Practices Are Local Value Systems: Domains, Policies and the End of 'Best Practices'

Full text

1 Introduction

Most organisations improve their work by borrowing. A team reads about a review routine, a planning cycle or an incident procedure that worked elsewhere and adopts it. The borrowed practice arrives with a reputation: it is a "best practice". Very often it disappoints. The ceremony is followed, the artefacts are produced, and yet the people who receive the team's output see no improvement, or see new friction they did not have before.

This paper explains that pattern with the Theory of Value Management (VMT) and the Æilus methodology built on it. Its claim is simple. A practice is not a neutral technique. It is a small value system that consumes some value elements and produces others, and whether that exchange is good depends on the larger value system it sits in. The same practice can therefore be admissible in one organisation and harmful in another, and no global ranking of practices can be defended.

The paper is a methodology paper. It defines a practice as a template with a declared value hypothesis, sets out when a practice is admissible, describes domains and domain policies as acceptance contexts, uses the partial commensurability axiom (A13) to show why practices cannot be ranked globally, and gives a structure for building and maintaining a practices collection. Definitions of value, anti-value, value flow, value system and value transformer follow the first paper of this series and are not repeated. Roles and processes are left to the eighth paper. The Æ formalization is not re-derived here: see Bolshakov (2026).

1.1 Whose value is measured

Throughout the paper the value system under analysis is a single organisational transformer, such as a team, a department or a service, together with the actors that send value into it and receive value from it. The boundary is drawn around that transformer and its direct inflows and outflows. Value is measured from two positions inside this boundary: the receiving actors, who interpret realized value and anti-value at consumption, and the transformer itself, whose capability and participation conditions a practice can strengthen or erode. The wider organisation enters only as an external system connected by flows (A10). Changing this boundary would change the conclusions about any specific practice (T15), which is itself one of the points of the paper.

2 Why "best practices" transfer badly

2.1 Same practice, different value systems

Under VMT, value exists only inside a value system (A1), and one reality can be interpreted as different value in different systems (A2). It follows that there is no "value in general" (T1). A practice cannot carry its value with it when it moves between organisations, because the value it produced in its origin was an interpretation by specific receivers, under specific constraints, with specific inflows.

Consider what actually travels when a practice is copied. The description travels: the steps, the roles, the artefacts. What does not travel is the value system that made those steps worthwhile: who consumed the outputs, what those consumers needed, which anti-value the practice was suppressing, and which other practices supplied its inputs. The adopting organisation reproduces the form and assumes the effect.

2.2 How the failure shows up

Three failure modes follow directly from the theory.

• Flow inversion. A flow that one system interprets as value can be interpreted as anti-value in another (T2). A detailed approval routine that protects a regulated receiver can be pure delay for a receiver who needed speed and carried no such risk.
• Local gain, system loss. A practice can optimise a fragment of the transformer without improving the whole (T5). A team can produce more estimates, more reports or more releases while the receivers' realized value stays flat or falls.
• Adoption by reputation. When a practice spreads because influential sources recommend it, not because realized value confirmed it, the system scales an interpretation without confirmation. VMT treats this as structurally unstable (T16) and as a source of accumulating anti-value (T17).

None of these failures means the practice was bad at its origin. They mean that "good" was never a property of the practice alone.

3 A practice as a local value system

3.1 Definition

Æilus defines a practice as a formal template of a local value system inside a transformer. When a transformer is an organisational unit and not a single person, it inevitably has an internal value system implemented through how work is done. Æilus describes those ways of working as practices. A practice template specifies:

• incoming elements of value and anti-value;
• how these elements are realized, transformed or stored;
• outgoing elements of value and anti-value;
• the conditions under which the practice is admissible in a value system.

A practice does not describe how to work correctly in the abstract. It describes what value it consumes and what value it produces under stated conditions. This follows from the transformation axiom (A7): a transformer realizes incoming value and produces new planned value. A practice is a reusable description of one such transformation.

3.2 The declared value hypothesis

Because a practice produces planned value, and planned value and realized value usually diverge (A4), every practice is a hypothesis. This paper proposes that each practice entry states that hypothesis explicitly, in a form that can be refuted by observation:

If this transformer consumes these inputs and applies this transformation, then these receivers will realize these value elements, these anti-value elements will decrease, and flow resistance at this point will fall, without violating these domain policies.

The hypothesis names receivers, so it states whose value the practice serves. It names anti-value separately, because anti-value is not the absence of value (A5) and a practice may succeed on value while generating anti-value elsewhere. It names a point of resistance, because the Æilus principles direct interventions at reducing flow resistance (P6) and not at moving it to another actor.

3.3 Position in the value lifecycle

Practices act on different forms of value. Discovery practices turn potential value into planned value by choosing a recipient and a flow. Delivery practices realize and transform planned value. Other practices reduce anti-value and resistance, or build the transformer's capability to deliver value sustainably. Stating which form a practice acts on prevents a common category error: judging a discovery practice by delivery output, or a capability practice by this quarter's throughput.

3.4 Interfaces and composition

Every practice must expose its interfaces: incoming value and anti-value, outgoing value and anti-value, and expected effects on resistance, throughput and participation conditions. Practices may be combined into a process only when the outputs of one connect coherently to the inputs of the next. Two practices can each be admissible and still be incompatible. When they are combined anyway, the mismatch appears as duplication, misalignment and hidden resistance. Much of what organisations experience as "process overhead" can be read as this kind of interface failure.

4 Admissibility

In Æilus a practice is never evaluated as good or bad in the abstract. It is evaluated as admissible or not admissible in a given value system. A practice is admissible if three conditions hold:

1. it satisfies the domain policies of the transformer;
2. it does not violate value interpretation for the receiving actors;
3. it does not lead to violations of participation conditions or of the sustainability of the value system.

The third condition rests on A8 and A9: actors participate only while their participation conditions hold, and a system is sustainable only while they hold for all actors. A practice that raises output by exhausting the people inside the transformer can pass the first two tests and still fail the third. VMT predicts that actors whose conditions are violated leave or change their form of participation (T7).

Admissibility is also consequence-based. Declared intent, perceived usefulness or a better narrative do not establish it. A practice is confirmed only through observable consequences of value realization: lower resistance, less anti-value, better realized outcomes for receivers. Value is realized at consumption (A6), so the evidence must come from the receiving side. Practices that mainly increase planned-value narratives or stakeholder enthusiasm, without improving realized or retrospective value, are treated as possible sources of interpretational distortion and should be reviewed or retired. This applies P2 and P10 of the Æilus principles to practices themselves.

Finally, the methodology distinguishes instrumental outcomes (internal artefacts and intermediate improvements) from terminal outcomes (value realized at points of consumption). A practice that improves instrumental outputs while terminal realization does not move tends to increase accumulation and resistance across the system. This is consistent with P5, throughput over accumulation, and with the observation that accumulated distortion reduces future capability (T6).

5 Domains and policies as acceptance contexts

5.1 Definitions

Value is never delivered in isolation. The same value element can be acceptable or unacceptable depending on the conditions of its delivery. Æilus formalizes these conditions as domains. A domain is a value-interpretation context that imposes requirements on the admissible practices, processes and outcomes of a transformer. A domain policy is the set of requirements that practices must satisfy to remain admissible within that domain.

Domains do not create value directly. Domain policies do not say which value to produce. They say how value may be produced without generating non-admissible anti-value. In the methodology's own short form: domains do not optimize value; they prevent value from becoming non-admissible.

5.2 The typical domains

Æilus does not define a closed list. The relevant domains depend on the value system and on the transformer's participation requirements. The methodology names typical domains:

Compliance is often added where external rules shape acceptance. A new domain is warranted whenever "good versus bad" can no longer be decided without naming a quality context.

5.3 How domains constrain practices

Domains act as a synchronization mechanism between a transformer and its receivers. They answer one question: is the delivered value acceptable under the constraints that matter to the receiver? Without explicit domains, transformers drift toward maximizing output and planned value, while receivers experience instability, risk and hidden anti-value. A typical sequence is familiar: delivery speed rises, planned value rises, realized value does not follow because a domain constraint was violated, and anti-value accumulates.

For practice selection this has a direct consequence. The admissibility of a practice is not a single verdict but a set of verdicts, one per relevant domain. The same practice may be admissible in one domain and not in another. Domain policies therefore function as filters that a practice must pass before its value hypothesis is even worth testing.

6 Partial commensurability and the impossibility of a global ranking

VMT holds that value types can be commensurable, partially commensurable or incommensurable (A13). Speed of delivery, continuity of service, protection from risk and preserved knowledge are not always convertible into one another at a fixed rate. Æilus treats domains as partial value spaces for exactly this reason: each domain carries its own admissibility criteria, because a single scalar score cannot represent conflicts between them.

From A13 the impossibility of a global ranking of practices follows in three steps.

1. A ranking needs one ordering of outcomes. Where outcomes span partially commensurable domains, an ordering exists only under an aggregation policy that states how trade-offs are resolved.
2. Aggregation policies belong to value systems. The methodology assigns them to the system level, where cross-domain trade-offs are decided. Different organisations legitimately hold different aggregation policies, because their receivers and participation conditions differ (A1, A2).
3. Therefore a ranking of practices is always relative to an aggregation policy, and no ranking holds across all value systems. This is a specific case of T13: there is no universal criterion of optimality.

Two practical points follow. First, "best practice" is meaningful only as an abbreviation of "best under this aggregation policy, for these receivers, within these domain policies". When those qualifiers are dropped, the claim cannot be checked. Second, a value system that has no explicit aggregation policy has no stable basis for choosing between practices at all. In that case selection degenerates into local preference, political influence or metric gaming, which is the pattern Goodhart (1975) described for indicators turned into targets.

Partial commensurability does not mean that nothing can be compared. Practices can still be compared within one domain, and they can be compared across domains inside one value system once its aggregation policy is stated. What the axiom rules out is a comparison that claims to hold everywhere.

7 Building and maintaining a practices collection

7.1 What the collection is

Æilus expects practice knowledge to accumulate in a catalogue curated by ÆVRI. The catalogue is not a list of universal solutions. It is a repository of documented local value systems, each with its context of applicability, observed effects and constraints. Adding a practice to the catalogue does not make it admissible anywhere. Admissibility is always decided in the adopting value system, against its domain policies and participation conditions. Selection, adaptation, composition and retirement are the responsibility of the Value Transformer Owner (VTO) of the adopting transformer, who is also accountable for the anti-value the practice produces. The same structure can be used for an organisation's internal collection.

7.2 Catalogue entry fields

The following fields are proposed as the minimum for one entry. They translate the definitions of sections 3 to 5 into a record that can be reviewed.

7.3 Evidence

Evidence in an entry should follow the value lifecycle. Planned value records what the practice was expected to produce. Realized value records what receivers interpreted at consumption. Retrospective value records how that interpretation changed later. An entry with only planned-value evidence is a proposal, not a validated practice. An entry with realized evidence from one value system is a local result and should say so. Evidence that comes only from the transformer's own reports, without validation by the receiving actor, does not meet the integrity requirement of P10.

7.4 Retirement

Practices have a lifecycle of their own. The methodology states that practices are valid only while they remain admissible and produce positive system-level effects. Retirement is therefore a normal event, not an admission of failure. Reasons for retirement include: realized value no longer follows the hypothesis; a domain policy has changed and the practice now violates it; its inputs are no longer supplied because a neighbouring practice changed; or it mainly produces instrumental outputs that accumulate. Flow resistance accumulates historically (T4), so a practice that is kept after it stops working leaves a residue that its replacement will also have to absorb. Recording why a practice was retired, and in which context, is as useful to the collection as recording why it was adopted.

8 Illustrative example

The following example is hypothetical. It is constructed to show the reasoning and does not describe any real organisation or measured result.

Two software teams consider the same practice: a mandatory second review of every change before release. Team X runs a service that many internal receivers depend on continuously. Its receivers interpret an outage as severe anti-value, and its Availability and Continuity policy requires that no single change can stop the service. Team Y builds short-lived internal prototypes. Its receivers are product designers who need to see working ideas quickly and discard most of them.

For Team X the declared hypothesis might read: receivers realize fewer interruptions, the anti-value element "failed release" decreases, resistance at the release step falls, and the Performance policy is not violated because review time stays within an agreed bound. The practice passes the domain filters and its hypothesis is worth testing.

For Team Y the same practice consumes reviewer time and delays feedback to the designers. Its main output, reduced release risk, protects against an anti-value element that Team Y's receivers barely register. Under Team Y's aggregation policy, which gives priority to speed of learning, the practice adds resistance without corresponding realized value. It is not admissible there, although it is the same practice.

Neither team is wrong. The example shows A2 and T2 at the level of a single practice, and shows why the catalogue entry must carry its context of origin and domain profile, not a verdict.

9 Limitations and research agenda

This paper is a conceptual and methodological analysis derived from VMT axioms and from the published Æilus methodology. It contains no empirical data, and none of its claims about practice transfer have been measured. Its contribution is a set of definitions and a structure that make such measurement possible.

Several limitations should be stated. Writing a refutable value hypothesis for every practice has a cost, and the paper does not show that the cost is lower than the anti-value it prevents. Domain lists are open, so two organisations may describe the same constraint under different domains, which complicates comparison across the catalogue. Aggregation policies are often implicit in practice, and making them explicit may itself be contested. Receiver-side evidence can be hard to collect when receivers are many or distant from the transformer.

The research agenda follows from these gaps:

• case studies that apply the catalogue entry structure to real practices and record planned, realized and retrospective value from the receivers' side;
• a shared vocabulary of domain policies, so that entries from different value systems can be compared within a domain;
• methods for stating aggregation policies explicitly, and for detecting when a practice decision rests on an unstated one;
• tests of interface compatibility between commonly combined practices, to locate hidden resistance;
• a retirement record format, and analysis of why practices are retired across different types of value systems.

Validation would require observing whether practices selected by explicit admissibility checks produce more stable realized value for receivers than practices adopted by reputation, in comparable value systems over time.

10 Conclusion

"Best practices" fail to transfer because a practice is a local value system, and its value belonged to the system it came from. Treating a practice as a template of inputs, transformations and outputs with a declared value hypothesis makes this visible and testable. Admissibility, not quality in the abstract, is the right question: a practice is admissible when it satisfies domain policies, respects the receivers' interpretation and preserves participation conditions, as confirmed by observed consequences. Domains turn quality concerns such as security, availability and continuity into explicit acceptance contexts. Partial commensurability (A13) shows that practices can be ranked only under an explicit aggregation policy, never globally. A practices collection built on these ideas records context, evidence and retirement, and leaves each value system to decide what it can admit.

References

• Bolshakov, V. Value, Anti-Value and Value Flows: A Working Vocabulary for Practitioners. ÆVRI, 2026.
• Bolshakov, V. Value Management Theory and Employee-Centric Competitive Advantage: A System-Level Perspective. ÆVRI, 2026.
• Bolshakov, V. Æilus Methodology 2.0: Principles; Domains and Policies; Practices Collection. aeilus.tech, 2026.
• Bolshakov, V. Theory of Value Management (VMT): axioms A1-A13 and theorems T1-T17. aeilus.tech, 2026.
• Goodhart, C. A. E. Problems of Monetary Management: The U.K. Experience. 1975.