# Scratch Pad *Quick notes, surfaced content, things to have on hand today* ## Nat's draft **Codes, Grammars, and What Happens When Frameworks Touch** In 2012, Davide Nicolini, Jeanne Mengis and Jacky Swan published a study of a cross-disciplinary team building a bioreactor — a device for growing stem cells under controlled conditions. The team comprised bioengineers, sensor specialists, electronics engineers, and molecular biologists, each working from their own lab and meeting monthly to coordinate. During one of these meetings, a junior bioengineer put up a slide showing results from a recent experiment. One of the sensor readings showed what she called "a weird trend" — an unexpected drop in ammonia concentration that nobody could immediately explain. What followed was striking. The senior electronics specialist asked whether the process accounted for calibration. The sensor specialist asked whether the raw voltage had been smoothed before conversion. The bioengineer asked everyone to go back a slide: "This noise, are we sure this is noise?" People started standing up, approaching the screen, pointing at specific parts of the curve, sitting back down. The electronics specialist brought up a story about gravitational wave detectors in the United States that had produced false readings because a power plant dropped off-line — suggesting the anomaly might be environmental electromagnetic interference. Someone asked whether the day of the week had been recorded, since power loads shift over weekends. The junior bioengineer took note. Nicolini and colleagues describe the interaction as resembling "a nicely choreographed dance." What is happening here? Different specialists are reading the same graph, but each is entering it from a different place, with a different set of concepts, different assumptions about what might explain the pattern. What situation was that? What was that room producing? **The organising problem** Organisations exist because things need doing that no individual can accomplish alone. Solving complex problems requires the coordinated commitments of multiple actors who control different resources - material and *cognitive*. The bioengineers in the bioreactor project thought in terms of cell behaviour and culture conditions. The electronics engineers thought in nanoseconds and signal integrity. The sensor specialists thought in terms of polymer chemistry and calibration drift. Different expertise, different training, different assumptions about what matters and why. As a result, diversity has always been considered a resource for organisations, particularly for its benefits in fostering adaptability and innovativeness. But this diversity is also a problem. Actors who see things differently need to build enough mutual understanding to act together. They need to communicate what they know, negotiate what matters, make sense of shared situations, align on what to do next. Communication is ubiquitous in organisations. In fact, there is a tradition in organisation theory that makes the point more forcefully. The argument is that communication is not simply something that happens *inside* organisations: it is the process through which organisations are established, composed, and sustained. Organisations are networks of commitments — requests that create obligations, promises that bind future action, declarations that bring new facts into existence — held together by ongoing conversation. As Dewey put it, society "not only continues to exist *by* communication, but may fairly be said to exist *in* communication."  What this means is that communication is not one factor among many in organisational life — alongside, say, incentive structures, resource allocation, or formal hierarchy. It is the process through which all of those things come into being and are reproduced. Incentive structures exist because someone articulated them and others interpreted them. Hierarchies hold because they are enacted through speech acts — directives, reports, requests — not because an org chart on a wall exerts gravitational force. When communication breaks down, these structures do not simply become harder to use. They cease to function, because they *are* communication. If communication is constitutive of organisation — if organisations exist *in* communication rather than merely using it — then the properties you attribute to communication are not incidental to your theory. They determine what kinds of organisational phenomena your theory can see. A framework that treats communication as the *transfer* of pre-formed information between minds will see organisations as coordination problems: the challenge is getting the right information to the right people. A framework that treats communication as a *meeting* of minds (the process through which meaning, structure, and collective action are brought into being) will see something different: organisations as sites where new possibilities can be produced, not just existing ones transmitted. In this post, I focus specifically on one form of communication: language. Communication, of course, is broader than language — it includes gesture, material artefacts, spatial arrangements, visual representations, and much else besides, all of which participate in constituting organisations.[^1] But language has a property that the others do not, at least not in the same way: *compositionality*. Language combines finite elements according to rules to produce expressions that have never existed before. A gesture communicates, but it doesn't generate novel expressions by combining discrete elements according to productive rules. A building communicates values, but it doesn't have a grammar. Broadly speaking, management theory has followed three philosophical traditions on the role of language in organisations. The first treats language as a *code* — a system of representation. The second treats language as *game* — something that does things in context. The third treats language as *grammar* - a generative system, something that produces things that didn't exist before. As we will see, each has very different implications for what you think can happen when people who see differently try to work together. **The code view** The first tradition descends from the work by economist and Nobel Prize laureate Kenneth Arrow. In *The Limits of Organization* (1974), Arrow argued that organisations develop specialised languages — *codes* — that allow members to communicate efficiently. A code is a classification scheme — a set of categories that allows members of an organisation to compress information, communicate efficiently, and act on what they hear. A company develops internal shorthand for its products, its processes, its problems. A profession develops technical vocabulary that makes communication among practitioners fast and precise. These are codes.  You can look at codes like investments. Once developed, they become entrenched — filtering what members perceive, what counts as relevant, which problems make it onto the agenda. They are also path-dependent: the conditions under which a code was created leave traces in its structure long after those conditions have changed. The code filters what members notice, what counts as relevant information, which problems are legible and which are invisible. This is why codes are efficient: they are selective. And this is why they are costly to change: the circumstances under which the code was developed leave traces in its structure long after those circumstances have passed. A code maps things in the world to labels, causes to consequences, actions to evaluations. When people share the same code, coordination works. When they do not, it breaks down. This connects directly to the organising problem. As I mentioned, organisations are diverse by design — different functions, different expertise, different ways of seeing. Each group develops its own code: engineers talk about tolerances and specifications, marketers talk about positioning and segments, finance talks about margins and exposure. These codes are what make each group internally effective. They are also what makes coordination across groups difficult. The same word — "quality," "risk," "value" — sits in different codes and does different work in each. The code view's answer to this problem comes in several forms. The most straightforward is convergence: build a shared interpretation of code through socialisation, training, co-location, repeated interaction. This works but it is expensive, slow, and sacrifices the diversity that made the different groups useful in the first place. If everyone shares the same code, you have solved the coordination problem by eliminating the diversity that created it. A more interesting answer comes from the literature on strategic ambiguity. Eisenberg, in a paper that has become foundational, argued that ambiguity in organisational communication is not always a failure of clarity. It can be a deliberate and effective strategy. When a leader articulates a vision in terms that are open to multiple readings, different groups can each find something in it that maps onto their own concerns. The vision holds the coalition together not because everyone agrees on what it means, but because everyone can project their own interpretation onto it. This perspective highlights an important property that I will discuss more in detail in the next section. Codes are *fuzzy*. Communication codes map stimuli to labels — but not crisply. The same word can activate several meanings; the same thing can attract several labels. Causal codes map actions to consequences, again with probabilistic weights. Moral codes map actions or outcomes to dimensions of desirability. Notice that this fuzziness is not a defect. It is what makes strategic communication possible. If a leader knows that "sustainability" maps to environmental stewardship in one person's code and to durability of competitive advantage in another's, she can deploy the term knowing it will do different work for each audience. She can switch codes when addressing different groups, or show someone a path from an action to a value they already hold but had not connected to that action — moral reframing. This is powerful. It explains a great deal about how organisations coordinate across internal diversity, how skilled actors navigate heterogeneous audiences, how culture functions as both shared understanding and strategic resource. But the underlying ontology remains representational: language maps things onto other things. What changes through interaction is the pattern of mappings — which associations strengthen, which weaken, which get activated. The concept sets themselves are given before the interaction begins. Strategic ambiguity works because different codes already exist, and the ambiguous message sits at their intersection. The communicator exploits the fact that a term has different mappings in different codes. Each actor reads it through their own. Nobody needs to change their code. Nobody needs to understand anyone else's. The diversity is navigated, not engaged with. When agents interact, their mappings can converge (through socialisation or learning), remain different but navigable (managed divergence), or be strategically exploited (ambiguity, code-switching, reframing). In all these cases the underlying picture is the same. There are codes. They map concepts onto labels. Each person reads the message through their own code. Coordination depends on the relationship between codes: whether they overlap, whether they can be bridged, whether their differences can be strategically exploited. The concepts themselves — what populates the codes — are given prior to the interaction. What changes is how the mappings are managed. The framework asks whether codes are similar enough for coordination or different enough for strategic exploitation. It does not ask what happens when different frameworks come into direct contact — when people with genuinely different conceptual architectures interact through conversation, building on and reacting to each other's contributions. This means the code view handles transmission, alignment, and strategic deployment. It does not handle emergence — the possibility that interaction itself produces something that no participant's code contained beforehand. **Language as action** The second tradition breaks with the first on a fundamental point. Language does not just represent things — it *does things*. Language, in this view, is performative: it constitutes social realities rather than merely referring to them. This insight has its roots in Wittgenstein's later philosophy, where meaning is not a fixed property of words but something that emerges through use. You do not understand a word by looking up its definition. You understand it by learning the "language game" in which it operates — the rules, the context, the forms of life in which it is embedded. The same word does different work in different games. "Open" means one thing in a restaurant and another in a chess match.  This is an important distinction from the code view. In the code view, a leader navigating diversity deploys ambiguous codes that map onto multiple representation and optimise their choice based on the optimal level of information: fuzzy enough to be broad and plausibly interpretable. Organisational communication is a matter of getting the right amount of information to the right people and manage (or indeed exploit) the potential mismatch to the organisation’s advantage. The pragmatic view suggests something different. Different groups are not just using different labels. They are playing different language games and context matters for meaning. When a manager says "we are restructuring for long-term value," the phrase does not work because it is vague or fuzzy enough for each group to project their own meaning onto it. It works because the utterance performs a different act in each game. To investors it functions as a commitment to returns. To employees it functions as a reassurance that change has a purpose. To regulators it functions as a signal of responsible governance. The words are identical. What they do — what obligations they establish, what responses they make appropriate — differs depending on the game in which they land. Communication succeeds when the right act is performed in the right game. Which means that in a diverse organisation, where different actors are playing different games, the same utterance can simultaneously succeed and fail — landing as a commitment for one audience and as mere noise for another. Diversity is not a problem of incompatible representations but a condition in which different actors are engaged in fundamentally different kinds of activity. But notice what this tradition still shares with the code view: it takes existing frameworks as given. The pragmatic view opens up what language can *do*. It does not yet ask what language can *produce*. It does not theorise what happens when different language games come into contact and the collision *generates something new*. It does not explain generativity. And it is generativity — the production of possibilities that no existing game contained — that the third tradition is about. **The other side of language** Remember the ammonia graph? What happened in that room? Each specialist read the ammonia graph through their own framework — calibration, voltage, noise. The two traditions we've traced each account for something here: one explains the divergent readings, the other explains what each specialist was doing with their reading — probing, framing, testing against what they knew. Between them, they cover a lot. But something else took place in that room. The electronics specialist, trying to make sense of the anomalous reading, brought up a case from physics: gravitational wave detectors that had produced false positives because a power plant dropped off-line, flooding the instruments with electromagnetic interference. The analogy was not obvious — ammonia sensors in a bioreactor and gravitational wave detectors have nothing to do with each other. But the story suggested a mechanism: that background electromagnetic noise, which varies with power load, might explain a sensor reading that didn't otherwise make sense. Someone asked whether the day of the week had been recorded. Power loads shift over weekends. Neither the code nor speech acts view accounts for is where the power-line hypothesis came from. It was not in anyone's framework when they walked in. The connection between weekend electromagnetic interference and an anomalous ammonia reading was not stored anywhere. That hypothesis — environmental electromagnetic interference as the source of the anomaly — wasn't in anyone's framework when the slide went up. It came from the friction between the electronics specialist's experience in a completely different domain, the sensor specialist's knowledge of voltage sensitivity, and the bioengineer's suspicion that what looked like noise might be signal. It emerged from the contact between frameworks, through an object none of them fully understood. A new possibility, one that wasn't in the room when the slide went up. There is a third way of thinking about language and organised activity that starts to make this visible. Warglien calls it treating organisations not as contexts in which language operates but as language themselves — generative systems whose defining property is not representation or performance but combinatorial productivity. A grammar, unlike a lexicon, does not store expressions. It produces them. Finite elements, finite rules, potentially infinite output — including output that has never existed before. He and Reinhard Selten tested what this looks like. They asked pairs of subjects to develop artificial languages for geometric figures. Given recurring figures, subjects built idiomatic codes — fixed pairings of label and referent, stable and efficient. Given figures they had never seen and would never see again, something shifted. The fixed pairings ran out. What subjects built instead were compositional systems: ways of combining elements to express objects encountered for the first time, rather than ways of retrieving the right label from memory. The grammar appeared where the code was insufficient. Generativity is needed, it seems, when the existing inventory runs out. In her work on groups that produce breakthrough ideas, Harvey argues that what distinguishes their output is not the variance in what individual members bring but what happens when those different perspectives are brought into genuine contact. The product of that contact is not a recombination of existing elements. It is, in her account, qualitatively different from what any participant contributed — a new framework that emerges from the friction between existing ones, containing them but reducible to neither. The source of novelty is the integration process itself.  The bioreactor room worked something like this. The electronics specialist's gravitational detector story, the sensor specialist's knowledge of voltage sensitivity, the bioengineer's suspicion about noise — these were not pooled. They were brought into contact through an object none of them fully possessed, and the contact generated a hypothesis that no single framework contained. **What this opens** The distinction between codes and grammars changes what you think is possible — and, as a consequence, what you think is worth asking. If your theory of language in organisations is code-based, the questions it generates are about alignment and management: how do we get codes close enough for coordination? How do we exploit gaps between them strategically? If your theory is pragmatic — language as action, as performance, as the doing of things in context — the questions shift toward how utterances land differently in different games, how the same words can simultaneously commit one audience and reassure another. These are real questions with real answers, and they account for a great deal of what actually happens in organisations. But both traditions share something: they take the existing frameworks as given. The code view asks how mappings relate to each other. The pragmatic view asks what acts are performed within which games. Neither quite asks what happens when frameworks meet and the meeting itself produces something — a concept, a hypothesis, a possibility — that none of the participants brought in. That is the question the grammar view opens up. Under what conditions does interpretive diversity become a resource for producing something new, rather than just a problem to navigate? The junior bioengineer took note. Nobody in that room had arrived with what she wrote down. Nobody left carrying only what they walked in with. *This connects to work I am developing on how language functions in contexts of interpretive plurality — how the same expressions enable diverse audiences to construct rich but distinct elaborations, and what happens when those elaborations come into contact. More on that, hopefully, in future posts.* **References** Arrow, K. J. (1974). *The Limits of Organization*. Norton. Warglien, M. (2013). Language and economic organization. In *Handbook of Economic Organization* (pp. 137–154). Edward Elgar. Koçak, Ö., & Puranam, P. (2024). Decoding culture: Tools for behavioral strategists. *Strategy Science*, 9(1), 18–37. Selten, R., & Warglien, M. (2007). The emergence of simple languages in an experimental coordination game. *Proceedings of the National Academy of Sciences*, 104, 7361–7366. Harvey, S. (2014). Creative synthesis: Exploring the process of extraordinary group creativity. *Academy of Management Review*, 39(3), 324–343. Star, S. L., & Griesemer, J. R. (1989). Institutional ecology, translations and boundary objects. *Social Studies of Science*, 19(3), 387–420. Nicolini, D., Mengis, J., & Swan, J. (2012). Understanding the role of objects in cross-disciplinary collaboration. *Organization Science*, 23(3), 612–629. Pentland, B. T. (1995). Grammatical models of organizational processes. *Organization Science*, 6, 541–556. Knorr Cetina, K. (1997). Sociality with objects. *Theory, Culture & Society*, 14(4), 1–30. Page, S. E. (2011). *Diversity and Complexity*. Princeton University Press. http://www.jstor.org/stable/j.ctt7pfdp Cooren, F., Kuhn, T., Cornelissen, J. P., & Clark, T. (2011). Communication, organizing and organization: An overview and introduction to the special issue. *Organization Studies*, 32(9), 1149–1170. Warglien, M., & Gärdenfors, P. (2013). Semantics, conceptual spaces, and the meeting of minds. *Synthese*, *190*(12), 2165-2193. Austin, J. L. (1962). *How to Do Things with Words*. Oxford University Press. Searle, J. R. (1995). *The Construction of Social Reality*. Free Press. Wittgenstein, Ludwig (1953). Philosophical Investigations. New York, NY, USA: Wiley-Blackwell. Edited by G. E. M. Anscombe. Winograd T. and F. Flores (1986), Understanding Computers and Cognition: A New Foundation for Design, Norwood, NJ: Ablex Publ Corp. Robichaud, D., H. Giroux and J.R. Taylor (2004), ‘The metaconversation: the recursive properties of language as a key to organizing’, Academy of Management Review, 29(4), 617–634. Leblebici, H. (2000), ‘Allocation of rights and the organization of transactions: elements of a generative approach to organizing’, Journal of Management and Governance, 4, 149–168. Astley, W.G. and R.F. Zammuto (1992), ‘Organization science, managers, and languages’, Organization Science, 3, 443–460. [^1]: The CCO tradition is explicitly ecumenical on this point. Cooren et al. (2011) argue that values, knowledge, and ideologies can be "conveyed, incarnated and constituted not only through what people say and write, but also through what they wear, how they look, and how they gesture or behave" — and through nonhuman agents such as documents, architectural elements, and technologies. --- # What's Happening *Active threads and today's activity — updated by /harvest* ## Active ## Sessions ## Actions ---