The Amazing, Shrinking Org Chart

About a year ago, an 1855 org chart of the New York and Erie railroad was cascaded worldwide by the VP of the Infographics Department of the Internet. There was a good deal of admiration as well as lamentation. Apparently we no longer care enough about our corporations to create beautiful depictions of their anatomy, ars gratia artis. Whatever else the shortcomings of mid-nineteenth century corporate management (they had a tendency to start wars and gun down workers in pursuit of their Missions and Visions among other things, and you had to be a quick-draw gunfighter to earn a Harvard MBA in those days), they clearly cared.  [caption id="attachment_4859" align="aligncenter" width="200"] Library of Congress (via McKinsey)[/caption] By contrast, a modern set of org charts is usually a showcase of apathetic PowerPoint banality. In fact, you rarely ever see a big global view anymore. Just little local views that could, in principle, be patched together into a global view, but in practice never are. Often, even CEOs only have a coarse, low-resolution view of the whole, with blocks representing entire huge divisions of thousands of humans and billions in capital assets. There is usually no operational capability for drilling down into finer points where the situation demands it (Proctor and Gamble, apparently, is an exception). Most senior executives -- VP and above in organizations of 1500 or more people say -- are in the position of surgeons operating on the basis of having played the kids' game Operation rather than on the basis of medical training and tools like MRI machines. There's a very good excuse for this though: the pace of organizational and environmental change today turns static maps into garbage very quickly. The part of the organization that is both possible and useful to represent using an org chart has been rapidly shrinking. What, if anything, should be done about it? The org chart has been the main orientation tool for corporate life for at least 160 years, going by the map I opened with. Probably longer. Certainly too long. But we rarely stop to consider what exactly an org-chart is, conceptually. The org chart is a visualization of a theatrical web of masks and power. We acknowledge the element of theater by calling each box a role. In your classical corporation, each box flows via a solid upward arrow into a unique superior box. The whole is a natural human social network pulled upward into a hierarchy by the bootstraps of the founder. Unlike a network, a hierarchy has a notion of up, defined and enforced by the converging flow of solid arrows towards an apex of power. A strict hierarchy also admits a notion of inside/outside, defined and enforced by means of a set of deliberately broken or weakened natural information flow links: a boundary. The process makes the underlying social reality simpler and more legible of course, but not entirely, or even primarily, in service of function. A good deal of the process is about imposing anxiety-alleviating platonic structural beauty. So to summarize, a hierarchy as represented in an org chart transforms its raw social-graph input in three ways:

1. It embodies a function that results in orientedness ("up") and boundedness ("inside")
2. It is more legible; you can read more meaning into it
3. It is more aesthetically pleasing when viewed from the top or outside

This theoretical structure of course, never entirely true in the first place, began to crumble in the 1970s. First came the bane of the cubicle-drone's life, the dotted-line reporting relationship. Then came the half-hearted stab at agility known as the matrix organization, with rows and columns. Then came outsourcing, keiretsu, core competencies, process re-engineerings, value-chain partners, unbundlings and rebundlings, and a steadily growing contingent labor force: the stagehands of the corporate theater. Up became increasingly a meaningless fiction. The CEO became increasingly a figure of heroic myth-making rather than the apex of the lived social reality of the corporation. The corporation's chief storyteller in the stock market (or, in more regulated and corporatized parts of the world, chief lobbyist in the corridors of power) rather than its leader. Inside/outside too, became an increasingly meaningless fiction, as natural connections reasserted themselves across nominally controlled boundaries. After a century of subservience, the social graph had reasserted its authority over the impoverished org chart. Up become the direction of increasing disorientedness. Inside became the direction of increasing disconnectedness from reality. In its terminal, mortally stricken form, this disoriented, disconnected entity turned into the zombie corporation -- living-dead corporations like Radio Shack -- that had failed a critical transformation challenge. The evidence of vast numbers of zombie corporations (and other organizations) around us suggests, prima facie, that the amazing, shrinking org chart is the symptom of a pervasive problem: that of adapting to a more dynamic business environment. Keeping Up versus Letting Go There are two basic (but not mutually exclusive) responses you can have to this phenomenon of shrinking org-chart views of corporate realities: turn to more powerful maps or step back and reconsider what sort of situation awareness you actually need in order to be effective. The two approaches are what I call the keeping up and letting go approaches. Keeping up means you improve your real-time data-gathering and dynamic mapping capabilities. It means trying to get ahead of change and staying ahead. It means signing up for an arms race with the environment. The faster things change, the more horsepower you bring to bear on the problem of keeping your global situation awareness current and accurate. And relevant individual views of ground truths globally aligned and synchronized. Letting go means you start figuring out management strategies and capabilities that do not require global situation awareness or tight synchronization and alignment of views, just a rough consensus on a few key beliefs. This means developing capabilities that are, in some sense, antifragile in the face of increasing VUCA (volatility, uncertainty, complexity, ambiguity). In the keeping-up department, one of the best attempts I've encountered is Simon Wardley's mapping technique, which traces the structural evolution of value chains (in roughly the sense of Michael Porter or early nineties business process re-engineering) over time. Wardley maps give up some structural completeness (since it's hard to represent a business with multiple value chains on a single 2d diagram) in order to represent corporate realities along a more useful dimension: Schumpeterian evolutionary time. A Wardley map is a map of creative-destruction in progress, in the form of a freeze-frame snapshot of an evolutionary cycle. To use Wardley maps, you need sufficiently detailed information about the internal and external environment. In the letting-go department, the best attempt I've seen is my favorite new workflow product, Slack (that's truly inspired naming incidentally). It is  not just the first credible threat to email. It is the medium-that-is-the-message of emerging patterns of corporate organization. I've been using it for multiple projects for a few months now, and it definitely requires a major, low-level set of behavioral changes in how you think about work. Not the least of the challenges is giving up the assumption that chatting/IM is some sort of childish and playful distraction activity rather than the mainstay of serious work (there are still people out there who don't think it's work unless it is in the form of a printable PDF that at least a hundred people do not read, preferably after printing). But once you make the mental shift, for many kinds of work, your effectiveness is vastly amplified. You won't be able to draw better org charts, but you will have less need for them. Depending on how you mix and match the two approaches in different areas of activity, you can generate behaviors ranging from Boydian maneuver warfare and guerrilla disruption attacks on industries to tightly-synchronized network centric warfare type approaches and Skynet-level gothic high-tech command and control systems. How do the relatively pure approaches fare when they go head-to-head? I did a little paper-napkin level inventory of cases I am familiar with (which I won't share since it's partly based on proprietary information) and concluded that there is no clear winner. Sometimes letting-go approaches beat keeping-up approaches, and sometimes it goes the other way. The outcome depends strongly on the state of the entire stack within which the competition is situated, and crucially, the authoritah of the leader. In brief, keeping-up approaches work better where you have a sufficiently high-authoritah leader, and letting-go approaches work better when the leader is not the high-authoritah type. I have a suspicion that letting-go approaches are gaining a small but decisive advantage that will compound to dominance in about fifty years (especially if many enabling technologies, Slack-like in spirit, continue to enter the workplace), but that's admittedly speculation. Beyond the Org Chart Whether you are keeping up or letting go, or some mix of the two, what you're doing is getting beyond org-chart mindsets. On Twitter recently, Simon was saying (heh!) rather rude things about my most sacred object, the 2x2 diagram, so I promised him I'd get my revenge by putting Wardley Maps in the bottom-left corner of one. As we all know, the bottom-left quadrant (or Q3) is the 2x2 equivalent of hell. Here is my revenge 2x2: a visualization of beyond-the-org-chart adaptation strategies: To complete the heaven/hell metaphor, the top right is obviously heaven, where righteous people like me live, and the off-diagonal quadrants are two kinds of purgatory. Kidding aside, this is a serious 2x2, and there's no particular reason to orient it this way. Among them, the four responses cover 90% of response patterns I've seen across organizations (not counting zombie corporations that have gone catatonic through repeated failure to transform). The x-axis ranges from fundamentally determinate approaches (those that aspire to a complete and canonical understanding of reality, in the hedgehog sense) to fundamentally indeterminate approaches (those that attempt to work with a plurality of partial models without privileging any particular one, in the fox sense). My use of the terms determinate/indeterminate is roughly the same as in Thiel's famous 2x2 of optimism/pessimism, determinate/indeterminate (the same heaven/purgatory/hell joke works there: Thiel is in the purgatory of determinate optimism, suspicious of the heaven of indeterminate optimism). The y-axis may be non-intuitive to some of you. I am borrowing the terminology from adaptive control theory. In adaptive control (the engineering discipline behind things like adaptive cruise control in cars), there are two basic ways to engineer adaptation when you anticipate constantly changing operating conditions.

• In direct adaptive control, you try to come up with control architectures that feed directly off (noisy, messy) feedback and environmental signals to detect changed conditions and reconfigure the control mechanism.
• In indirect adaptive control, you come up with an explicit model of the system, and use feedback and environmental signals to "identify" the parameters of the model (using model-fitting techniques). You then reconfigure the controller by pretending that the model is reality.

Both architectures are present in nature (in the human brain for instance). Both involve an outer adaptation loop that "supervises" (or "reorients" in a Boydian-OODA sense) an inner feedback loop, tuned for locally static conditions. Both can fail when the adaptation/reconfiguration ability is pushed beyond the designed range of variation, requiring "open loop" creative  thinking for recovery. Both can also fail due to insufficient speed of adaptation, causing the system to crash during reconfiguration (related to the problem of "fast transients" in Boydian theory, also known as the "dual control problem" in control theory where there is a tradeoff between adaptation in the outer loop and stabilization in the inner loop. An example is an automative adaptive cruise control mechanism that does not retune the braking/acceleration algorithm for wet conditions quickly enough, and rams into the car in front as a result). A classic org chart represents a non-adaptive control scheme: An observe-decide-act loop incapable of reorientation (or equivalently, an OODA loop with unusably slow transients). The advantage of direct approaches is that they are often more robust and often involve fewer modeling assumptions that can break. At a people level, they involving breaking and forming habits via direct conditioning. Their disadvantage is that they are non-intuitive and harder to fix when the assumptions do break down. They don't give you much of an understanding of the system. Only the means to control it effectively. If you want understanding and appreciation, that's a separate problem. The advantage of indirect approaches is that they give you an appreciation of how the system works (the model) as a byproduct of trying to control it, but give up some of the capability envelope in order to do so. Sometimes a very significant amount of capability. At a people level, indirect approaches adapt via cognitive reframing and conscious, self-managed attempts at behavior change. Each of the four quadrants represents a pure approach that works under some narrow range of circumstances. Indirect-determinate techniques like Wardley Maps apply when you there is low ambiguity about the structure of the value chain (there is a discoverable "ground truth" about the environment), and you don't have complete control over all the moving parts in the environment. If you do have control (or can acquire control), you can operate in direct-determinate (Q2) ways: lower VUCA by simply buying up and integrating everything you need, soup to nuts, making the competition less relevant, and using more traditional management methods (the analogy in control is adding more actuators to control previously uncontrolled behaviors, lowering the need for adaptation). The reason this works was pointed out by Chris Dixon in a tweet: "In tech, the most intense competition is between complements not substitutes." Eliminate this in-stack competition locally, and you free up a great deal of energy and simplify control needs. This was (for different reasons) Carnegie's playbook in the early days of steel, and appears to be the playbook of many Internet era companies like Apple, as well as the new breed of "full-stack startups." In some ways, this is approach is a throwback to an earlier era of management, even though it is very effective, which is why I classify it as a purgatory state. It works by simplifying the problem through the application of more economic energy (i.e. money) rather than increasing the sophistication of management methods. Q4, the portfolio strategy, which combats VUCA by hedging in the right ways, is less popular now than it used to be at the peak of conglomerate thinking in the West, so I won't say much about it (though it appears to be undergoing a renaissance in Asia). Q1 though, is in my opinion, the most powerful response: direct and indeterminate. Change the nature of low-level tools, and you change the nature of the beast itself, making it fundamentally more capable and able to handle unspecified new patterns of change, within a larger operating envelope. When you trade up from a bicycle to a car, you may still have no idea where to go, but your effective range has increased. Corporations that use tools like Slack as the foundation of their operational structure are what I've taken to calling stream corporations.  Flux and Change in Stream Corporations Today, even the tiny, local, partial view of the corporation -- say a departmental You Are Here view -- is becoming rare. Even where it is available, it is generally useless.  Debates about flat versus tall structures and titles often miss the actual patterns of ongoing activity. In emerging stream corporations, running on Slack instead of email, staffed in opt-in ways, and invisible ecosystems on Github and social media activity backing up visible tip-of-the-iceberg engineering and marketing operations, org charts are a museum technology. Fluid role definitions and transient relationships states matter more than nominal boxes and reporting relationships. For a stream corporation, an org chart is increasingly a not-even-wrong construct for thinking about corporate anatomy. Because most of the structure is not formally created or under deterministic control. For example, a marketing department that attempts to "map" its social media promoters and influencers in detail and "manage" the underlying social graph, as though it were the org-chart of an in-house marketing staff, is doomed to failure. "Let's make a viral video" is an example of the basic category error of imposing a deterministic mental model onto a probabilistic control mechanism. It's a mistake comparable to thinking poker is chess. Stepping back from the specifics of conceptual and technological tools, stream corporations are what emerge when your dominant metaphor of organization is the flux/change metaphor rather than the more traditional mechanistic or organic metaphors. Flux/change is one of the eight metaphors of organization exhaustively studied by Gareth Morgan. What distinguishes it from the other seven is that it is elemental rather than anthropocentric: it has more in common with earth/fire/water/wind type gestalts than living-thing gestalts. In important ways, the flux/change metaphor is the most basic of Morgan's eight. Since the other seven are all anthropocentric to varying degrees (as the very words organization and corporation suggest), they tend to share the biases and blindspots that are absent in the flux/change model. The biggest shared bias of the anthropocentric-seven (mechanistic, organic, instrument of domination, brain, culture, political system and psychic prison) is that they tend to confuse social-psychological reality with all reality, and reductively think and manage in terms of purely social-psychological realities. This leads, for example, to the self-righteous culture-over-strategy meme common to all seven, and to the deeply misguided idea that preserving and extending the "lives" of corporations (and organizations in general) is ipso facto a good idea. The rise of the stream corporation and the flux/change metaphor is a challenge to anthropocentrism in business. Whether or not the challenge is Copernican, Darwinian or Watson-and-Crick-ian in profundity remains to be seen. All four of the adaptation approaches in the 2x2 above lean more or less strongly towards flux/change metaphors compared to org charts, but flux/change metaphors designed into low-level communication tools provide the most leverage for serious transformation. To understand tool-level transformation of corporations, you have to adopt an immersive perspective. In the stream, or flux/change metaphor, we let go of the desire for an up altogether, and orient instead around the direction of accelerating creative-destruction. The right question about a stream, or a set of streams, is not what is the hydrological map here? It is what is it like to swim in this particular stream? Creative destruction is something best felt viscerally rather than observed visually. In the flux/change metaphor, situation awareness lives primarily in the gut rather than in the mind. Inhabiting Stream Realities In a flux/change metaphor, structurally, the organization  is whoever happens to be present in this meeting, inhabiting  this patch of curved local space created by two pizzas on the table, and this frequency band of the atemporal Slack event stream  providing global context. Morale is the feel of this meeting or video chat. Most-used emoji are a better measure of culture than the success of the company picnic. Everything beyond the felt-present is speculation, and rumors of its "reality" have historically been greatly exaggerated. Mapping stream realities is hard, even with a lot of data and powerful dynamic tools. As the very name Slack suggests, often a smarter, lower-effort approach is to manage anxieties rather than realities. For those new to tools like Slack (and more specialized tools with similar DNA, such as Github), this anxiety often manifests as the urge to simply recreate the organization chart in the lineup of Slack channels (the basic building block of Slack is chat-like streams of interaction reminiscent of IRC and role-playing games called channels). Slack best practices suggest that this is a mistake. It is better to keep activity in general-purpose channels and only fork off new channels when sustained bottom-up conversational activity justifies it. These may or may not line up neatly with imagined organizational boundaries. Managing through Slack is more like administering a busy bulletin board (something I did back in the day) than managing a team in the traditional sense. Boundaries go from highly consequential constructs that control and direct activity to background book-keeping artifacts for the decreasing number of things that aren't stream-like yet. In the future, I suppose layoffs will become bleed-offs. Reorgs will become reroutings of flows. A snapshot of activity on a tool like Slack may not provide a satisfying understanding of what is going on. In fact, I'll make an even stronger statement: tools like Slack will work for you only to the extent they do not provide such an understanding. The functions of accurate appreciation of reality and effective manipulation of reality do not get along very well. One requires the detachment of distance, the other requires the live anxiety of immersion. One requires you to bracket your biases and broaden your vision. The other requires you to manage stress and master emotions. Tools that try to support both functions typically do both very badly (think steering wheel that also serves as a disc-shaped map display). Under low-VUCA conditions, this does not matter. Under high-VUCA conditions, this is fatal. So good flux/change metaphor products like Slack and Github are highly opinionated: they have features designed to actively resist the artificial imposition of anxiety-alleviating "understanding" on operational structures. My favorite features of this sort include the witty/philosophical aphorisms that show up on the Slack loading screen, gently nudging your thinking into a more relaxed state, and the personality of slack bots (sure to be of increasing importance as we employ more robots and algorithms in the workplace). This resistance combats the false sense of security fostered by inaccurate maps, and also helps combat the authoritarian high-modernist instincts of leaders inclined to be dictatorial in clueless ways. This means Slack, despite its name and persona is by design more anxiety-inducing than email, which may partly explain its success over other email-slayer products. By increasing the firehose volume, velocity and variety and taking away comforting organization tools like folders and attachments, Slack makes avoidance tactics harder, and forces you to actively manage information anxiety and stress by tackling it head on. In the Slack world, there is no Inbox Zero. Only a min-stress-max-flow state, if you will forgive the geek joke. And that is a good thing. Immersion in a flourishing flux/change product may feel like chaos, but in fact creates situation awareness of a Borgesian encyclopedia variety:

These ambiguities, redundancies and deficiencies remind us of those which doctor Franz Kuhn attributes to a certain Chinese encyclopaedia entitled 'Celestial Empire of benevolent Knowledge'. In its remote pages it is written that the animals are divided into: (a) belonging to the emperor, (b) embalmed, (c) tame, (d) sucking pigs, (e) sirens, (f) fabulous, (g) stray dogs, (h) included in the present classification, (i) frenzied, (j) innumerable, (k) drawn with a very fine camelhair brush, (l) et cetera, (m) having just broken the water pitcher, (n) that from a long way off look like flies.

(Aside, anybody want to create and run a performance-art slack site in honor of Borges, with these channel names?) This is not the ontology-folksonomy distinction. Both top-down ontologies and folksonomies are org-chart-like constructs, meant for appreciation. But this is not chaos either. The distinction here is the distinction between a model of a finished building and an inventory of an active construction site for that building. One contains things like columns and doorways. The other contains things like excavators and cranes. Because they focusing on flux/change and creative destruction, the tools of stream corporations highlight its nature as a constantly under-reconstruction entity. Whether through indirect or direct adaptation, under VUCA, the control structure embodied by a non-zombie corporation is being constantly reconfigured. Without opinionated workflow tools that "believe" in the corresponding adaptation model, this degenerates into ineffective thrashing. The sign is decreasing rather than increasing mastery of emotional churn and stress (note that I didn't say reduction of emotional churn and stress; that is a sign of disconnection and disorientation: going zombie).   When immersed in a stream, your view of the global environment is something like an evolving gestalt of emotion and tempo, rather than a mental model you can articulate or document. You may be able to answer questions like Is the organization healthy? How should we respond to this development? and Is morale high? without being able to answer questions like What's the overall structure of this organization? One way to think of this condition is this: you have perspective and control, even if you don't have an intellectually satisfying and global understanding of your situation. Perspective as Managed Anxiety In my last big post on corporations a few years ago, I argued that we're past Peak Attention and slowly climbing the slope of Peak Perspective. The shift to the flux/stream metaphor is one sign of that broader shift. When you embrace direct-indeterminate adaptation and immersive orientations, you gradually become capable of operating under increasing VUCA conditions, far past the point at which extrinsic orientation aids fail. Under such extreme conditions, your backstop is not objective knowledge but subjective perspective. So what is perspective? A 2x2 gets at the answer. Specifically, David Allen's 2x2 of perspective and control, which I offered as an example of a good one in How to Draw and Judge Quadrant Diagrams:   Whatever your chosen pattern of adaptive organizational response to organizational VUCA, the test of whether it works is whether it puts you in the top right quadrant of high perspective and high control. The master-and-commander zone, where you can, as David once put it (iirc), "steer with a light touch with one hand, with a glass of wine in the other." High perspective is not the same as accurate and detailed situation awareness (though David Allen's GTD is a way of getting there via accurate and detailed situation awareness). From conversations with David, I've come to think of perspective as a state of successfully managed stress and anxiety. If perspective is valuable for extending the range in which you can safely operate, it is because the real problem is not ignorance, but dealing with the anxiety and stress of not knowing. Of course, ignorance can kill, but more often, inability to manage anxiety and stress will kill you first. Having a perspective means you have identified and occupied a point of view that allows you to stay calm and in control of your emotions despite VUCA. You do not need to impose order and calm in order to function. Statements like in the long-term, we're all dead and don't sweat the small stuff are indicators of perspective states that are appropriate for specific situations. The serenity prayer is another example of a high perspective state. High perspective can exist despite extreme and perhaps insurmountable ambiguity in the environment and uncertainty about what to do. And it can be achieved without burying your head in the sand and going into denial (blindness is not a perspective). High perspective, high control is the cognitive stance associated with high effectiveness. Whatever your adaptation strategy and tools, it means you have self-awareness about your capabilities, and are mentally prepared to deal with chaos without flinching or resorting to knee-jerk emotional responses. Structure and Strategy Reconsidered The increasing importance of direct-and-indeterminate models of adaptation and associated tools signal a secular, rather than cyclic, shift in the economic environment. This is one reason this whole direction of thinking has been an active research and consulting interest of mine in the last couple of years. Within my modest means, I've been  rethinking Alfred Chandler's structure-follows-strategy thesis for the emerging era of stream-like organizations. The (largely invisible) rise of management philosophies grounded in acceptance of ambiguity and uncertainty rather than determinate control is not just a wild gonzo "phase" the new economy is going through, to be replaced by a more measured and "mature" phase in 2025 and a New Organization Man era. Like a stream transitioning from laminar to turbulent flow, we've moved to a qualitatively different economic environment, and there's really no going back. The collapse of organization charts in our thinking reflects a deeper collapse of the underlying imagined organizational realities. The organization chart is shrinking because we are slowly recognizing what has always been true: there is much less "organization" to chart than we'd like.