--- title: "Corporations and the Workplace" url: "https://toddpaulbrownjr.com/corrigibility/corporations-and-the-workplace/" author: "Todd Paul Brown Jr." description: "The dashboard that ate the mission, status rounding up the chain, and a company going deaf one reorg at a time, with an audit you can run from any seat." kind: "guide-chapter" updated: "2026-09-26T02:50:17+00:00" --- # Corporations and the Workplace ## A normal Monday Picture a support team at a mid-size company — software, insurance, logistics, it doesn't much matter. On a shared screen in the open-plan office sits a dashboard nobody remembers deciding to build; it just accreted, one metric at a time, until it became the way the team understands itself. Average time to close a ticket: down again this quarter. Tickets closed per agent per day: up, third quarter running. First-contact resolution rate: green, and it has been green for months. The support director walks these numbers into the Monday leadership meeting and gets an approving nod from the COO. Nobody in that room is lying. Nobody is cutting a corner they know is a corner. By every number the company has agreed to track, it is a good quarter, and the director has every reason to be proud of the team she built. Meanwhile, a slower number never made it onto the dashboard: the rate at which the company's best accounts quietly fail to renew. If someone had asked, in that same meeting, "why are we losing them," the honest answer would start with the tickets. Nobody taught the agents to game the count. They learned it the way people learn most of what actually governs their behavior, by watching what gets rewarded. A closed ticket counts. A reopened one drags the average down and gets flagged in a one-on-one. So a hard, half-solved problem gets marked resolved before the call ends, the customer gives up explaining it a second time, and eventually stops calling at all. A ticket that never comes back in looks, on a dashboard, exactly like a problem that got solved. The two are indistinguishable from where the director is standing, and she is standing exactly where the company put her. Ask any individual agent about a specific case and they'll tell you the truth without hesitation: that account's billing issue never actually got fixed, it just stopped coming up. Ask why they closed it anyway and the answer is not evasive. It's almost bureaucratic: the queue was long, the metric was watched, and a closed ticket is what "done" looks like on the only scoreboard anyone hands out bonuses against. Multiply that one small, sensible decision by forty agents and eighteen months and you get a support organization that is, on paper, the best-performing team the company has ever had, managing the exit of its most valuable customers one satisfied-looking ticket at a time. This is not a story about villains, and that's the point of opening here instead of with a scandal. Nobody in this building set out to hollow out a customer relationship. The director earned her pride on the terms she was handed. The COO isn't a fool; he is looking, diligently, at the number he was told to look at. This is where most readers of this book actually work: not in a courtroom, not in a headline, but in an ordinary building on an ordinary Monday, where every pathology this book has named is already active, dressed in office clothes, doing no harm that anyone in the room can currently see. ## The metric that ate the mission The support team's dashboard is a small, clean case of what Chapter 4 named *sucralosis*: a system optimizes a stand-in for a real result until the pressure of optimizing makes the stand-in and the result pull apart, and the system keeps optimizing the stand-in anyway. The name borrows from the artificial sweetener: all the sweetness, none of the nutrition. In an office, the shape is exactly this. Closing tickets was always supposed to be a proxy for solving customers' problems, a fast, countable stand-in for something slower and harder to measure. For a while the two moved together; most tickets that closed really were resolved, so the count was a decent mirror of the mission. Then the team got good enough, and pressured enough, at hitting the number that the fastest way to move it stopped being "solve the problem" and started being "close the ticket." The two paths diverged without anyone voting on it or announcing it. The same shape shows up wherever a team gets handed a countable proxy for something that resists counting. A consulting team's utilization rate is supposed to track whether the firm is deploying its people well; push hard enough and it starts tracking whether people can be kept visibly busy, whether or not the work is worth doing. A software team's velocity, story points shipped per sprint, is supposed to track whether the product is getting better; push hard enough and it starts tracking whether tickets are sized small enough to close fast, whether or not the software works. A sales team's quota is supposed to track whether the company is winning customers who will stay; push hard enough on the quota alone and it starts tracking how many contracts can be signed before the onboarding team discovers the customer was never going to be a fit. None of this requires a bad actor. It requires only a proxy, sustained pressure, and enough time for people to discover, individually and without collusion, which behaviors move the number that gets watched. The test that keeps this diagnosis from turning into a reflex ("the boss measures things, therefore sucralosis") is the one Chapter 4 built. Ask what the metric was adopted to track in the first place. Then ask which one wins when the metric and the thing it was tracking pull in different directions, and whether anyone in the building is allowed to say so out loud. Ticket closure was adopted because it correlated, once, with solved problems. The divergence is measurable, not mystical: pull thirty "resolved" tickets and count how many of those customers called back about the same issue within a month. If the metric keeps climbing while that number climbs with it, the proxy has come loose from the mission, and the honest question is whether the team is allowed to say so in a review, or whether saying so reads as an excuse for missing target. The same test draws the line this book insists on drawing every time. A team that says, out loud and on a recurring cadence, "we track ticket closure as an imperfect signal, and we check it every quarter against callback rate and churn," is *using* a proxy, not captured by it. So is a team that consciously trades quality for speed to hit a launch date and says so in the postmortem. That's a trade-off made in daylight, not a substitution smuggled past anyone's notice. Sucralosis needs the substitution to be quiet. The moment a team can name the gap between the number and the mission in a room where it costs them nothing to say it, they've done the thing that makes the proxy safe to use. ## The manage-upward engine Here is the idea in this book most corporate readers will recognize before they finish the sentence. Sucralosis doesn't only happen to teams optimizing a metric. It happens to individuals optimizing a person: specifically, their manager's *perception* of their work, rather than the work. Most people have watched this happen, and most have done a version of it themselves. An employee learns, correctly, that what determines their raise, their next assignment and their standing in the room is not the quality of the underlying work but how that work reads to the person one level up. So the behavior that actually gets optimized is the update, the framing, the well-timed visibility of effort, rather than the contribution. This is individually rational. It is also, on its own, forgivable: most people are responding to the incentives the building actually runs on, and doing otherwise would mean losing out to the person who doesn't. The trouble starts when you notice that the manager doing the perceiving is also, one layer up, managing someone else's perception of *her*. She isn't reporting the raw state of her team; she's reporting the version of it that reads well to her boss. Her boss does the same thing one layer up. Nobody at any layer is lying, exactly. Most of what travels upward is true, selectively framed, optimistically rounded, timed for a good week. But each layer is a proxy for the layer below it, and each proxy is under the same pressure the ticket count was under: to drift from what it was supposed to represent, quietly, without anyone deciding to let it. Run that up five or six layers of a real organization and the arithmetic gets brutal. Picture it moving one rung at a time. An engineer knows a project is six weeks behind and tells her team lead it's "on track, minor risk." The team lead, who has his own quarterly review coming up, tells the director it's "green with a small watch item." The director, briefing the VP, drops the watch item entirely; it isn't worth the VP's limited attention for something that will probably resolve itself. The VP tells the executive team the portfolio is healthy. Nobody in that chain said anything false. Each rounding was small, defensible, and made in good faith by someone protecting a reasonable amount of their own credibility. By the time the picture reaches the top, the distortion isn't small; it compounds, each layer's honest rounding added to the last. The executive team ends up looking at a clean, internally consistent dashboard of a company that, in the ways that matter, no longer exists: executing well, morale fine, roadmap on track, according to a chain of reasonable people each optimizing the perception one level up rather than the reality one level down. This is one mechanism behind a familiar corporate experience: the missed product cycle or the customer exodus that reaches leadership as a sudden surprise, when people three and four layers down saw it coming for a year and had no channel that would carry the news upward without smoothing it on the way. The test still applies here, exactly as it did with the ticket count: which wins when the accurate picture and the flattering one collide, and can anyone in the chain say so without cost? A department where bad news travels upward at the same speed as good news is using status reports as an imperfect but working proxy. A department where bad news arrives late, softened, or not at all is not being run by dishonest people. It is being run by people responding, individually and sensibly, to what happened to the last person who reported something the layer above didn't want to hear. That question, what happened to the last person who said the uncomfortable thing, is worth holding onto. It returns later in this chapter as the sharpest tool in the workplace kit. ## Cancrity at company scale A team can also do to the company what the ticket count did to the customer relationship: hit its own numbers by consuming something the company depends on faster than that something gets replaced. Chapter 5 named this *cancrity*. Its source model is a single cell thriving locally while degrading the tissue it depends on; here the same failure shape appears in entirely different machinery, a balance sheet. A sales team that burns through its best account managers to make quota, with six-day weeks, unreachable targets and a revolving door, can hit every number on its own dashboard while the company's hiring pipeline can't refill the seats fast enough to keep pace. A marketing team can spend down a brand's earned trust for one strong quarter of conversions. A leadership team can make next quarter's margin look excellent by cutting deeply into the people who hold the company's working knowledge, and discover the quarter after that what the knowledge was holding up. The discipline that keeps this diagnosis honest, the one Chapter 5 insisted on, is naming the substrate specifically. Not "this team is bad for the company," but "this team is hitting its numbers by consuming ______ faster than it's being replenished," filled in with something you can actually watch move: senior-engineer retention, customer trust as measured by repeat purchase, the on-call rotation's sustainable headcount. A diagnosis that can't fill in that blank with something trackable isn't a cancrity diagnosis yet; it's a complaint wearing the vocabulary. This is also what separates cancrity from ordinary competition. A rival team that wins your best engineer in an internal transfer, fair and square, hasn't consumed anything; the company still has the engineer. A team that burns an engineer out until they leave the company entirely has consumed something that doesn't come back, and the difference between those two outcomes is the whole diagnosis. Chapter 5 told the story of a major department store chain that split itself into dozens of internal units, each run as its own profit-and-loss center and competing with the others for capital, marketing and executive attention inside the same company. Each unit could look, on its own dashboard, like a disciplined performer. Critics of the arrangement, including former executives, argued that the units were collectively drawing down the shared stores, shared brand and shared trust that let any of them function as a retailer at all. Because each unit was measured against the others rather than against the health of the whole, nothing in the structure was positioned to notice the shared substrate thinning. The exclusion case matters as much as the diagnosis, because this is exactly where the word gets misused into a slur. A startup that spends down its runway deliberately, with a stated plan and a board that signed off on the burn rate, is not cancritic. The coupling between spending and outcome is intact and spoken out loud; everyone involved knows which substrate is being drawn down and has agreed to the terms. Cancrity requires the consumption to be decoupled from anyone tracking it: not a chosen bet, but an unexamined one. ## One reorg at a time The harder failure isn't a bad number; it's a company that has quietly lost the ability to hear that the number is bad. This rarely happens as one decision. It happens the way the manage-upward engine happens, one individually defensible act at a time, none of them sinister on its own. A team runs an honest internal review and flags a real structural problem with how the product is built; six weeks later a restructuring is announced, and the team that wrote the report no longer exists in its old form, its members scattered to other groups "to better align capabilities." An engagement survey comes back with a bad score on one specific question, whether people feel safe raising concerns, and the response is a communication initiative: a town hall, a new internal newsletter, a poster campaign about open doors. The question doesn't get asked a second time the next year, because the initiative has become the answer. A senior engineer who has raised the same uncomfortable point in three consecutive planning cycles is managed out as "not a culture fit," and the point goes with him. No single one of these events looks alarming. Reorgs happen for real reasons. Communication initiatives are sometimes exactly what's needed. Not every departure is a purge. That is what makes the pattern hard to see from inside: each event has an innocent, plausible explanation, and each explanation is even true, as far as it goes. The team really was inefficiently structured. The newsletter really might help morale. The senior engineer really had become difficult to work with, in ways that had nothing to do with the point he kept raising. But run enough of these reasonable events through a company over a few years and the aggregate is unmistakable: every channel through which bad news could travel upward without being deflected, absorbed or disbanded has, one at a time, stopped working. Nobody signed an order to disable the company's correction machinery. It disabled itself the way rust happens, one small, locally sensible decision after another. Chapter 6 built the distinction that separates two workplaces that can look identical from outside. A company is *incorrigible* when it genuinely can't see the problem: the information isn't reaching the room, the metric hasn't been built, nobody has connected the dots. That's fixable. Hand them the missing information, build the missing dashboard, and a company that is merely incorrigible will generally use it. A company is *contracorrigible* when the review process itself is owned by the thing that most needs reviewing: raising a concern through the official channel routes it to the very people whose performance the concern implicates, and the channel reliably produces a restructuring, a rebrand or a departure instead of a fix. That is not fixable from inside, because the tool you'd use to fix it has been repurposed. The practical tell, usable by anyone without access to the org chart or the board minutes, is the question already introduced: what happened to the last person who raised this? An answer like "it got addressed, imperfectly, but addressed" points away from contracorrigibility: the channel works, so push on it. An answer like "they don't work here anymore" or "their team got reorganized" points the other way. The pattern takes a recognizable shape at the top of a company whose identity has fused with its founder's. Chapter 7 named the mechanism by which a serious structural critique becomes invisible to exactly the person who most needs to hear it: not rejected, which would require the critique to arrive as what it is, but converted on contact into something smaller and safer to process. A founder for whom the company is not a project but a self hears "the strategy is failing" and cannot let that sentence land as stated, because there is no self standing apart from the company to receive it. What lands instead, reliably, is "here are three tactical issues to fix": a marketing tweak, a hiring plan, a pricing adjustment. The correction-shaped conversation happens. The correction does not. Everyone in the room can point to the meeting where the hard feedback was "addressed," and the strategy keeps failing on schedule. ## Running the audit when you don't own the place Most readers of this book are not the CEO, and the natural response to four pages about a company's correction machinery breaking is: fine, but I can't restructure the org chart. You don't need to. Chapter 8's telonic audit is three questions, and all three can be asked from wherever you sit. The first question, what does this team or company claim to be for and what does it actually reward, is answerable from your own desk. Compare the mission statement to the last five promotions. What got someone promoted, and does it match what the mission statement says the company is for? The second question, what is my part of the organization consuming and is it being replenished, is answerable by naming the substrate the way Chapter 5 insists: not "my team is toxic" but "my team is hitting deadline targets by consuming the on-call rotation's sustainable staffing faster than we're hiring for it," a sentence you can test against a headcount spreadsheet. The third question costs the most and tells you the most: who has the standing to raise the first two out loud, and what happens to them when they do? You don't need subpoena power to answer it. You need one data point, the last person in your building who said the uncomfortable true thing in a room where it mattered, and an honest look at where they are now. The three answers together tell you something concrete: whether you're in a system that's incorrigible and can still be helped by better information reaching the right room, one you should quietly route around while doing your actual job, or one worth leaving before its trajectory becomes your résumé's problem too. None of this requires organizational power. It requires being willing to look, and willing to write the answer down instead of letting it stay a vague feeling you carry into every Monday. A vague feeling is easy to talk yourself out of. Three specific answers, dated and reviewed again in six months, are not. It does require one thing that isn't information at all: the standing to survive being right. That is what Chapter 12 called *telonic grounding*, and Chapter 14 develops it fully: an identity spread across enough independent commitments that losing any one of them wouldn't dissolve the person. At work the payoff is blunt. The employee whose whole sense of self runs through the job can't afford to see the job clearly, because seeing it clearly might cost them the job. The employee with a life outside the org chart can afford to ask the third audit question out loud. That isn't a mindset to adopt the morning of a hard conversation. It's a position built in advance. ## The corrigible team None of this is only diagnostic. Chapter 9 described what a healthy system's correction machinery looks like from the inside, and at team scale it is not abstract. It is a short list of things an actual manager can build. First, dissent has to be *exercised*, not merely permitted. A dissent channel nobody uses is functionally identical to one that doesn't exist; the objection has to be a normal, expected, minuted part of how the team works, not a brave exception. A team where the junior engineer's objection in the design review gets written down, tested against the plan, and sometimes wins is exercising the machinery that catches drift before it compounds. A pre-mortem, sitting down before a launch and asking in writing "imagine this failed; why did it fail," does the same work on a schedule instead of waiting for a crisis to force the question. Second, the team needs enough internal variety to notice its own blind spot, a version of what Chapter 9 called requisite variety. A team whose members trained in the same discipline, came up through the same kind of role, and share the same assumptions about what matters cannot see the class of failure that falls outside those assumptions, because nobody in the room is positioned to notice it. This is a hiring and staffing decision, not a slogan: deliberately putting people with different disciplines, different prior jobs and different experience of failure into the room, and giving them real weight, specifically because the room's current center of gravity has a blind spot it cannot see from inside. Third, the health check has to be about presence, not absence. "No complaints reached me" is not evidence of a healthy team; it's equally consistent with a healthy team and with one where raising a complaint has become expensive. The better question, asked on a recurring basis: when did we last change course because someone junior turned out to be right? A team that can answer with a specific instance from the last quarter has working correction machinery. A team that has to reach back years, or can't answer at all, has a channel that looks open and isn't being used, which, per the first point, means it has already atrophied into looking exactly like a channel that doesn't exist. A manager who wants a harder version of the test can ask it about themselves: when did a direct report last change my mind about something I'd already decided, in front of the rest of the team? If the honest answer is "I can't remember," the team has learned something true about what happens to disagreement in that room, whether or not it was ever said out loud. One habit ties all three together and costs almost nothing: before a big initiative launches, write down what would tell the team it's failing. Not after the fact, when every result can be reframed as progress by someone invested in the outcome, but before, while the team can still be honest about what a bad sign would look like. This is the disconfirming-conditions test from Chapter 9, applied to a single product bet instead of a whole institution. It is cheap, hard to fake after the fact, and does more to shape a team's culture over a year than any values poster on the wall. ## The workplace kit A short set of questions, worth keeping somewhere you'll actually see them again: - **The manage-upward mirror.** What am I currently optimizing, the work or my boss's perception of the work? When those point in different directions, which one am I actually choosing, and would I say so out loud? - **The one-line cancrity form.** Fill in the blank for your own team: we are hitting our numbers by consuming ______ faster than it's being replenished. If you can't fill it with something you could plot on a chart, the diagnosis isn't ready yet. - **The standing question.** What happened to the last person in this building who raised the uncomfortable true thing? That single answer tells you more about whether the place can hear correction than any stated value ever will. - **The pre-registered failure test.** For whatever initiative is currently the big bet, write down now, not later, what result would tell you it's failing. If nobody can name one, the initiative can't lose, which means it also can't be corrected. ## Where this leaves you The workplace is usually where these forces are first felt, not as ideas in a book but as something happening *to* someone: a good employee's honest report deflected by a reorg, a manager's real concern converted into three tactical fixes, a team's mission quietly replaced by its dashboard. The patterns this book has traced through cells, companies and crowds show up here first because this is the building most readers spend the most waking hours inside. The tools in this chapter don't require you to run the place. They require you to look at the place you're already in, clearly, and to keep enough of yourself outside it that looking clearly stays affordable. The next chapter turns the same instruments on the one system every reader runs: not the team, not the company, but the person reading this sentence, a system that can be captured by the same mechanisms and can, with the same tools, stay correctable.