The Transition Most Organizations Haven't Made

Most organizations that adopted OKRs over the last decade or two report something is wrong with them. The Objectives sound aspirational. The Key Results get hit. The dashboards report green. Year after year, the work does not move the organization in the direction the Objectives named. The vocabulary of purpose, alignment, and outcomes is everywhere in modern management. The results that vocabulary was meant to produce are not.

The standard explanations for this gap are operational. Goals were not specific enough. Alignment was not cascaded properly. OKRs were used wrongly. Managers did not communicate the why. Each of these explanations is plausible. None is the structural mechanism. The mechanism is not in any of these places. It is in something that happened to management theory between roughly 1954 and 1968, and that has been quietly determining how organizations evaluate work ever since.

The claim is straightforward, and somewhat uncomfortable. The structural distinction between a goal and a task is the management-theory expression of Russell Ackoff's Machine Age to System Age transition. Most organizations live in 2026 with System Age vocabulary running on Machine Age infrastructure. Goals become tasks at evaluation because the infrastructure can only evaluate task-shape. Telling the two apart, structurally apart, is itself the move that the management field crossed in 1954, walked back from in 1968, and has not collectively re-crossed since.

This article is longer than usual because three ideas need careful introduction: what a real commitment is, what makes a goal structurally different from a task, and what Ackoff meant by Machine Age and System Age. With those in place, the diagnosis is straightforward, and considerably more useful than "your OKRs are bad," because it identifies what would actually have to change for them to stop being bad.

What a commitment is

Tom Jans applies Margaret Gilbert's philosophical definition of commitment to organizational teamwork: a person P is committed to act A if and only if P has sufficient reason to do A. A commitment begins with a decision to commit and ends with a decision to release. Three features matter:

A commitment is to an act, not a result. You commit to acting in pursuit of a state, not to the state itself. The state is what your acts may or may not produce. "Commit to a trend," "commit to an outcome," "commit to a number" are category errors; the real commitment is to the acts that aim at them. The result is what gets evaluated; the commitment is what gets undertaken.

Commitment requires a real decision. A genuine "no" must be a real option. You cannot commit to gravity (no choice). You cannot commit to never eating (no choice that does not kill you). In employment, the test is rarely a clean binary; "no" sits on a spectrum from "can walk away without consequence" to "the mortgage depends on this cycle." Most cascaded organizational commitments sit at the low-no end of the spectrum: the worker has no real "no" available. What they have is an instruction wearing the word.

Strong commitment requires clarity. If you do not know what the act requires, what difficulties will arise, or what competing commitments will pull at you, the commitment will be brittle when the unknown turns out to matter.

A "commitment" that fails any of these is not a commitment in Gilbert's sense. It is a wish, an instruction, or an aspiration wearing the word.

The two commitment-shapes

A real commitment, then, has a shape, and the shape determines what kind of evaluation makes sense for it. Two shapes are structurally distinct, and the difference is in what the acts aim at and how the work is evaluated against the aim.

A task is a commitment to act in pursuit of a specified state. Ship Feature X by Q3. Migrate the platform to Kubernetes by Q2. Win three enterprise deals this quarter. Hire fifteen engineers by year-end. The state is fixed in advance and the evaluation is obvious: was the state reached? At assessment, the question is binary. Hit or miss.

Tasks are correct for a large fraction of organizational work. Most operational delivery, most contractually-scoped engagements, most repeatable execution is correctly task-shaped. Calling such work a task removes ambiguity at evaluation. There is nothing wrong with tasks; the failure mode this article describes is not "having tasks."

A goal is structurally different. A goal is a commitment to act in pursuit of moving outcomes in an intended direction. The shape has three structural parts, all of which must be present.

First, a trend: the intended direction of movement in outcomes (the trajectory the work is supposed to produce, not whatever happens by accident). Without it, no result can be evaluated as desired or undesired; the number alone tells you nothing.

Second, a target: a magnitude that gives the trend measurable shape, not as a contractual threshold but as the worker's current best estimate of where the trend should land.

Third, an evaluation that respects both: at assessment, the trend direction, the target magnitude, and the learning the work produced are all checked. Did we move the right way? Did we approach the target? What did we discover about whether the target itself was right?

A goal also requires revision authority: the worker has standing to revise the target mid-cycle when learning suggests the target was wrong. Without revision authority, the worker is locked into a target that may have been set on incomplete information, which structurally collapses the goal back into a task. (This applies to learning-context goals, where the target is a hypothesis being tested. Constraint-context commitments such as safety thresholds, regulatory floors, and contractual SLAs are correctly non-revisable; their target is a non-negotiable boundary, not a hypothesis. Constraints are a third commitment-shape distinct from both task and goal, and they are evaluated as binary thresholds, which is task-evaluation logic applied to a different kind of work.)

Goals are correct for work where the desired direction is known but the path must adapt to learning. Reducing customer churn. Improving onboarding effectiveness. Strengthening a capability. Expanding into a new market. The path has to unfold; the target is a hypothesis about where the trend should land.

These two shapes are not the same kind of thing. A goal is not "a big task" or "a long-term task." A goal is structurally distinct from a task, requires different evaluation infrastructure, and produces different artifacts (learning, direction-revisions, trend judgments) that a task does not produce. At the planning layer, real work often sits on a continuum between pure task and pure goal; at the evaluation layer, the system collapses every commitment to one shape or the other, which is what makes the binary decisive. What follows depends on this distinction holding.

Two ages

The Machine Age and System Age frame comes from Russell Ackoff's Redesigning the Future (1974), applied specifically to organizational design in Re-Creating the Corporation (1999).

In Ackoff's frame, the Machine Age is the dominant intellectual mode of roughly the seventeenth through mid-twentieth centuries: the era of Newton, the Industrial Revolution, and Frederick Taylor's scientific management. The Machine Age treats organizations as machines. A machine has no purpose of its own; it is built by a designer to serve a purpose external to itself. Its parts are interchangeable, decomposable, and individually replaceable. Each part executes a prescribed function. The machine is understood by analyzing the parts; the whole is the sum of the parts.

In the System Age, by contrast, organizations are understood as purposeful social systems composed of purposeful parts. A purposeful system has its own aims and can choose among them. Its parts are not interchangeable; each is itself a system with its own aims. The whole cannot be understood by analyzing the parts in isolation, because the parts' behavior depends on their role in the whole, and the whole's behavior depends on the system's purpose.

The transition in intellectual terms is well-named and largely complete. Anyone who has read modern management literature has encountered the System Age vocabulary: purpose, mission, outcomes, alignment, emergence, complexity, learning. The transition in operational terms has barely happened. Most organizational structures (org chart hierarchy, performance review software, OKR cascade, bonus calculation, board reporting) are built on Machine Age assumptions. They are good at evaluating tasks. They cannot evaluate goals.

This vocabulary-infrastructure mismatch is what produces the OKR rollout pattern described in the opening. The vocabulary is System Age. The infrastructure is Machine Age. The vocabulary describes work that the infrastructure cannot evaluate, so the infrastructure quietly converts the work into the shape it can evaluate, which is task-shape. The dashboards stay green; the work does not move the organization. This is not a misunderstanding; it is exactly what the structures were built to produce.

Why Machine Age has only one commitment-shape

Machine Age ontology has only one valid commitment-shape: decomposable attainment-units. Every commitment is, in the end, a unit of decomposition with a measurable state to reach. You can vary scope (a quarter, a year, a decade). You can vary specificity (tight versus loose). You can vary scale (a single deliverable, a multi-team initiative). But the underlying claim is always the same: work consists of units that can be specified in advance, decomposed into subunits, and evaluated by whether each unit's output matched its specification.

In Machine Age, "goal" and "task" are not structurally distinct. They are different scopes of the same kind of object. A goal is a "big task." A long-term objective is a "task with a longer cycle." A strategic outcome is a "task at the level of the whole organization." All work shares the same shape; only scale varies. There is no need for "goal" as a separate structural category, because the system has no direction-setting capacity distinct from the parts' execution. The system's purpose, in Machine Age, is just the sum of the parts' attainment.

This is the operating ontology behind classical management theory. Taylor's task management treated the firm as a designed machine whose parts execute against specification. Fayol's planning function added administrative direction-setting on top, but the direction-setting decomposed into tasks below. Two parallel vocabularies coexisted for four decades (scientific management spoke of tasks, administrative management spoke of objectives), but neither distinguished goal-shape from task-shape structurally. Machine Age had no slot for the distinction.

Why System Age requires two commitment-shapes

To distinguish goal from task structurally, you have to first hold the system as purposeful. You have to commit to the claim that the system has direction-setting capacity that is not reducible to the parts' execution. This is the System Age commitment.

Once the system is purposeful, the distinction becomes meaningful. Some commitments are correctly task-shape: work where the deliverable is known and the path can be planned up front. Other commitments are correctly goal-shape: work where the desired direction is known but the path must adapt to learning, where the target is a hypothesis rather than a contracted threshold, where the worker must have revision authority because the system is moving through an uncertain environment.

The recognition itself is the threshold. Machine Age cannot make this move because Machine Age has no concept of system purpose distinct from part execution. System Age requires the move because purposeful systems have direction-setting needs that task-shape commitments cannot meet.

This is why the act of codifying the goal-vs-task distinction is itself a System Age move. It is not a vocabulary preference. It is the operational expression of having crossed the intellectual threshold. An organization that genuinely operates in the System Age will have evaluation infrastructure that distinguishes goal-shape from task-shape, will treat them differently, and will preserve direction-revision authority where it is needed. An organization that has crossed only linguistically will have System Age vocabulary applied to Machine Age evaluation, and goals will become tasks at quarter-end, every quarter, regardless of how they were set.

The historical timeline as Age transition

The history of management thought on goals and tasks reads, with one important irregularity, as the Machine Age to System Age transition playing out in commitment-shape.

Pre-1950: Machine Age explicit. Taylor's task management (1911) and Fayol's administrative management (1916) are pure Machine Age. Workers as parts. Prescribed execution. Attainment-graded. Goals not in the vocabulary as a structurally distinct category.

1926: Mary Parker Follett. Follett's essay "The Giving of Orders" argued that authority should derive from the situation rather than from hierarchical position, and that workers should participate in defining their own objectives. Drucker later credited Follett as a source for what became Management by Objectives. Her work foreshadowed the System Age commitment-shape but did not name a system.

1954: Drucker as System Age threshold expressed in pre-Ackoffian language. Peter Drucker's The Practice of Management coined "Management by Objectives." The chapter title was "Management by Objectives and Self-Control." Drucker's claim was structural: objectives only function structurally when paired with self-control, where the worker measures their own performance against the objectives and uses the measurement to direct their own work. In Drucker's words: "Each manager should have the information he needs to measure his own performance and should receive it soon enough to make any changes necessary for the desired results."

This is System Age commitment-shape articulated twenty years before Ackoff named the System Age. The worker has context the cascade does not. The worker revises. The system needs the worker's purpose to function. Drucker did not have the philosophical apparatus to name what he was doing; the apparatus would arrive with Bertalanffy's General System Theory (1968) and Ackoff's Redesigning the Future (1974).

1968: Locke as Machine Age regression. Edwin Locke's foundational paper was titled "Toward a Theory of Task Motivation and Incentives". The title is decisive. Locke was studying tasks. The key finding was that specific, difficult targets within tasks produced higher performance than vague urging. Locke called those targets "goals."

This is the academic legitimization point for treating goals as specific difficult states to attain. The research tradition that became "goal-setting theory" started by studying tasks and named the specific-target feature of tasks as "goal." From its origin in management psychology, "goal" was structurally a difficult task target. Locke's 1981 formal definition of goal: "what an individual is trying to accomplish; it is the object or aim of an action." Object, aim, accomplishment, all state-shape. The trend dimension is absent from the definition.

By 1990, Locke and Gary Latham's A Theory of Goal Setting and Task Performance consolidated three decades of research into a single book. The title pairs goal setting with task performance as the same domain. Their 50-year retrospective (2019) confirms specificity and difficulty as the central performance findings, with no structural slot for trend evaluation or revision authority.

Specificity, in Locke's empirical research, is what made targets effective. In structural terms, specificity is what makes a goal task-shaped (a precise state to attain). The dominant academic tradition of "goal-setting" therefore walked the management research field back to Machine Age structure under System Age branding. The research tradition that practitioners cite when defending "goal-setting" as a serious discipline is, structurally, a Machine Age research tradition.

1974: Ackoff names the Age transition. Redesigning the Future makes the Machine Age vs System Age distinction explicit. By this point, in management theory, two traditions are alive simultaneously: Drucker's structurally System Age framing of objectives plus self-control, and Locke's structurally Machine Age framing of specific difficult goals. Ackoff's frame would let you tell them apart, but the frame did not get applied to the goal/task question by Ackoff or his immediate followers, and the management field as a whole did not connect the two.

Late 1970s: Andy Grove at Intel. Grove introduced "iMBO" (Intel Management by Objectives) at Intel, later renamed Objectives and Key Results. Grove documented the system in High Output Management (1983). The framework was Drucker's MBO adapted for tech: quarterly cycles, key results graded as attainment scores, cascade through the organization. Grove's adaptation kept the "objectives" vocabulary; the operational structure was attainment-graded.

1981: SMART codifies specificity. George Doran's Management Review article "There's a S.M.A.R.T. way to write management's goals and objectives" translated Locke's specificity finding into a popular checklist. Doran's original "A" was Assignable (who will do it), making the task-execution origin explicit. The drift to "Achievable" came later via training-industry adoption. SMART made specificity mandatory in any framework that adopted it. Specificity is what makes a goal task-shaped. The framework that became the canonical "goal-setting" tool worldwide was structurally task-grammar.

1999: Doerr brings OKR to Google. John Doerr, who had encountered iMBO at Intel in 1975, introduced OKR to Google in 1999. The framework spread rapidly through Silicon Valley and then globally. By the 2010s, OKR cascades had become the default goal-management infrastructure across tech, with downstream adoption in finance, healthcare, government, and non-profits.

2010s: infrastructure lock-in. Performance review software, OKR vendor platforms, bonus calculation systems, and board reporting infrastructure all built around binary attainment. The Objective layer declared a trend; the Key Result layer specified a number; the evaluation, at quarter-end, asked one question: did you hit it? David Hurst's analysis of MBO's failure modes documents that the self-control aspect was "lost in translation from a German cultural context (shared by both Drucker and Follett) to an American one." Drucker's emphasis on self-discipline got swamped by his injunction to "set clear objectives," and the objectives won.

The intellectual movement was therefore not unidirectional. Drucker articulated System Age commitment-shape before Ackoff named the Age. Locke walked the field back to Machine Age structure in the same year Bertalanffy formalized General System Theory. The Locke tradition won the operational battle (SMART, OKR cascade, performance review software). The Drucker tradition survived as quotable rhetoric in management literature.

Why the conflation persists

The conflation between goal and task in modern organizational practice is not corporate sloppiness or post-Drucker drift. It is the empirical evidence that the Machine Age to System Age transition has not happened in most organizations.

The vocabulary moved. Purpose. Mission. Outcomes. Goals. Alignment. Emergence. Complexity. Learning. These are System Age words; they appear in every modern management text, in every OKR vendor's marketing, in every board-level strategy document.

The structures stayed. Cascade. Attainment grading. Hit-or-miss bonuses. Performance review software designed around legible binary signals. Annual goal-setting cycles that lock targets at the start. These are Machine Age infrastructures.

The persistence is not only intellectual inertia. External pressures select for Machine Age infrastructure regardless of what the firm believes internally. Public-company quarterly reporting demands legible, comparable, audit-grade numbers. Boards need binary signals to assess executive performance. Compensation systems need defensible numerical criteria for legal and pay-equity reasons. Investor relations needs forward guidance that is measurable. Each of these pressures rewards task-shape evaluation infrastructure and penalizes anything more nuanced. A firm that intellectually accepts the System Age framing will still build Machine Age infrastructure if its capital markets, regulators, and compensation lawyers all reward Machine Age legibility.

When System Age vocabulary is applied to Machine Age infrastructure, the words function as decoration. The infrastructure can only evaluate task-shape, because task-shape is the only commitment-shape Machine Age structures know how to assess. Whatever was set as an Objective becomes evaluated as an attainment. The trend lives in the words. The target lives in the evaluation. The dashboards report green; the work does not move the organization. This pattern is not a misunderstanding of OKR. It is exactly what Machine Age structures were built to produce when given System Age inputs.

This is why fixing the conflation requires more than "better goal-setting" or "more sophisticated OKR practice." Better practice does not change the structure. Better goals do not survive a Machine Age evaluation infrastructure; they get converted into tasks at the moment of assessment. The intervention that would actually work is structural: complete the Machine Age to System Age transition for the specific subsystem where direction-setting is needed.

The diagnosis upgrade

This frame upgrades the goal-task conversation from a goal-setting critique into something more useful: a structural-Age diagnostic.

When a client organization reports that their OKRs are not producing the directional change the Objectives named, the surface diagnosis is "your goals became tasks at evaluation." The deeper diagnosis is "your evaluation infrastructure was built in the Machine Age and cannot assess goal-shape commitments, regardless of how they were set." The deeper diagnosis is more useful because it identifies what would actually have to be redesigned.

The diagnosis can be tested without a long engagement. Pick one Objective in the current cycle. Ask three questions of how it will be evaluated.

First, will the evaluation check whether outcomes moved in the direction the Objective named, or only whether the Key Result number was hit? If only the number, the evaluation is task-shape, regardless of how the Objective was written.

Second, is the Key Result treated as the worker's current best estimate of where the trend should land, or as a contractual threshold whose miss equals failure? If the latter, the evaluation is task-shape, and the trend dimension has been stripped.

Third, can the worker revise the Key Result mid-cycle when learning suggests the original number was wrong? If not, the worker has no goal-shape authority; what they have is a task with a number on it.

In almost every case, the answers will sit on the task-side for all three questions. This means the work is structurally task-shape regardless of what was written at the start of the cycle. This means, by extension, that the organization's goal-evaluation infrastructure is Machine Age, regardless of how much System Age vocabulary surrounds it.

The diagnosis "your goals are tasks" is structurally equivalent to the diagnosis "your organization is still operating in the Machine Age." This is a stronger claim than a goal-setting critique. It names the underlying ontological commitment that produced the surface symptom.

What it means for organization design

For readers who have not encountered the term, organization design names the discipline of building the structure in which an organization runs work and sets direction at every level. It is structural rather than behavioral; it is about how the organization is built, not about how individuals in it are trained to behave. The structure is what determines whether a particular kind of commitment-shape can survive evaluation.

In the framing I work with, organization design is the structure in which two distinct functions both happen at every level. The first function is management: running the work, reducing uncertainty about what is happening, keeping the work moving against its commitments. The second function is leadership: setting direction, reducing uncertainty about what should be happening, revising the commitments themselves when learning suggests the current direction is wrong. These are not personality traits or job titles; they are functions, and a healthy organization needs both at every level where work is committed.

This framing depends entirely on the structural distinction between goal and task. Management runs against tasks. Leadership sets and revises goals. If the organization's commitment-shape is exclusively task (because the evaluation infrastructure can only assess task-shape), then leadership has nowhere to live in the structure. The leadership function gets squeezed into the cracks: people who do it informally, sponsors who carve out exceptions to the OKR cascade, founder-CEOs who keep direction-setting at the top because no one below them has the structural conditions to do it.

This is the structural meaning of "Machine Age organization." It is not a pejorative; it is a description of an organization whose structure can run work but cannot set direction at the level where the work happens. Such an organization can be very large, very profitable, and very famous. It can use System Age vocabulary fluently. It cannot, structurally, do what the vocabulary describes.

Goal-shape commitments require infrastructure that Machine Age organizations do not have: revision authority distributed to where the direction-setting work happens, evaluation that can assess trend rather than only attainment, mid-cycle direction-checks that treat new information as input to revising direction rather than as deviation from schedule, performance review processes that respect trend evidence and learning alongside numbers. These are not soft requirements. They are structural conditions. Without them, goal-shape commitments cannot survive contact with the organization's evaluation infrastructure, and they will be converted to tasks at quarter-end regardless of how they were set or what vocabulary surrounds them.

This is why "fix your OKRs" engagements typically fail. The OKR is the symptom of a structural absence; you cannot fix the symptom by improving the artifact. The structural absence is that the organization has no infrastructure for goal-evaluation, because the underlying ontology (Machine Age) has no slot for goal-shape distinct from task-shape. The intervention that produces durable change is the structural one: build the goal-evaluation infrastructure for the subsystems that need direction-setting.

This is also why most "agile transformations" and "OKR rollouts" produce the same disappointing result. They import System Age vocabulary into Machine Age structures and expect the vocabulary to do the structural work. The vocabulary cannot do the work. The structure has to be redesigned, which is a much larger and more deliberate project than adopting a new framework, and which most organizations are not in a position to commission.

A closing question

If you are reading this and you have an OKR cycle running in your organization, here is a useful question to sit with. When the cycle ends, will the evaluation tell you whether the work moved the organization in the directions the Objectives named, or will it only tell you which Key Results got hit? The honest answer, in almost every modern organization, is the second.

This is not a complaint about OKRs as a framework. It is an observation about what kinds of evaluation a Machine Age organization is structurally capable of performing. If the evaluation can only assess task-shape, then no amount of System Age vocabulary at the Objective layer will produce System Age results. The vocabulary will be decoration; the work will be tasks; the dashboards will be green; the organization will continue to wonder why its strategy is not moving.

The transition that would change this is not a vocabulary upgrade. It is a structural redesign of the evaluation infrastructure for the subsystems where direction-setting is needed. That redesign is harder than a framework rollout and more durable than a vocabulary refresh. It is also the actual System Age transition for that organization, the one Drucker pointed at in 1954, that Ackoff named in 1974, and that the management field has spoken about for half a century without operationalizing.

Most organizations have not made it. Recognizing this gives you the right thing to work on.