Why I Am Writing This

I have been thinking about analogy since 2005, longer than I have been thinking about most things professionally.

At the time, I was working as a researcher in Québec City, trying to build a system that could assist creative professionals in finding visual inspiration. The technical problem was tractable. But solving it required first answering a question I had not expected: what is creativity, actually? Not as a personality trait or a cultural value, but as a cognitive mechanism.

The answer I arrived at led to an architecture, a patent specification, and a set of ideas that I could not discuss publicly for fifteen years due to contractual obligations.

By the time those constraints lifted, the world had changed considerably. LLMs had appeared, AI had become a public conversation, and the organizations I was working with were exhibiting exactly the pathologies I had been thinking about. I finally had both the freedom and the context to say so.

This essay is where those two threads meet.

The solutions to most organizational problems already exist. They were developed in other domains, under a different name.

Analogy versus Metaphor: A Distinction That Matters

There is a common confusion between metaphor and analogy that matters more than it might appear. A decorative metaphor makes an idea accessible; it creates recognition, sometimes resonance. A structural analogy reveals a real correspondence between two systems: it diagnoses, and it points toward solutions. The difference is epistemological, not stylistic.

Saying “an organization is like an organism” is a metaphor. It invites a certain way of seeing. Demonstrating that the mechanism of cellular apoptosis and the dysfunction of project management share exactly the same causal structure (a local unit disconnected from systemic feedback, optimizing for its own survival at the cost of the whole) is a structural analogy. The first inspires; the second diagnoses.

The core formulation is this: human creativity is combinatorial, not generative ex nihilo. The only way to produce something genuinely new within a bounded experiential space is to discover a structural correspondence between two distant domains that are already known. Novelty is a function of semantic distance between domains. Utility is a function of the match with the problem and with the evaluator’s own criteria.

This formulation connects to a tradition of research in cognitive science that has approached the same question with more formal rigor. Dedre Gentner’s structure-mapping theory (1983) established that what gets mapped in an analogy is not object attributes but relational structure: the causal and functional architecture of a domain, independent of its surface content. Douglas Hofstadter and Emmanuel Sander, in Surfaces and Essences (2013), argued that analogy is not a special cognitive operation but the very engine of thought, the mechanism by which every act of categorization and understanding operates. Arthur Koestler, in The Act of Creation (1964), called the productive collision of two incompatible frames of reference “bisociation,” the mechanism behind both humor and discovery.

What this research tradition establishes, and what the diagnostic method proposed here builds on, is that structural correspondence is the operative criterion. The question is not whether “these domains feel similar” but whether “these domains share the same relational architecture.” The implication for organizational diagnosis is direct: the value of an analogy is structural rather than rhetorical, and it can be evaluated as such.

A Taxonomy of Analogical Domains

Not all distant domains are equivalent as diagnostic tools, and not all combinations of domains are equally powerful. The patent specification formalized a taxonomy for a metaphor-based search system, and that taxonomy maps directly onto the diagnostic method proposed here.

Three types of relationships between analogical domains matter:

Alternative domains cover the same structural dimension of the problem. They are interchangeable: selecting one or the other yields the same diagnostic contribution. Using both is redundant. They compete rather than combine.

Complementary domains cover distinct structural dimensions of the problem. They are combinable, and their combination produces a structural coverage that neither achieves alone. This is where the power of the method concentrates: finding two or three domains that together cover all dimensions of the problem.

Linked domains are complementary domains that also share an internal overlap with each other. That coherence strengthens the combination and makes it more than the sum of its parts.

The selection criterion for a strong set of analogical domains is therefore maximum structural coverage with maximum internal coherence, not quantity. Three well-chosen complementary domains with strong internal links produce a richer diagnostic than ten loosely related ones.

This taxonomy disciplines the method. Without it, analogical reasoning risks becoming what it seeks to transcend: the accumulation of decorative metaphors that feel illuminating without actually diagnosing anything.

The Constraint of the Judge

A structurally rigorous analogy is not automatically an adopted solution. Its value is co-constructed between the structural correspondence and the human evaluator who receives it.

The 2005 theory was precise on this point: utility is a function of the match with the problem and with the evaluator’s own criteria. For organizational diagnostics, this introduces a constraint that the method cannot dissolve but can make visible: the judges of a proposed solution are often precisely the people whose reward system is aligned with the status quo that the solution diagnoses.

A chief risk officer evaluating a proposal to extend project manager accountability beyond delivery dates is simultaneously the person whose career was built on the existing governance model. A board reviewing a recommendation to shift from process-driven to principle-driven governance is the same board that approved the processes now being questioned.

The method does not resolve this. What it does is name it with precision, which is the first condition for treating it. It also shifts the framing: the resistance is structurally predictable rather than irrational. That means it can be anticipated and designed around, and sometimes dissolved by finding the frame in which the solution resonates with the judge’s own criteria.

This is itself an analogical problem: finding the presentation of the solution that achieves maximum structural coverage of the judge’s own conceptual framework. The method recurses.

The Demonstration: Project Management and the Short Horizon

The Observation

A technology asset with a multi-year lifecycle is managed through a sequence of fixed-budget, fixed-date projects. Each project team is constituted for the duration of the project and evaluated on its ability to deliver within its commitments. Decisions are made to protect delivery dates and budget envelopes, and those decisions degrade the asset over the longer term. The team’s focus has shifted from the organization’s interest to the project’s success, which are not the same thing.

The pathology is consistent and well-recognized by practitioners. What is less well-understood is its causal structure, and therefore what kinds of interventions might actually address it rather than merely managing its symptoms.

Applying the Taxonomy: Three Complementary Domains

The search for analogical domains begins with a structural characterization of the dysfunction: a local unit optimizes for its own constraints, producing decisions whose costs are externalized to the larger system and to the future. The local unit has no mechanism to perceive or weight these externalized costs. Its reward system reinforces the optimization.

Domain 1, Biology (apoptosis): the mechanism

A cancer cell is not malfunctioning within its own frame of reference. It is optimizing perfectly for its own reproduction, which is precisely what makes it pathological for the organism. What is absent is apoptosis: the signal received from the organism that tells the cell when to stop reproducing, when to sacrifice itself for the integrity of the whole. Cancer is cellular disconnection rather than cellular rebellion: the rupture of the feedback channel through which the organism communicates systemic state to its constituent units.

The project team is in the structurally identical situation. It is not irresponsible by nature; it is disconnected from the signals that would allow it to integrate the systemic impact of its decisions. The organizational system has not built the equivalent of apoptosis, the mechanism by which a local unit receives information about what its decisions cost the larger whole.

What this domain contributes: the dimension of mechanism. The pathology is architectural, not motivational. The solution direction is not better incentives alone but the construction of a feedback channel.

Taxonomy check: this domain covers the mechanism dimension. It does not address who bears responsibility for the effects, nor why the disconnection is individually rational.

Domain 2, Law (deferred responsibility): the governance gap

Law has progressively recognized that retrospective responsibility is insufficient when effects are diffuse and temporally distant. The evolution from simple liability to fiduciary duty, from shareholder primacy to stakeholder obligation, from individual fault to systemic accountability, represents a legal tradition grappling with exactly the problem at hand: how to assign responsibility for effects that materialize outside the perimeter and after the horizon of the actor who caused them.

The project manager’s perimeter of responsibility does not correspond to the perimeter of impact of their decisions. That is a governance design problem rather than a competence problem, and law has spent two centuries developing frameworks for precisely this mismatch.

What this domain contributes: the dimension of governance. It reframes the question from “how do we motivate better behavior” to “how do we design a responsibility structure whose perimeter matches the impact perimeter of the decisions being made.”

Taxonomy check: complementary to biology. Biology covers mechanism; law covers governance. Their structural dimensions are distinct, so they combine instead of competing.

Domain 3, Neuroeconomics (reward systems): the deep cause

The human brain discounts the future at a high rate: a certain reward today is structurally worth more than an uncertain consequence in eighteen months. This is a cognitive architecture, not a character flaw. And organizational reward systems, when they encode short evaluation horizons, amplify this natural discounting rather than counteracting it.

Nassim Taleb’s skin in the game formulates the structural response: the person who makes the decision must bear a portion of the risk of its future consequences, as an alignment mechanism rather than a punitive measure. When the decision-maker’s long-term interest is structurally coupled to the long-term consequences of their decisions, the cognitive discounting of the future is partially corrected by the system design.

What this domain contributes: the dimension of deep cause. The disconnection is not merely architectural (biology) or ungoverned (law); it is also individually rational given the reward structure.

Taxonomy check: linked to biology. Both neuroeconomics and biology address adaptive mechanisms, one at the cellular level and one at the cognitive level. This internal overlap strengthens the coherence of the combination without making them redundant. Together, all three domains cover: mechanism + governance + deep cause.

The Synthetic Diagnostic

The local unit is disconnected from systemic feedback (biology). The organizational system has no mechanism to assign it responsibility for its deferred effects (law). And its reward system gives it no reason to seek reconnection (neuroeconomics). Each domain alone produces a partial diagnosis. Together, they produce a complete one.

Solution Directions: What the Source Domains Suggest

The biological analogy suggests constructing a systemic feedback channel: a technology asset health metric that follows the project manager after delivery, visible to them during the project as a forward-looking indicator of the consequences their decisions are accumulating.

The legal analogy suggests redefining the responsibility perimeter. The next team that inherits the asset should have a formal mechanism to evaluate and record the state in which they received it, and that evaluation should be part of the delivering team’s record. Accountability extends to the handoff instead of ending at the launch.

The neuroeconomic analogy suggests realigning the reward system with the asset’s horizon rather than the project’s: evaluating the project manager’s performance at 24 months post-delivery, weighted toward asset quality indicators, as incentive alignment rather than punishment.

These three solution directions are as complementary as the domains that generated them. Each addresses a dimension the others do not, and their combination amounts to a governance redesign rather than a management improvement.

What the Method Does, and What It Does Not

The method produces precise diagnostics and solution directions that the source domains have already explored. It does not produce turnkey solutions: the transposition to the specific organizational context remains a work of judgment that the method cannot replace.

What it avoids: reinventing wheels that other domains have already manufactured, and remaining imprisoned in the managerial vocabulary that masks the real structure of the problem. A problem named in managerial terms produces managerial solutions. The same problem named in biological, legal, and neuroeconomic terms produces a radically different solution space.

The method also has direct implications for agentic AI systems. An agent that reasons purely by similarity, retrieving the nearest known pattern and applying it, is executing the organizational equivalent of process-following. An agent capable of structural analogy (identifying correspondence across distant domains, composing complementary representations, evaluating coverage rather than proximity) would be executing something closer to genuine reasoning. In AI systems, the gap between similarity and analogy is the gap between a sophisticated retrieval engine and a system capable of diagnosis.

Organizations do not lack solutions. They lack the ability to recognize them across domains.

The method remains constant. The domains change. And each time, the question is the same: what is the structural correspondence, who has already solved this problem, and what does that suggest here?

If the pathologies of organizations have already been solved (in biology, in law, in physics, in software design), what are we waiting for to transpose those solutions? And what does it say about our field that we keep reinventing the same diagnoses without looking next door?

Gabriel Mailhot
Software Architect and Software Engineering Quality Advisor
Series: Structural Analogy as an Organizational Diagnostic

A Theory of Creativity First

Before building any system, I needed to understand what creativity actually is.

Here is the formulation I arrived at around 2005, working as a researcher in Québec City:

Humans cannot create something entirely new from nothing.

We can only perceive reality through our senses. You cannot imagine a color that does not exist. You cannot conceive a sound you have never heard. The entire domain available to human creativity is bounded by accumulated sensory experience.

Creative thinking is therefore combinatorial rather than generative ex nihilo. The only way to produce something genuinely new within a bounded experiential space is to discover a structural correspondence between two distant domains that are already known.

When that correspondence holds, it is a metaphor.

But novelty alone is insufficient. A creative idea must also be useful, and usefulness is not intrinsic to the idea itself. It is evaluated by a human observer with a specific problem and a specific set of objectives. The same association can be brilliant in one context and irrelevant in another.

This leads to a precise formulation:

A creative idea is an association between two distant known domains whose relevance is judged by a human evaluator as both novel and useful within a specific context.

Novelty is a function of semantic distance between domains.

Utility is a function of the match with the problem and with the evaluator’s own criteria.

From this theory, an architecture follows naturally.

The Architecture (2005–2010)

If creativity is the identification of structural correspondences across distant domains, then a system designed to assist creativity should not search for what is similar. It should search for what is structurally equivalent across distance.

Between 2005 and 2010, working at a private imaging company in Québec City, I designed and implemented such an architecture.

Each object in the system (initially images, but the framework was modality-independent) was described through three structured vectors:

• A composition vector capturing spatial organization, perspective, and luminosity
• An object-property vector encoding detected geometries, shapes, and physical attributes
• A semantic vector containing qualitative keyword descriptors

These vectors were computed automatically using classical computer vision techniques: color histograms, wavelet decompositions, geometric segmentation, texture analysis, spatial frequency distributions. Each image received a multidimensional signature, which we called a “genetic code,” stored in a database and queryable in real time across a collection of 100,000 images.

In today’s terminology: a hand-crafted feature-embedding pipeline feeding a structured vector database, with sub-second query response time. Built in 2008, before those words existed in this context.

But the architectural intent was quite different from similarity search.

The system did not ask: What is nearest?

It asked:

• Which objects share equivalent structural overlaps with a reference? (alternative objects: substitutable, not combinable)
• Which objects contribute complementary, non-overlapping components of a reference? (complementary objects: combinable to form a stronger metaphor)
• Which combinations increase analogical coherence rather than surface resemblance? (linked objects: internal reinforcement as a tiebreaker)

The metaphorical strength of a combination was computed as the aggregate overlap across structured dimensions. Combinations were ranked by analogical coverage, not by proximity.

A red apple and a white tablecloth, taken separately, are only partial metaphors for the Japanese flag. Combined, they cover it completely: one contributes the red circle, the other the white rectangle. Their combination scores higher not because they resemble the flag, but because together they account for its structure. That is analogy, not similarity.

Similar questions were being explored across content-based retrieval and computational creativity research during that period. The problem was recognized; what was missing were the representational tools to address it at scale.

Why This Distinction Still Matters

Modern embedding systems, including the best multimodal foundation models, are extraordinarily capable. They organize high-dimensional spaces where related concepts cluster together. Distance minimization in embedding space has proven to be a remarkably powerful paradigm.

But the analogical operation itself remains largely latent in these systems.

Current architectures optimize for distance, retrieve nearest neighbors, and interpolate within learned manifolds. They do not explicitly evaluate which pairs or sets of representations together form a structurally stronger mapping to a reference than any single element alone.

Foundation models can produce metaphor-like outputs through statistical association, sometimes impressively so. But the mechanism is emergent rather than structured, which means it cannot be controlled, composed, or inspected.

Making analogy a first-class, explicit, steerable operation in AI systems remains an open architectural problem.

What AI Still Needs

Revisiting this research today, three limitations of current AI systems stand out, each rooted in the same theory of creativity I was working from in 2005.

1. Implicit context inference

Human creativity does not operate in a vacuum. The relevance of any idea depends heavily on the context in which the human is operating: their constraints, objectives, mental state, environment. A skilled collaborator reads this context without being explicitly told.

Current AI systems depend almost entirely on what the user explicitly provides. They do not perceive the context in which the human evolves. For AI to become a genuine creative partner, it needs access to richer contextual signals, along with the capacity to infer meaning from them without requiring the human to articulate everything. That is a problem of situated cognition rather than retrieval.

2. Memory as a knowledge reservoir

In my theory of creativity, the quality of associative reasoning is directly proportional to the richness of available memory. More diverse experience means more potential connections across distant domains.

Current AI systems have impressive parametric knowledge, but their memory architecture is largely static between training runs. The strategic management of episodic, contextual, and long-term memory, treated as a dynamic, queryable reservoir that evolves with use, remains underdeveloped. Memory, in this view, is the substrate of association and not just storage.

3. Emotional memory

This is perhaps the most significant gap, and the one most likely to trigger an objection from researchers in reinforcement learning or affective computing.

The objection is predictable: AI systems do incorporate affective signals, through reward modeling, reinforcement learning from human feedback, and preference learning. This is true, and the work is serious.

But it addresses a different problem. These approaches treat affect as an external training signal: feedback injected during learning to shape behavior. Human emotional memory is structurally different. Rather than a signal applied to knowledge from the outside, it is woven into knowledge itself.

The neuroscientist Antonio Damasio demonstrated this precisely. Emotion is not separate from rational cognition; it is constitutive of it. Patients with damage to emotional memory centers do not become more rational. They lose the capacity to make decisions at all, because they can no longer assign differential weight to recalled experiences.

Every human memory carries affective encoding (fear, joy, discomfort, wonder) that influences which associations surface, which combinations feel meaningful, and which ideas resonate. This weighting is embedded in the retrieval structure itself rather than applied post-hoc.

Current AI systems process semantic relationships without this integrated affective architecture. The gap may therefore be architectural rather than merely incremental.

Building systems where affective structure is embedded in memory, instead of applied to it from outside, may be one of the most important open problems in AI research on creativity and reasoning.

Why I Am Writing This Now

The research described here was conducted between 2005 and 2010 and was not publicly discussed at the time due to contractual obligations.

Revisiting it now is less an exercise in retrospection than a recognition that the underlying questions remain open, and that the technological conditions to investigate them meaningfully finally exist.

I am not writing to claim priority. Parallel discovery is the norm in research, and I have no interest in relitigating timing. I am writing because the problem is still in front of us.

If you are working on reasoning architectures, compositional intelligence, contextual AI systems, or human–AI collaboration, I would like to talk.

The research described above is documented in an unpublished patent specification and a research poster produced in 2008.

Gabriel Mailhot
Software Architect and Software Engineering Quality Advisor
Series: Organizational Engineering for the AI Era