A team receives a score of three.
No one is quite sure what the three means, but the slide looks authoritative. Level one was reactive. Level two was managed. Level three is "integrated." Levels four and five sound increasingly impressive. The organization begins planning how to move upward.
Before the next workshop, someone asks a less satisfying question.
What exactly matured?
Human capability maturity assessment is a useful methodological idea only if that question stays at the center. There is no universal validated instrument for measuring the maturity of human capability across the domains of Inner Technology. The term should therefore be used for a family of staged organizational assessment approaches, not for a psychometric scale of how mature people are.
The closest established lineage comes from capability maturity models, especially the People Capability Maturity Model developed at Carnegie Mellon University's Software Engineering Institute. The P-CMM applies a maturity framework to organizational practices for managing and developing a workforce. It is not a theory of individual human maturity.
That boundary is non-negotiable.
What maturity models try to do
Maturity models organize development into structured levels or stages so an organization can compare current practice with a more systematic target state.
The attraction is obvious. Complex systems are difficult to improve when practices are inconsistent and dependencies are invisible. A maturity framework can create a common language for asking whether important practices are improvised, repeatable, institutionalized, measured, and continuously improved.
The People CMM uses this logic for workforce practices. Its 2009 second edition describes a process maturity framework intended to help organizations improve their ability to manage and develop workforce capability. The associated assessment method was designed as a diagnostic for identifying strengths and weaknesses in current practices and prioritizing improvement.
This is materially different from rating employees as immature or mature.
The unit of assessment is organizational practice capability.
A maturity level is a model, not a natural fact
The seduction of a maturity model is that levels look like reality rather than an analytical device.
But the level structure is constructed. Designers decide which practices belong together, what order is expected, what evidence counts, and which threshold separates one level from the next.
Those decisions require theory and validation.
Research on software capability maturity models illustrates the point. Jung and Goldenson examined the internal consistency of key process areas in the Software CMM using hundreds of assessments. Their work did not assume that the maturity structure was self-validating because it had five levels. They tested dimensions underlying the construct and found evidence that required a more nuanced interpretation.
A maturity model therefore has measurement obligations like any other structured assessment.
Level labels are hypotheses about development.
Applying the idea to human capability
For the Institute, a human capability maturity assessment could be useful if it assesses the maturity of the systems that cultivate and protect capacities rather than claiming to score the maturity of the human being.
Consider attention. An immature organizational approach might treat attention failures as individual weakness while allowing constant interruption. A more developed approach could establish norms for focus, task design, interruption, recovery, and tool use, then evaluate whether those practices are consistently enacted and reviewed.
For ethical discernment, assessment might examine whether high-stakes decisions have defined review points, contestability, accountability, stakeholder input, and permission to slow down. For learning, it might examine whether people receive practice, feedback, challenge, and opportunities for transfer rather than only content.
The maturity object is the practice system.
This prevents the framework from becoming a disguised ranking of character.
A defensible assessment process
A maturity assessment should begin with a validated or at least explicitly theorized domain model.
First, define the construct and unit of analysis. Is the assessment about workforce-development practices, AI oversight practices, learning architecture, or another bounded system?
Second, define maturity dimensions independently of the level names. A single level may hide different patterns. One organization might have strong governance and weak learning; another the reverse.
Third, specify observable evidence for each practice. Policies alone are weak evidence. Interviews alone are vulnerable to impression management. Stronger assessment triangulates documents, workflow evidence, observations, outcome data, and stakeholder experience where appropriate.
Fourth, define rating rules. Assessors need anchors clear enough that independent reviewers can reach reasonably consistent judgments.
Fifth, document uncertainty and disagreement. A maturity score should never be more precise than the evidence supporting it.
Sixth, separate diagnosis from certification. Internal learning creates different incentives from external ranking, procurement, or public claims.
Seventh, repeat assessment only when change could reasonably have occurred and preserve the evidence trail so apparent progress can be audited.
Reliability before prestige
Organizations like maturity levels because they simplify communication.
That simplicity can conceal low reliability.
If two trained assessors examining the same evidence produce very different ratings, the level is not stable enough to carry high-stakes decisions. If a level changes depending on who is interviewed, the sampling method matters. If organizations learn to stage documents and performances for assessment, the measure may become vulnerable to gaming.
Reliability is not sufficient for validity, but without adequate reliability there is little basis for interpreting differences.
Validity asks the deeper question: does the model represent the capability it claims to represent, and do higher levels relate to outcomes in the way the theory predicts?
A maturity model may be internally coherent and still fail to predict better judgment, learning, safety, agency, or human outcomes.
The outcome must not be smuggled into the level name.
The danger of developmental moralism
The word maturity carries moral weight.
Applied carelessly to people, it can turn disagreement into deficiency. Someone who resists a change program becomes "low maturity." A team that questions automation becomes "not ready." An employee who needs boundaries becomes "less adaptive."
This is scientifically weak and institutionally dangerous.
Human development is multidimensional. People can show expertise in one context and novice behavior in another. Stress, power, disability, culture, role, opportunity, and domain knowledge affect performance. There is no defensible single ladder on which a whole person can be placed.
The Institute should therefore reserve maturity assessment for bounded practices or organizational systems and use direct measures for specific individual capacities where validated instruments exist.
Do not turn a stage metaphor into a hierarchy of human worth.
Relationship to capacity mapping
Capacity mapping and maturity assessment answer different questions.
Mapping asks: what capacities and related conditions belong in the model, how are they defined, and how do they relate?
Maturity assessment asks: how systematically has a defined practice system developed relative to explicit criteria?
A map can exist without levels. A maturity framework should not be built before the map underneath it is coherent.
This ordering matters. Otherwise organizations create five levels of a construct no one has properly defined.
The same caution applies to the Institute's future Human Readiness Baseline. Baseline, mapping, and maturity are related methods but not synonyms. A baseline describes a current state. A maturity model proposes developmental structure. A map describes the conceptual terrain.
Each carries different evidence requirements.
Maturity models should be tested against consequences
A maturity framework becomes more credible when it can survive a question beyond internal consistency: do the levels matter outside the model?
If organizations rated at a higher level of AI-review maturity are supposed to make fewer unexamined high-stakes deployments, researchers should test that relationship. If a learning maturity model claims that higher levels create better transfer, then transfer should be measured. If a capability-building system claims to strengthen retention, agency, safety, or judgment, those outcomes need independent evidence.
This is criterion validity in practical form.
The relationship may not be linear. A formalized process can initially improve reliability and later become bureaucracy. Organizations can score well on documentation while everyday practice remains weak. A highly adaptive small team may lack formal artifacts that a maturity model rewards. Context therefore has to enter interpretation.
Longitudinal evidence is especially valuable. Cross-sectional comparisons between high- and low-scoring organizations can be distorted by size, resources, leadership, sector, regulation, or selection. Repeated assessment can show whether changes in practices precede changes in relevant outcomes, although causal attribution will still require care.
This is why the Institute should resist releasing a five-level model before the constructs underneath it are stable. A staged framework is visually attractive and operationally convenient. It can also fossilize premature theory.
A better development path is evidence-first: define dimensions, collect descriptive data, test rating anchors, examine inter-rater agreement, study relationships with outcomes, and only then decide whether stages add explanatory value beyond continuous or multidimensional profiles.
The result may be that no single overall maturity score is warranted.
That would not be a failure. It would be the method learning from reality.
Fairness and incentive effects
Any maturity assessment used for funding, procurement, accreditation, or public comparison changes the behavior of the assessed organization.
Once a level becomes a target, people learn to produce evidence for the level. Documents multiply. Language standardizes. Weaknesses become reputational risks. This can improve discipline and also encourage gaming.
Assessment design should therefore separate evidence generation from impression management as far as possible, protect reporting of weakness, and make clear that a lower score in one domain can be more honest and useful than a polished claim of maturity.
Reporting should preserve the profile
Even when a framework ultimately uses levels, reports should preserve dimension-level evidence.
A single overall number can hide asymmetric development. An institution may have sophisticated policy and weak enactment, strong learning systems and weak accountability, or excellent review capacity concentrated in one team. Collapsing these differences into "Level 3" makes the result easier to communicate and less useful to improve.
A strong report therefore shows the evidence behind each domain, confidence in the rating, material disagreements among assessors, and the conditions that could make the result non-generalizable. It also records what was not assessed.
This makes maturity assessment compatible with intellectual honesty. The organization receives a map of where practices are dependable and where they are fragile, rather than a badge that invites comparison beyond what the evidence can support.
Maturity should become harder to fake as the stakes rise
Early levels in a maturity model can often be demonstrated through artifacts: a policy exists, a training has been delivered, a process has been named. Higher claims require stronger evidence. If an organization says discernment is embedded, the question is whether people can challenge a fluent machine recommendation under deadline pressure. If it says accountability is mature, the question is whether responsibility remains traceable after failure. If it says reflection is routine, the evidence should include changed decisions, not only retrospective language.
This suggests a useful design principle for any Institute maturity assessment: evidentiary burden should rise with the maturity claim. Documentation may establish formalization. Repeated behavior across settings may establish use. Outcomes, counterexamples, and the institution's response to breakdown are needed before claiming resilience. Such a model would still be a designed heuristic, not a natural developmental law. But it would make ceremonial maturity harder to achieve. The assessment would reward institutions for becoming more corrigible in reality, not merely more sophisticated in describing themselves.
What the evidence does not show
The existence and use of P-CMM does not validate a generic Human Capability Maturity Assessment for Inner Technology.
P-CMM is a specific organizational workforce-practice model with its own architecture and assessment method. Its concepts cannot simply be renamed and transferred to attention, embodiment, ethical discernment, or relational maturity.
Evidence from software or process maturity models does not show that five levels are universally correct. It does not show that higher maturity scores always cause better outcomes. Internal consistency within one model does not establish external validity for another.
The evidence also does not justify individual maturity rankings derived from organizational frameworks.
A new Institute maturity assessment would require explicit construct definition, evidence models, assessor training, reliability work, validity studies, fairness analysis, and revision before high-stakes use. Until then, it should be presented as a developmental audit framework, not a validated diagnostic instrument.
Maturity as the ability to learn from reality
A useful maturity assessment does not reward an organization for sounding sophisticated.
It asks whether practices have become dependable enough to survive pressure and corrigible enough to change when evidence contradicts them.
The highest level should therefore never mean "finished."
In human systems, maturity includes the capacity to discover that the model itself is wrong.
The team looking at its level-three slide may indeed have made progress. But the most mature question in the room is the one that interrupts the ladder:
What does this score let us know, and what would we need to observe before we are entitled to believe it?
Research references
Curtis, B., Hefley, W. E., & Miller, S. A. (2009). People Capability Maturity Model (P-CMM), Version 2.0, Second Edition. Software Engineering Institute. https://doi.org/10.1184/R1/6581894.v1
Hefley, W. E., & Curtis, B. (1998). People CMM-Based Assessment Method Description. Software Engineering Institute. https://doi.org/10.1184/R1/6581912.v1
Jung, H.-W., & Goldenson, D. R. (2002). The Internal Consistency of Key Process Areas in the Capability Maturity Model for Software. Software Engineering Institute. https://doi.org/10.1184/R1/6585155.v1
Campion, M. A., et al. (2011). Doing Competencies Well: Best Practices in Competency Modeling. Personnel Psychology, 64(1), 225-262. https://doi.org/10.1111/j.1744-6570.2010.01207.x
Shippmann, J. S., et al. (2000). The Practice of Competency Modeling. Personnel Psychology, 53(3), 703-740. https://doi.org/10.1111/j.1744-6570.2000.tb00220.x