handoff there is a courtesy to the next person in the flow - optimised for their inbox, not for what survives. The crossing has no owner. No design. No budget line. Not because the capability isn’t there. But because nobody named it as the same problem. Having the word changes that. Not because naming something magically fixes it - but because you cannot design for a phenomenon that is illegible to your system of logic. Transduction is the crossing. The crossing is where things go wrong. Now you can point at it.

And now, there’s AI.

The act of prompting a language model is pure transduction. Intention crosses into language. Language crosses a membrane into a different substrate. Something returns that carries the structure of what you meant - or it doesn’t. The output is fluent and confident regardless of what was lost in transit. That’s the problem. Fluency is not fidelity. The stakes are higher. Not lower.

An AI summary that quietly erases the customer’s most vital friction is worse than no summary at all - because it doesn’t look like a loss; it looks like an answer. Every knowledge worker navigating this daily is doing so inside structures that weren’t designed for it - structures that default to deduction, that optimise for the next stage, that mistake fluency for fidelity. A team that prompts well but feeds the output into a deductive process hasn’t solved anything; they’ve just moved the crossing one step earlier. The signal still loses.

How would Klopp do an enterprise? He said it himself: “No playmaker in the world can be as good as a good gegenpressing situation." He wouldn’t hire a smarter analyst. He’d build the conditions.

More on that - what I’ve been building and what the first artifact looks like - in the next few posts.

Follow along: mattburgess.micro.blog/subscribe… · mattburgess.micro.blog/feed.xml · micro.blog/mattburge… · Mastodon @mattburgess@micro.blog

The Elephant in the Room Has No Name

Thu, 26 Mar 2026 16:13:42 +0100

It is almost a requirement of the genre that any discussion of enterprise fragmentation begins with the parable of the Blind Men and the Elephant. You know the story: one touches the trunk and proclaims it a snake; another touches the ear and calls it a fan; a third touches the leg and insists it is a tree, it’s tail is a rope and so on. They aren’t wrong; their ontologies are coherent but bounded. It is the whole that is, as a consequence, fragmented.

In the modern firm, we have industrialised this blindness and turned the parable into a permanent operating model. Designers touch the user journey. Engineers touch the tech stack. Finance touches the cost centre. Each silo describes a different thing, and the Boardroom is left trying to mandate efficiency and value release to steer a thing it can neither see, nor name.

1. Analytical Autopsy

One observes a repeating pattern since the mid-2010s. Digital transformation has largely been an exercise in crossing fingers and making a wish.

Firms have spent trillions globally on going digital genuinely without much of an understanding of where or how that transformation would occur and where it would end up. Somehow, the elephant was just going to emerge.

But the friction runs deeper than a few failed projects. It stems from a way of processing reality that predates digital entirely, the analytical autopsy.

Driven by industrial need for linear outputs and annual budgets, firms can’t help but flatten customer intent - analysing research and feedback until customer intent is gone. Teams in production workflows then have to ingest those inert parts. Agility and product ways of working haven’t much hope of fixing this; they simply speed up the consumption of those autopsied products. There is still no shared understanding of how the parts works together in the cause of the whole firm serving the whole customer.

2. A Topology of Value

A way out isn’t a better roadmap or more integration. It’s a recognition of the actual Topology of Value. A firm may represent itself as an Org Chart. But its topology of value is an Assemblage.

The Org Chart is a map of command and control. If one’s primary concern is status, or the preservation of a rigid hierarchy, then the pyramid is a perfect topology. It is designed for reporting and oversight. But if one cares about value, the pyramid is a policy constraint. It forces a focus on local optimisation making silos efficient - but strangling global throughput of customer intent. Value involves a different map, an Assemblage.

To describe it in its very simplest terms, an assemblage is a functional-composition-of-heterogeneous-parts. It is an intensive sociotechnical entity where the parts (people, code, data) retain their own properties and their ‘living’ signal, and are synthesised into a functional whole. Right now, this composition is invisible to the boardroom. It persist in the shadows, sustained by informal workarounds and heroic individual efforts that are never funded or formalised because they don’t fit on the grid. But at least you now have a name for it. And fortunately, it is not just a label, it has its own rigorous dynamics.

Where the Org Chart is a diagram of static boundaries, the Assemblage is a map of flow. By naming the Assemblage, one makes the topology of value across the firm legible. One stops trying to manage a collection of inert parts (the consequences of that analytical autopsy) and one starts managing a synthesis.

3. The Neutrality Trap

We often treat our corporate tools and structures as neutral objects. We fall into the trap of thinking, ‘the process doesn’t run the company, people do’. On the face of it, this is true. A CRM or a project framework sitting idle is just a dormant set of rules.

What this fails to recognise is that when a person is plugged into a specific structure, they are fundamentally changed. They become a person + tool + environment unit. This unit - the Assemblage—has its own agency. It transforms the people within it, shaping habits, practices, and subjectivities. It makes certain outcomes (like siloed thinking or slow approvals) far more likely than others. The Assemblage is never neutral, and it is never passive. It has a dominant propensity. If you inhabit a modern enterprise, you should already recognise this.

Look at the diagrams of the most valuable sociotechnical practices being shared and carried out right now - whether it is Team Topologies, Value Stream Mapping, or Domain-Driven Design. These are not just new ways of working; they are the first attempts to diagram and manipulate the Assemblage(s). They are attempts to move from the internalisation of siloed knowledge to the externalisation of a shared flow. They are naming and framing the domains, boundaries, and flows that keep the ‘molar’ (the hard, structural stuff) from crushing the ‘molecular’ (the wet, living signal of the customer). This, the Assemblage, is the strategic language of the new sociotechnical enterprise. It is the thing the firm couldn’t see and couldn’t name.

4. A Strategic Scaffolding

This is why we have been insisting on the term Assemblage. It is the foundation of a strategic scaffolding and the deliberate construction of the environment that authorises new habits. A dynamic, adaptive topology that provides coherence without stifling the autonomy of the parts.

Operating as an Assemblage unlocks a spectrum of capabilities: from precision of resource allocation to compression of the feedback loop, all while transforming around the integrity of the customer signals from first insight to final output. It defines the firm’s capacity to affect and be affected by the market. And it is the difference between a firm that is stuck in its own internal representation and one that is finally understanding its own topology of value.

Foundations & Further Reading

For those wanting the more on the Assemblage and scaffolding, the following sources provide the primary theoretical and scientific rigour:

Manuel DeLanda: Specifically his Lectures on Assemblage Theory, which provide the materialist logic for the capacity to affect and be affected.

Dave Snowden (The Cynefin Co): On Scaffolding and the shift from fail-safe to safe-to-fail experiments in complex domains.

Emergent Futures Lab: A contemporary and highly accessible diagnostic on The Neutrality of the Assemblage.

Ben Zweibelson (Military Design/JSOC): On the use of a Tornado Metaphor to Build an Assemblage Concept to diagram the kinetic and non-linear flow of assemblages in high-stakes environments.

Gilles Deleuze & Félix Guattari: A Thousand Plateaus The nonlinear emergence of the Assemblage (agencement). This is the definitive materialist critique of the Org Chart, though it is written with a deliberate, maddening resistance to being managed. It’s less a textbook and more like a terrain. If you have the appetite, go straight to Plateau 12: Treatise on Nomadology, which contrasts the static, bureaucratic state apparatus (the Org Chart) with the war machine - a fluid, goal-oriented assemblage designed for speed and reconfigurability.

Source: London Review Bookshop.

A Cheat Sheet: Marc Ngui’s Illustrated Plateaus provides a visual, topological breakdown for those who prefer the blueprint to the prose. Ngui’s favourite (and most cited) illustration for this is the Plateau 1: The Rhizome diagram. It depicts a system that has no central core but is held together by its connections - a map of flow and a visual blueprint for the strategic scaffolding described in this post.

If you want to track how one actually understands the dynamics of the assemblage, as a topology of value, you will find my ongoing explorations and blueprinting here and on LinkedIn.

Taming the Tempest: What a 1981 arcade game teaches us about the topology of AI innovation

Wed, 11 Mar 2026 21:39:00 +0100

We are currently living through an era of AI brute force. Most of us are stuck in a cycle of blind prompting - throwing more adjectives at a Large Language Model, adding more compute, and hoping that if we just vibe hard enough, the machine will eventually spit out a miracle. We treat AI like a black box to be bargained with, rather than a system to be navigated.

But last week, Dave Plummer (the legendary Microsoft engineer behind the Windows Task Manager) posted a video that should be the North Star for anyone trying to scale AI strategy. He didn’t just beat his own world record in the Atari classic Tempest; he broke through a year-long performance plateau by making a fundamental shift in his conceptual geometry.

To be clear: Dave didn’t use the lexicon I’m about to describe. He arrived at his brilliant breakthrough through world-class engineering intuition. However, Dave’s story is the perfect vehicle for demonstrating a powerful, strategic, but latent lexicon: the philosophy of Deleuze and his modern-day interpreter, the philosopher and polymath DeLanda.

By adopting their vocabulary - specifically the mechanics of Assemblages, Intensities, Territorialisation and Singularities - we move beyond the vibe-coding of the amateur and begin engineering breakthroughs with the surgical clarity of a Tier-1 strategist.

This isn’t just about using fancy terms; it’s about having a precise map for how systems actually evolve and break through plateaus. Here is how that Tier-1 lexicon applies to Dave’s Tempest story:

The Heuristic: Dave’s Wall and the Polar Breakthrough

For a year, Dave’s AI was ‘competent-ish.’ It played like a talented human but hit a hard ceiling - a Plateau. Dave realised the problem wasn’t a lack of data; it was a mismatch of worldviews. The AI was trying to navigate a circular, 3D tube using a flat grid map. It was like trying to sail the globe while insisting the Earth is flat.

The breakthrough came from two specific shifts:

Polar Mapping: He stopped forcing the AI to see ‘X and Y’ and let it see ‘Angle and Depth.’
Prioritised Replay: He stopped letting the AI treat every second of footage as equal and forced it to obsess over the ‘surprising’ moments of failure.

Here’s how that scales from a niche optimisation to a universal strategy for innovation.

Phase 1: Joining the Assemblage

Before code was even written, Dave is immersing himself in the body of the game. He didn’t just treat the AI as a separate tool; he formed what Deleuze calls an Assemblage. To be overly reductive, an Assemblage is a-thing-composed-of-lots-of-different-things-that-can-still-be-decomposed and in this case, it is the functional heterogenous group where Dave (a human), the code, and the environment (the game’s memory) became a single moving entity.

Dave achieved this through an A-signifying Rupture. Most vibe-coders look at the Signifiers - the pixels on the screen, the ‘visuals’ of the game. Dave bypassed the skin of the game entirely. He didn’t care about what the game looked like (the representation); he hooked directly into the RAM (the raw intensities). He went beneath the surface meaning to engage with the Machinic Phylum - the raw data streams of the 6502 machine code.

The Strategic Pivot: Success in AI isn’t really about ‘using a tool’, it’s about how you seamlessly integrate the AI into the actual physics of your project or business. If you treat it as an outsider, it will always hallucinate an outside perspective.

Phase 2: Moving from Striated to Smooth Space

Once inside the system, Dave realised he had trapped his AI in Striated Space - a rigid, artificial grid. He was feeding it data in squares, but the game was a circle.

By switching to Smooth Space (the Polar Coordinates of the Tempest tube), he gave the AI legibility. This is the principle of Decalcomania vs. Mapping: Most AI projects at the moment are ‘decalcomania’ - they try to trace a pre-existing image (the grid) onto a problem. ’Decal’ as in mass-produced art transfer. Dave stopped tracing (copying the past) and started mapping. He allowed the AI to navigate the multiplicity of the tube - where every lane and depth is connected in a continuous web rather than just a series of isolated boxes.

The Strategic Pivot: Stop ‘blind prompting’ and start mapping the real ‘geometry’ of your problem. Are you asking your AI to solve a circular problem using square instructions?

Phase 3: The Nomad Scientist and the Difference Engine The real magic happened when Dave shifted his methodology from that of a Royal Scientist to a Nomad Scientist.

The Royal Scientist (The Grid-Binder) seeks the Universal Law. They want a model that handles the average case perfectly. They treat data as Extensive - focusing on volume, length, and “more-ness” (e.g., 15 million frames). This approach hits a plateau because it is obsessed with the typical.
The Nomad Scientist (The Singularity-Chaser) was Dave’s breakthrough. The Nomad follows the Singularities - the accidents, the surprises, and the TD Errors. This is what DeLanda calls Intensive thinking.

By building a Prioritized Replay system, Dave created a Difference Engine. DeLanda argues that change is driven by intensities - differences in pressure or ‘surprise.’ Dave realised that Extensive volume hides Intensive truth. He stopped trying to boil the entire ocean (a waste of compute) and instead built a Difference Engine that only applied heat to the points of highest pressure.

[Marc Ngui, Drawing A Thousand Plateaus, Introduction paragraph 8c]

Prioritizing surprising frames, Dave created a Line of Flight. He allowed the AI to escape the boring, repetitive loops of average play and focus only on the Singularities - the high-intensity moments of near-death.

The Strategic Pivot: You don’t need more data, you need intensive data. Stop seeking the universal averages, the Extensive, the low-pressure, high-volume, and energetically expensive. Shift to focusing on the Intensive, the high-pressure singularities.

Phase 4: The Folding of Time and Becoming-Superhuman

By focusing on high-intensity singularities, the AI underwent what Delueze and DeLanda call Deterritorialization - shattering the ‘human’ performance benchmark to enter a state of Becoming-Machine. Dave’s prioritised buffer acts as a Deleuzian Fold, where the most surprising moments of the past are folded directly into the present. This creates a Rhizome between the code and the game’s internal geometry. The AI ceases to play Tempest as an outsider, it becomes an extension of the tube itself, navigating the smooth space of the machine code with a brutally correct precision that connects any two points in the game-space instantaneously.

Why this Vocabulary is Your Secret Weapon

You might well ask: Why do I need these fancy French words? Well, because words like Assemblage and Singularity aren’t just jargon, they are navigational coordinates. And to be provocative, what else have you got? What language does one use that encompasses the dynamic and spatial and ontological? The properties and capacities of human and machine that we are witnessing?

If you’re at a point over lunch and you say, “the AI is acting weird,” you have no solution. But if you say, “The model has plateaued because we’ve trapped it in a striated grid; we need to prioritise its replay of high-intensity singularities,” you have a very clear articulation and enunciation of a situational awareness. You have strategy.

Digital and internet terminology became ubiquitous because we needed a way to talk about the (then) new world. In 2026, we need a Cartography of the Machine and Deleuzian philosophy and its remarkably lucid terminology - refined by DeLanda - is the strategic operating system for AI innovation (and a whole lot more). It’s not only thoroughly thought-through but it’s a stunningly appropriate topology for where-we-are and where-we-might (or might-not) want to go.

Dave Plummer didn’t just beat the game; he showed us that when you align your thinking with the true geometry of a problem, the Superhuman results follow as a matter of course. So the next time you’re stuck, don’t just add more prompts. Ask yourself: “Is your model tracing a habit or mapping a flow?"

[Machinic Assemblages of Desire, Barcelona 2016]

Service Design in the Machinic Age: Can we use the machine to do better reconnaissance?

Wed, 11 Mar 2026 16:46:58 +0100

Service design is hard. I’ve worked with brilliant practitioners who spend their days tracking down disparate parts of an enterprise just to pull a legible picture of the system together. It is a role fraught with peril, often forced into a ‘compliance’ mindset - tasked with making what the system should do look plausible, rather than being empowered to surface what the material reality of the situation actually necessitates.

The Service Blueprint is our most worthy attempt to bridge this. It is intended to connect the Front Stage (the customer journey) with the Backstage (the technical stack). But because it is so effortful to generate, the meaning-gap between these two silos is a chasm.

Why Blueprints are Hard

A Service Blueprint is not a ‘drawing.’ It is a high-fidelity reconciliation of a thousand contradictions. To build one, you must manually track the Materiality of the System - tracing a customer ‘click’ through five APIs and a legacy database that hasn’t been documented since 2012. Without this, the designer is denied the literacy required to actually form the material they are designing.

Tethering Intent to Materiality

Product ways of working and enterprise agility are theoretically positioned to incrementally fix some of this, but progress is glacial. Often because there simply isn’t the situational awareness to ‘simultaneously hold twenty-one moving parts in one’s head’. A Service Blueprint provides the necessary ‘technical reconnaissance’ - the ‘spatial memory’ anchoring the Why to the How.

In most enterprises, a customer journey map - the molecular expression of a thin slice of interactions - usually has a startlingly empty Backstage row. Conversely, the molar expression - the technical architecture - is located deep down away from the customer (its reason-for-being). Systems allegedly exist to serve the customer, but the reality is often a process of sedimentation; layers of new tech built on old workarounds, bleaching away original purpose.

This leaves an ontological void where the connection to the customer should be.

I got to thinking: what if we used the machine to weave these disparate ontologies into a shared plane for the first time?

I ran an experiment using Gemini as a sort of Machinic Reconciler to catalyse the Customer Journey and the Technical Stack to individuate together. By scaffolding the tech stack against the customer journey, we synthesise a Metastable Assemblage: a functional and expressive whole where front and back stage are structurally coupled while maintaining the autonomous expertise of their respective silos.

The Experiment

I used the AI as a machine in two distinct roles to perform this ontological ‘cat’s-cradling’:

AI as Legend Generator: Machines cannot feel the structural logic of a visual customer journey map. To address this, I prompted the AI to translate the visual layout into a Machine-Readable Legend - a sort of Rosetta Stone. By discretising the visual map (the nodes, the flows, the emotional peaks) into a structured text pattern, we created Digital Hypomnemata. This isn’t just a description; it’s a technical memory aid that allows the machine to hold the service logic in its context window so it can be manipulated alongside me.
AI as Reconciler: Using this legend, the AI acted as the Intercessor between two disparate data sets. I gave it the raw application data and asked it to map them against the customer journey using the Legend.

It performed a bi-directional synthesis: it identified which molecular customer touchpoints (e.g., “User enters credit card”) were dependent on molar system applications (e.g., “Legacy SOAP service latency”). It didn’t just list them; it exposed the reciprocal visibility - revealing how the tech stack enables or constrains the flow of the experience.

The Result: Sufficient Fidelity

The output was a functional, first-iteration Service Blueprint that did more than just ‘look right.’ By traditional design standards, it wasn’t ‘high-fidelity,’ and by architectural standards, it wasn’t yet a ‘system of record’. But it was just what was needed: a sufficiently familiar landscape that mapped the adjacent-possible for both sides. It had tied discrete customer interaction to the precise rows of application data and system dependencies required to support them. It was a beginning - a way for the work of collaboration and negotiation to progress without the usual “meaning-gap” that forces different disciplines into a position of defensive compromise.

Beyond mere scaffolding, the AI acted as a diagnostic layer, surfacing friction points that had been obscured. This allowed a team to move away from binary ‘can/can’t’ arguments and into a triage of imperatives, weighing customer needs against the actual material cost of architectural change. And because the output was just a HTML file, it moved from a machine-readable concept to a collaborative artifact in seconds.

Because the Why and the How were structurally coupled, one could move beyond the friction of translation and into the flow of Noetic Agency. The blueprint became a shared, material reality that allowed disparate teams to deliberate and begin to form the service, rather than just documenting a series of disconnected workarounds.

This shift is not about AI ‘solving' the problem of complexity; it is about the human-machine assemblage unlocking a new kind of situational awareness. By providing a necessary artifact for exteriorised memory, the machine handles the sheer scale of the technical stack, freeing the whole team to focus on the high-value work of negotiation and design. It turns the service into material that can be tuned - functionally and expressively - in real-time.

From Artifact to Catalyst

The value here isn’t just speed; it’s the iterative surfacing of what sociotechnologists call a Boundary Object (Star & Griesemer, 1989). As Jabe Bloom emphasises, these objects allow teams to cooperate effectively without the exhausting overhead of forced consensus or learning each other’s specialised languages. By projecting the tech stack and the customer journey onto the same plane, we create a common catalyst - a shared map that allows silos to remain experts in their own fields while providing a clear, material ground for collective action.

A Note of Caution

I was satisfied with this as a piece of sociotechnical scaffolding. It worked. It reconciled. It brings the silos into the same room. It was a good day. But as the blueprint emerged, so did a sense of dissatisfaction.

In a Delandian framing, the “molar” side of the map - the applications and hard architectural constraints - possesses a genuine capacity to affect and be affected. It has weight; you can change a line of code and the world moves.

In a coupled assemblage, the molecular side - the fluid, intensive human signals - should exist in a reciprocal relationship with that structure, acting as the catalyst for architectural change. However, this felt different. I realised that the PDF customer journey maps we reconciled are just a simulated molecular. The raw, vital signal has been processed, bleached, and categorised until it is no longer an intensive force, but a static representation of one.

Even with a high-fidelity machinic bridge, we would still be transporting a simulation, a human signal fragmented into bits, across the gap.

Beyond Reconciliation

If we move beyond just ‘making things match,’ how do we project human signals into the system so they retain their original weight? In a truly coupled assemblage, why is the customer’s voice or inferred intent, relegated to a passive data point rather than the primary force it should be - one that the technical architecture is reciprocally obliged to respond to?

I know engineering teams that are desperate for this context. They want to move away from the ‘pointless production’ of features that don’t land, toward a system where the customer signal is high-quality enough to actually architect against.

If we don’t solve for the quality of that signal, are we really forming and delivering services? Or are we just building high-fidelity conduits for low-fidelity ghosts?

The scaffolding is in place, but it’s time to talk about the state of the signals we’re sending across it. This realisation is leading to the next stage of enquiry: Transduction.

Follow the Enquiry

I’m documenting this experiment in human-machine collaboration - not as an attempt to automate design, but as an evolution in how we think and remember together.

If you’re interested in how we might use these tools as a practice of care - ensuring that our shared technical reality amplifies human agency rather than automating it away - follow me on Micro.blog or here on LinkedIn

Analysis as Autopsy

Mon, 09 Mar 2026 17:00:17 +0100

In my last post, I talked about the ontology of the tomato - how it’s a fruit to a botanist but a vegetable to a chef. In a healthy kitchen, that’s just a conversation. In a modern enterprise, however, it’s an autopsy.

We often talk about ‘Customer Insight’ as if we’ve captured a living, breathing person. But the reality is that our methods for gathering feedback are as fragmented as our departments. We don’t get the whole “animal.” Instead, we buy disconnected bits from the butcher shop: a 4-star rating click through from an email here, a line from a video of a paid testing session there, an angry tweet from last Tuesday.

We then try to “reconstitute” these dead parts into an aggregate monster that we call “The Customer Persona.”

The Enterprise Solvent

The moment any insight enters the enterprise, it hits the Enterprise Solvent. We “analyze” it (from the Greek analuein, meaning “to undo”). We slice the aggregate feedback into shards that fit our silos:

Product gets a feature request.
Engineering gets a bug report.
Marketing gets a sentiment score.

The connective tissue - the “why” that held the customer’s experience together - is dissolved. We then spend months in concurrent, disconnected meetings, moving the insight further and further away from its original self, attempting to reconstitute, recombine, and ship what-the-customer-wants.

The “Human in the Corner” Provocation

This is because the customer can’t be present throughout the process though. Right? Just imagine, for a moment, that a real customer was allowed to sit in the corner of every meeting for a month. Whose office would they sit in? Would they be allowed to attend the “Architecture Review” and the “Budget Planning”?

If they did, they’d eventually end up as the kid in The Emperor’s New Clothes. They’d look at the fractured, contradictory decisions being made and yell: “FFS I’ve already told you this ten times! If I wasn’t here to keep you honest, you lot would do yourselves some harm!”

The Reconstitution Trap

Because we can’t actually have customers sitting in every corner, all enterprise work becomes reconstitutive. We are constantly trying to stitch value back together from severed fragments. And because we have words for the “organs” (Sales, Tech, Ops), but no words for the “nervous system” that keeps them coupled to the customer’s reality, it stays disconnected. Until it makes contact with a customer.

Photo: Tim Klein

Maintaining the Life

We need to stop treating analysis like a dissection. If you have to destroy the context to understand the data, you’ve already lost the game. We need a language that describes things-in-motion - a way to discuss shared artifacts and goals without perpetual autopsy. We need a way to achieve ‘Customer Coupling’, maintaining the “connective tissue” of intent across the entire delivery lifecycle.

What’s Next?

Photo: Nathan Sawaya

To stop the autopsy, we need a different situational awareness. Because an enterprise doesn’t really have descriptors for the thing that it’s inadvertently dissected. In another post, we’ll start building that vocabulary with a look at ‘Assemblage’ - and why your company can be more like a Lego set than a marble statue. We need to understand the difference between when everything-is-fused and when things are different-but-connected. To find those words, we have to look toward a specific branch of philosophy…

Why The O-Word Isn't as Scary as it Sounds

Sun, 08 Mar 2026 12:11:00 +0100

If you grew up in the UK in the 80s, you likely remember Maureen Lipman in the classic BT advert, beaming with pride because her grandson passed his exams. “He’s got an ‘Ology!" she beamed. “People will always need an ‘Ology!"

She was right. The suffix -ology simply means ‘the study of.’ Biology is the study of life; Sociology is the study of society. Ontology (derived from the Greek onto-, meaning ‘being’ or ‘that which exists’), then, is the study of being - or more simply, the study of what a thing actually is.

The Tomato

The reason ‘Ontology’ should matter more in business is that we rarely agree on what a thing is. To explain what we mean, take the humble tomato:

To a Botanist, it is a fruit.
To a Chef, it is a vegetable.
To a Film Critic, it is a metric of failure.
To a Spaniard in late August, it is ammunition for La Tomatina festival.

The tomato hasn’t changed, but the ontology - the framework of ‘what it is’ - changes completely depending on the context. In the enterprise, we run into trouble because folks work in silos where there is the one ‘right’ way to see the tomato. We lack a language that allows the Chef and the Botanist to collaborate without one of them being ‘wrong’.

Enter “Multi-Ontology”

While computer scientists in the 90s used the term to get disparate databases to talk to one another, it was Dave Snowden who revolutionised the field by introducing multi-ontology sense-making to management theory.

Snowden’s insight was a diagnostic breakthrough: he argued that we fail when we apply the logic of an ordered system (like an engine) to a complex system (like a culture). When a leader tries to use a spreadsheet (ordered ontology) to fix a morale problem (complex ontology), they aren’t just using the wrong tool; they are misidentifying the very nature of the thing they are looking at. To use the tomato again, if the head chef barks at his sous-chef to get “a load of tomatoes in” but they arrive by the ton in a truck from Buñol, then that’s an ontological error. Because they didn’t agree on the ontology (Are we fine-dining or are we hosting a riot?), the communication and subsequent operation action failed.

You are Multi-Ontological

You already navigate these ‘different ways of being’ every day. I am a Blogger, a Father, a Customer, and a Male in his mid-fifties.

To my daughter, my ‘being’ is defined by my role as a parent.
To a health insurance company, I am a demographic data point.
To you, reading this, I am a source of strategic insight.

I am all of these things at once. I am multi-ontological.

The Enterprise “Reconstitution” Trap

The tragedy of the modern enterprise is that it forces everything into single ontologies. When a ‘Customer’ (a complex human being) interacts with a company, the Billing department sees a ‘Debtor’ and Support sees a ‘Ticket.’ Each silo performs an Analysis as Autopsy, cutting away the parts of the customer that don’t fit their specific departmental dictionary.

When we talk about ‘Transformation,’ one of the things we are really challenged with is the need for a shared language - a way to discuss the ‘thing-to-be-done’ that survives the journey from the customer’s lived experience to the productionised product service experience. We need to be able to refer to the ‘in-between thing’ as a-mixture-of-different-stuff-in-motion-without-killing-and-dissecting-it-and-ending-up-back-where-we-started. It’s a bit of a Goldilocks test, how do we talk about things precisely enough for the engineer, tangible enough for the business lead, without ending up down rabbit holes or in the weeds when there’s lots of work to be done. Fortunately, and this is the idiot-savant point of my post, French philosophy (with some precision Mexican fine-tuning) has already provided the groundwork.

What’s Next?!

Now that we’ve hopefully demystified the ‘O-word’ and realised why those single ontology silos are failing, how do we actually name the shared work?

What if Mathematics and Philosophy Drew the Same Pictures?

Sat, 13 Dec 2025 16:26:00 +0100

Think about it: two radically different fields, arriving at the same picture for completely different reasons. Two decades ago, I was deep into Deleuze and Guattari’s A Thousand Plateaus (ATP). For the uninitiated, ATP is a wild ride - philosophy on a mission, exploring complex systems, networks of power, and how everything is connected. It’s dense, difficult, and brain-bleedingly brilliant.

Then I found the bridge. A Delandian bridge, if you will.

I’m referring to Manuel DeLanda’s Assemblage Theory, as presented in his EGS lectures. Assemblage theory explains how entities - ideas, social movements, physical objects - come together to form a coherent whole and then just as easily break apart. It’s about the emergent properties or rather capacities of these assemblages and how they affect and interact. But even with my Deleuze background, it was tough going until in 2018 I encountered the work of Marc Ngui.

Ngui’s illustrations of ATP are remarkable. They aren’t just art; they are visual maps of abstract Deleuzian concepts. I’ll be honest: they are really, really hard to understand at first. One should definitely not try too hard. In one sitting. The ATP book is dense, and the illustrations follow suit. You have to let them seep in, or soak. It takes a while. Anything to do with Deleuze may sometimes be difficult, but is most certainly worth the effort.

But here is what I can tell you: in the theory and the illustrations lies a way to represent the topology of things we encounter and witness every day but lack the lexicon to articulate. Social networks, platforms, self-directed learning, pseudonymous identities, the network effect, and all the laws, Poe’s, Godwin’s, Cunningham’s and The GIFT….

In assemblage theory, whatever ‘it’ is, there’s a topology for that. And, it turns out, a diagram. It’s a deeply architectural way of seeing the world.

During this soaking period, I then come across Category Theory applied to the Glass Bead Game Now, bear with.

Category Theory is a highly abstract branch of mathematics that looks at structures to find commonalities between them. It’s the ultimate tool for finding the underlying pattern between things that look completely different. Category Theory is often called ‘the mathematics of mathematics’. (The Glass Bead Game is, for the purposes of this post, just a novel about everything - a fictional, high-stakes intellectual contest that serves as the perfect vehicle for the mathematics of complexity).

And that readers, was when I had what alcoholics refer to as, a moment of clarity.

The visual interpretations of Deleuze and Guattari’s philosophy are remarkably similar to the diagrams of Ehresmann’s Memory Evolutive Systems (MES), a framework within Category Theory used to model complex, multi-level evolutionary systems.

It was like seeing two people on opposite sides of a wall, but in a room in another part of the world, drawing the same thing.

The Delandian Warning

Now, a word of caution. Manuel DeLanda explicitly warns us about the similarity of appearance. He argues that grouping things together because they look the same on the surface is a trap - it’s the taxidermy of thought. Just because two things share a shape, (or share properties), doesn’t mean they share capacities. To DeLanda, the real value lies in the similarity of relations. It’s not about how it looks, it’s about how the parts of the assemblage come together to affect or be affected.

So, do they?

Do the parts of Category Theory and the parts of Assemblage Theory connect and interact in the same way, even though they come from such different ontologies?

Well, they are both describing the mechanics of emergence. In Assemblage Theory, the ‘parts’ are heterogeneous (people, tools, ideas) and their interaction creates a ‘whole with new capacities’. In Category Theory , the ‘parts’ are objects and their interactions are morphisms; and just for sh*ts and giggles, when they huddle together correctly, they form a Colimit - a new, higher-level object that represents the unity of the parts. But that doesn’t matter right now.

What the MES system Ehresmann describes is strikingly similar to the concept of Strata in Deleuze & Guattari’s A Thousand Plateaus. In fact the graphic convention that Ngui came up with for Strata is very similar to the conventions Ehresmann/Béjean used in their diagrams to show how complexity emerges.

Here is the first MES image that features multiple “cones” (clusters) converging into a single point.

Here is paragraph 9a, which establishes the “Strata.” Ngui draws horizontal layers that mimic the leveled structure of MES. It visually represents the idea of “sedimentation” - where lower-level elements (Level 0) are captured and organised into higher-order strata (Level n+1), exactly like the “cones” in the MES image.

And here is the development of Strata, in A Thousand Plateaus paragraph 24.

Visually, these drawings use a split or dual-panel logic to show how one level of organisation (the molecular) is translated into another (the molar). This mirrors the MES diagram showing the transition and transformation of components across different time intervals.

The invariant of these diagrams is the same Abstract Machine.

In 2019, I talked about this with Marc Ngui who excitedly agreed and remarked that ‘It was rather uncanny.’

”The MES system described in the [Glass Bead Game] article is really closely aligned to the dynamic structures D&G are describing in 10,000 BC. It is like MES is a mathematical/logical description of Strata. ..There is probably a chance that Guattari was taking ideas from Category theory to incorporate into their concept of Strata. The concept of moving from inorganic material to complex organic structures as described in the Revisited Bead Game is mirrored in D&G’s description of the progression of matter from the plane of consistency to the human organism and then to culture and language. So interesting!” Marc Ngui

A Tale of Two Approaches

In reality, both fields are observing the same hybrid reality - a Flat Ontology - for us, that means where humans, code, and hardware exist on the same plane. They simply offer two different ways of looking at it:

-1. The Assemblage Lens: Context and Capacity

Think of a sports team during a match on the pitch. You have players, the ball, the grass, and the ‘vibe.’ This is an Assemblage. It doesn’t distinguish between the biological (the players) and the technical (the pitch). It is is topological and historical. It asks: How did these specific parts come together right now? What can this combination of humans and gear actually do? We are mapping the territory of emergence. For the business, this is the “as-is” state of a transformation - the messy reality of human culture, market pressure, and existing workflows.

-2. Category Theory & MES (The Formula)

Now, look at that same match through the lens of Category Theory. It doesn’t care if the star player is wearing the No. 10 shirt; it cares about the Position (the ‘Playmaker’). It ignores the _flavour _of the components and looks at the universal functions, the morphisms connecting them. For Tech, this is the enterprise architecture - the blueprinted logic that should hold true regardless of who is performing the task or which server is running the code.

It isn’t that one theory is for humans or one visual is for machines; it’s that they are two different ways of observing the same hybrid reality. And as the diagrams and drawings reveal, fascinatingly, they coincide and corroborate each other.

-Assemblage Theory (The Map of Capacities): AT is about History and Affect. It looks at a sociotechnical system (like a DAO or a smart city) and asks: ‘What are the specific components here, and what can they actually do together?’ It recognises that a server and a human operator form a new machinic phylum It’s about the territory - the messy, shifting, specific reality of how things have actually plugged into each other.

-Category Theory (The Map of Functions): CT is about Structure and Transformation. It looks at that same system and asks: ‘What is the universal logic that governs how these parts interact?’ It doesn’t care if the object is a human or a database; it cares about the morphism, the change, between them. It’s about the rules of the game - the abstract blueprints that remain true even as the parts swap out.

The Convergence: Poised-for-the-Next-Move-

Despite their different origins, both the philosophical diagrams of Deleuze/DeLanda and Ehresmann/Béjean’s mathematical diagrams both depict a system that is stable enough to exist, but fluid enough to change. They represent a network that is poised. Architecturally, and as a student of Libeskind, I find the similarities in the diagrammatic projections of both theories and their drawings fascinating because they are a betweeness. In a Delandian sense they have - both Similarity of Relations and Similarity of Appearance.

-The Context (The Philosophical Patterns): This is offering a path for transformation. By viewing the system as an assemblage of capacities, one sees the lines of flight - the places and spaces where the culture and the tech are already shifting and ready to self-organise into a new way of working

-The Logic (The Mathematical Patterns): This is the tool for managing complexity and technical debt. By viewing your system as a category of positions and interfaces, one finds the common patterns that allow for interoperability across different eras.

By multiplexing the Teleology of Category Theory (the functional logic) with the Topology of Assemblage Theory (the historical context), we might even stop treating ‘Tech’ and ‘Business’ as separate worlds. We might start seeing them as a single Sociotechnical Assemblage poised for transformation.

And then, at that moment, I stop. Because here I start thinking about Cynefin and liminality, and Dave Snowden’s brilliant Three A’s frameworks (Agency, Assemblage, Affordance), and it’s territory that has been interrogated by folks-smarter-than-moi.

But the questions this diagrammatic comparison prompts are compelling and invite further lines of latent enquiry like:

Are there niche diagrams in Assemblage Theory that might affect Category Theory? What is the diagrammatic capacity between the theories to affect or be affected?
Can we use the Similarity of Relations in either theory to predict or anticipate when a system is about to shift from a static, legacy block to a capacitive assemblage?
Does the geometry of architectural projection and drawings - which the architect Libeskind argued were autonomous spaces of exploration - bring a new territory?

Deleuze wrote a great deal about the diagram but that’s for another post.

So, what if mathematics and philosophy did draw the same diagrams?

Well, we get patterns. But these aren’t just pretty shapes on a whiteboard or wallpaper. These patterns appear as a similarity of relations. By diagramming the teleology of mathematics (the logical where are we going?) with the topology of philosophy (the structural where are we now?), might we map a new or different kind of diagram?

The diagrams coincide here because we’ve held off looking at the things and opened ourselves to looking at the relationships. And in a world of rapid transformation, whilst we think it’s the “things” that matter - the patterns tell a different story.

Libeskind's Drawings and Revolt Against The Masterplan

Tue, 11 Jul 2023 15:20:00 +0100

Daniel Libeskind’s Micromegas (published in the late 1970s, specifically 1979) remains one of the most disruptive “anti-blueprints” ever conceived. At the time, Libeskind was the head of the Department of Architecture at Cranbrook, and he was effectively staging a revolt against the “Master Plan.”

He didn’t see these drawings as illustrations of buildings, but as “Mathematical Meditations” on the act of space-making itself.

The “Conflict of the Vector” Libeskind described Micromegas as an exploration of the “End of Space." He wasn’t interested in the “Static Object” (the building); he was interested in the Point of Collision.

What he said: He argued that traditional architectural drawing had become a “servant of the industry” - a dead language used to provide instructions for construction. Micromegas was his attempt to liberate the drawing. He described the work as a “Calculus of the Architectural” - a way to map the “invisible forces” and “vectors of intent” that a standard floor plan ignores.

The Name: Voltaire and the “Tiny-Giant” The title refers to Voltaire’s 1752 novella Micromégas (literally “Small-Large”), which features a giant from Sirius visiting Earth. Libeskind used this name to signal a Scale Collapse. He claimed that in these drawings, “Scale is an illusion." A single line could represent a city wall or a microscopic fracture in a database. There’s a transductive link. He was “leading energy across” from the astronomical to the infinitesimal. He said the drawings were about “The Verticality of the Mind” crashing into the “Horizontality of the Site."

The “Polyphonic” Drawing Libeskind explicitly linked the work to music (specifically his Chamberworks series).

What he said: He described the drawings as “Polyphonic," meaning they contained multiple, simultaneous truths that didn’t resolve into a single “Target State.”

How this Sutures into my “Applied Transduction” thinking Libeskind’s 1970s swagger helps to dismantle the modern “Target State” architecture of the enterprise. In 1979, Libeskind gave us Micromegas to show us that the ‘Master Plan’ is a tomb. He showed us that the real work happens in the ‘Small-Large’ - the infinitesimal moment where a vector of human intent hits a technical constraint. We shouldn’t be drawing a roadmap; we need to be mapping the polyphonic collisions of the Cloud monolith.

The Polymathic Matrix: Architecture as "Cartographed History"

Tue, 07 Feb 2023 16:03:00 +0100

Daniel Libeskind’s early work remains lodged in my architectural consciousness. Having written my dissertation on him years ago, I’m still finding that his “praxis” - the translation of dense, polymathic theory into physical form - has only become more relevant as technological architecture struggles to balance digital abstraction with genuine human memory.

Libeskind didn’t just build (in fact for quite a while he really didn’t!); he deciphers. His signature ideas represent a career-long effort to turn the “invisible” data of history into a “visible” tectonic language. Rather than a fixed blueprint, his work functions as a rhizome - a Deleuzian map of connections that has no beginning or end, only middle-points and offshoots.

1. Architecture as Notation: Micromegas and Chamber Works

The Exploded Geometry (Micromegas, 1979): Named after Voltaire’s satire, these “mathematical meditations” function as a Derridean trace, where the presence of the drawing is always haunted by the absence of traditional Euclidean space. They shatter the perspective of the “Enlightenment observer,” replacing it with a multi-layered, shattered geometry that anticipates his later built works.

The Musical Score (Chamber Works, 1983): These are not representations of buildings; they are explorations of how mathematical proportions can move between the ear and the eye. By using the line as a “vibrational” element, Libeskind treats the paper as a field where music is decoded into a spatial frequency.

2. The Polymathic Matrix: Mapping the Gedenkbuch The design of the Jewish Museum Berlin represents the pinnacle of his polymathic approach. He didn’t just design a building; he “mined” the city of Berlin for its lost data to create a matrix of absence.

The Gedenkbuch (Memorial Book): Libeskind utilised the names, birthdates, and deportation sites of murdered German Jews as spatial coordinates.
[Urban Vectors]: He plotted the addresses of prominent Jewish and non-Jewish Berliners (such as Paul Celan, Mies van der Rohe, and Rahel Varnhagen). By drawing lines between these points across the city, he created an “irrational” web of connections. A map of intensive processes - the pressures, flows, and historical “speeds” that shaped Berlin before they were frozen into the **“extensive” **(physical) form of the building. By drawing lines between these points, Libeskind isn’t just making a pattern; he is tracing the gradients of Berlin’s trauma.
The Collapsed Site: He then “collapsed” this city-wide map onto the museum’s footprint. The building’s famous zigzag shape is a physical distillation of these urban vectors - a 3D diagram of the broken relationships between the city’s citizens.

3. The Architecture of Absence and “The Void” Perhaps the most persistent notion in Libeskind’s lexicon is The Void. This concept moved from theoretical drawings into the physical reality of Berlin and the World Trade Center Master Plan.

The Visible vs. The Invisible: In Berlin, the “Void” is a straight, empty line of raw concrete that slices through the zigzagging building. It represents the “erasure” of Jewish life—a space that is physically present but functionally inaccessible.

“Between the Lines”: Libeskind’s philosophy posits two lines of logic: one is the “visible” path of history we walk, and the other is the “invisible” line of what has been lost. The building exists in the tension (the “interstitial space”) between these two.

4. The Garden of Exile: A Polymathic Critique The Garden of Exile (originally the E.T.A. Hoffmann Garden) is where Libeskind’s architectural praxis takes a sharp, intellectual “swipe” at historical trivialization.

The Structural Critique: By invoking E.T.A. Hoffmann - the 19th-century Prussian judge and author - Libeskind critiques the “Universal Man” of the Enlightenment. He takes aim at a culture that celebrated intellectualism while simultaneously bureaucratizing Jewish identity.
The Physical Nausea: The garden consists of 49 concrete pillars on a $12^{\circ}$ tilted foundation. Walking through it causes literal physical disorientation. It forces the visitor’s body to feel the “instability” of exile - a direct challenge to the notion that history can be neatly filed away or rendered “safe” through simple naming.

5. From Shards to the Global Studio Since founding Studio Libeskind in 1989, he has scaled these radical ideas into global landmarks like the Royal Ontario Museum (ROM) and the Danish Jewish Museum.

The “Crystal” and the “Shard”: These forms break the “box” of modernism, using sharp angles to create energy and tension.
The 17 Words: His studio maintains a vocabulary of 17 core “spiritual” words (such as Memory, Hope, Shard, Void) to ensure that even commercial projects retain his poetic and philosophical DNA.

Libeskind’s legacy is the transformation of architecture into a narrative medium - where a building is not just a structure, but a “calculated history” of absence - what is no longer there. If one were to ask Manuel DeLanda to describe the “Between the Lines” diagram, he would see it as a phase space - a map of all the possible historical and emotional states of Berlin, “actualized” into a singular, jarring concrete form. It is the architecture of material complexity where the “irrational” vector is the only honest way to map a system that has undergone a catastrophic state-change.

Architecture as an Ontological Rescue Mission

Fri, 03 Feb 2023 16:03:00 +0100

An architect who wandered into the wasteland of the modern enterprise some thirty-odd years ago and, largely through a mixture of morbid curiosity and sheer inertia, stayed to flâneur the scene.

Formative years were spent in the necessary, if daunting, pursuit of unlearning the Master Plan. Tutored in the deconstruction of Libeskind - a true polymath who understood that a building is not a box, but is both polyphonic and a confrontation. That education, supplemented by the socially contingent, everyday architecture of Wigglesworth, taught me a singular truth: the most vital part of any system is not the stultifying hierarchy etched into an org chart. It is found in the gaps, the absences, and the graphic, bloody traces of human movement between them.

A front seat in the content industry ‘boom and bust’, (Anderson and DeGrandis were having a fun time too), witnessing first-hand how the internet simultaneously created exponential value and then commoditised it into infinitesimal margins, a full decade before the rest of the world used the word ‘digital’. For two decades, I’ve observed the “Transformation” industry from a position of detached, perplexed skepticism:

The Martech Explosion: Where the uninitiated saw a bloated mess, we saw a frantic, heterogeneous, and entirely necessary response to the “connected customer” rightly identified by @chiefmartec. While Marketing was confronted with the messy reality of the human-still-fickle-but-now-online, the rest of the enterprise was occupied with the expensive hobby of hoisting dead monoliths into the “Cloud” without altering a single syllable of the internal, rotting logic.
The “Digital Transformation” Era: billions of pounds squandered on “Modernization” programs that were, in point of fact, nothing more than corporate autopsies - grotesque obsessions with a “Target State” that was dead long before the pitch deck had reached its merciful conclusion.
The Design Acquisition Wave: witnessing first hand the furious period - documented by John Maeda annually at SXSW - where behemoth firms grabbed and swallowed design houses and agencies in theory for their ‘wicked problem-solving’ intellect, but in practice cause-everyone-was-at-it. Only to command them to sit in the corner and make the interface look more user friendly, more sharper (sic), more better (more sic). A failure of imagination on a truly industrial scale.

We have come full circle. Architecture - the real centuries-old, philosophically-wrought sort - is precisely the apprenticeship the enterprise requires. Frank Lloyd Wright said: “Architecture is the scientific art of making structure express ideas.” To which the follow-up has to be: “…so have you got any?”

A friend and steely-eyed missile man engineering leader once told me, that an IT architect is judged largely on their ability to ‘know what is in production.’ If you don’t know the grain of the wood, you’ve no business at the bench. But one must aspire to more than mere inventory. In practice, this isn’t about high-theory; it’s about the coalface reality of the enterprise. Have spent years navigating labyrinthine ecosystems - tracking down siloed systems and reclaiming access for teams who had fallen into a state of learned helplessness.

When a non-technical team is paralysed by something like, I dunno, a rigid SharePoint or Sitecore implementation, the architect’s job should be to bridge that chasm. Figuring out and coaching folks on a journey to mastery, finding the automations and notifications the technical teams never volunteered because no one asked. This is the sociotechnical heuristic in action: shifting the power dynamic so the machine serves the human intent.

As Ann Pendleton-Julien observes in Design Unbound, while the engineer is desperate for the cold comfort of certainty, the architect seeks the productive, luminous territory of ambiguity.

My intention now is an ontological rescue mission, though I wish that didn’t sound so insufferably pretentious. To do so, I’m leaning on a few reliable principles to transduce the human signal with the machinery:

The Ground: advocate good sociotechnical practices, utilising Simon Wardley’s mapping to explore the terrain and the profoundly important stuff from the definitive name in anthro-complexity Dave Snowden. Separate from any formal institutional certification I refer to Cynefin and more recently his Estuarine framework. Exploring with these frameworks and tools with their inherent strategic utility in navigating complexity, especially via the lenses of the 3As (Agency, Affordance, Assemblage).
The Material: the Rhizomes and Machinic of Deleuze & Guattari - the becoming-connections that make difficult things make sense - via the brilliant, materialist and highly accessible clarity of Manuel DeLanda.
The Strategy: inspired by the deeply considered sociotechnicality of Jabe Bloom to move beyond just fixing things towards designing for good.

We don’t need a dour, three-year roadmap to Cloud Native salvation. We need to move away from performing blind analytical autopsies on the parts of an elephant on the abattoir floor. The energy is already latently present - it is in the gaps, the synapses, and the traces. We could just do with releasing it in places where it might do some good.