Upgraded Decision Science for Humans and AI: Applying the deVinery Grammar of Desire

The deVinery Grammar of Desire offers decision science a structural upgrade by identifying and operationalising the upstream Desire Signatures that shape the very pressures Decision Science aims to solve. The grammar identifies the core tensions every decision-making body lives inside as three locked paradoxes rather than as pressure points that can be balanced or optimised away. 

Because the paradoxes cannot be solved, every situation produces only a limited set of real moves. Once that finite set is legible, the characteristic pattern of how a given body navigates it — its desire signature — can be read, diagnosed, and worked with.

The practical result is tighter diagnosis, a constrained option set that no longer invents solutions the geometry cannot support, visible costs so high-damage versions can be refused, and the ability to turn the same instrument onto the adoption problem itself. Good analysis stops dying in implementation for mysterious reasons; the pattern blocking it becomes mappable. The upgrade does not replace existing decision tools. It constrains and orients them so that more capacity is kept alive and less gratuitous harm is produced. Further, because the paradoxes are upstream from current systems thinking and psychology, it applies equally to mapping, predicting and diagnosing AI decisions, as well as human ones.

Decision science as it is now

Decision science helps people and organisations choose under uncertainty. In practice it usually does some version of this:

- Clarify the problem and who is deciding.

- Gather what people want and what the constraints are.

- Generate options.

- Model likely consequences (forecasts, scenarios, costs, risks).

- Try to correct for bias and group process failures.

- Choose, act, measure, adjust.

At its best this is careful and evidence-based. It already knows that people are not purely rational, that organisations have politics, and that incentives matter. It tries to reduce harm and improve outcomes.

Modern decision science is a toolkit for choosing under uncertainty and competing objectives. In practice it usually runs some version of this sequence:

1. Define the problem and the decision-maker(s).

2. Elicit or model preferences / utilities / criteria.

3. Generate options.

4. Model consequences (forecasts, scenarios, simulations, multi-criteria analysis, cost–benefit, risk matrices).

5. Surface biases and process failures (behavioural decision research, groupthink checks, red-teaming).

6. Choose, implement, monitor, adjust.

At its best this is rigorous. It has absorbed systems thinking, complexity, incentives, stakeholder mapping, and psychological safety. It knows that humans are not rational agents and that organisations have politics. It still treats the core tensions as solvable or optimisable pressure points: more information, better incentives, better process, better alignment, and the right trade-off can be found or approximated.

Sacrifice, when it appears, is usually treated as a cost to be minimised or a failure of process. The implicit hope remains that with enough skill we can keep the current identity of the decision-making body and get the better outcome.

Its recurring limit is practical: even excellent analysis often fails to shift the biggest patterns. Institutions keep failing to course-correct on known high-stakes risks. New programmes appear; the underlying loop continues. Something in the method is still missing.

What the deVinery Grammar of Desire adds

Every decision-making body — a person, a team, a board, an agency, a government — is trying to do three things at once that cannot all be fully satisfied at the same time:

1. Keep its current form and identity intact (its methods, status, story about itself, what it already knows how to do).

2. Stay in workable relationship with the others who matter (stakeholders, partners, rivals, the people affected by the decision).

3. Stay coherent with the larger constraints it cannot escape (law, markets, physics, demographics, public legitimacy, the actual limits of the situation).

These three pressures are always present. Current decision science treats them as competing factors that can be balanced or optimised with enough skill — so that, ideally, no hard loss is required. The Grammar treats them as a locked set that cannot be fully solved. Something always gives. The practical question is not whether a cost will fall somewhere; it is where it is already falling, whether that cost is making the whole more capable or less capable, and which of the still-available moves would produce less unnecessary damage and more lasting capacity.

What changes in the working process when we apply the deVinery Grammar of Desire?

Diagnosis 

Alongside the usual models and preference data, we also ask plain questions:

- What is this decision body most invested in protecting right now (its current form, status, method, story)?

- Who or what is currently carrying the real cost of that protection?

- Is that arrangement making the whole system more able to act and adapt, or is it slowly hollowing it out?

- Which options on the table still pretend that no real change to the current form is required?

These questions do not replace forecasting or stakeholder analysis. They constrain them. They stop the process from treating “keep everything important the same and get the better outcome” as a realistic design goal when the situation has already made that impossible.

The set of real options shrinks  

Instead of generating an open-ended list of new programmes, the process works with the moves that are still actually available under the present pressures. Options that require a necessary transformation not to happen are set aside as structurally unreal, even if they look attractive on a spreadsheet. That reduces wasted effort and the later cynicism that comes from programmes that were never going to hold.

Costs become visible so they can be reduced

Making the existing cost visible is the opposite of being indifferent to harm. When the cost stays hidden, it still lands — usually on the people or capacities with the least power to refuse it, or on the future. Naming where it is landing lets the decision body refuse the high-damage versions and prefer the versions that keep more of the whole alive and capable.

A practical way to find a better orientation from inside  

When the stated goal is followed one or two layers deeper — “If we got what we say we want, what would that actually give us? And what would that give us?” — a clearer orientation often appears. The decision body can then re-examine the original situation from that clearer orientation. The same limited set of moves then produces different results because the priorities have shifted. This is not an external moral lecture. It is an internal method for locating what the effort is actually for.

What this produces in practice

- Fewer decisions that look rigorous while slowly damaging the systems and people they claim to serve.

- Earlier sight of patterns that will produce large failures later, while there is still room to change course.

- Less energy spent on options that were never structurally available.

- More decisions that expand real capacity instead of defending a form that is already costing too much.

- A way for care to land accurately instead of being spent on process that cannot resolve the underlying bind.

This is how the method reduces suffering rather than increasing it. It does not claim that hard choices disappear. It claims that many of the most damaging outcomes come from refusing to see the bind clearly, and that seeing it clearly is what makes better navigations possible.

The deeper claim — that these three pressures are locked invariants at the level of how any self exists at all, and that desire itself belongs at that same level — is what makes the finite move-set and the clean distinction between generative and damaging patterns hold across domains. A decision scientist does not have to start there. They can start with the observable change in diagnosis, option set, and cost visibility. The category claim is why the practical upgrade does not collapse back into “it depends” once the workshop ends.

---

Applying the same method to the adoption problem itself

Most decision processes treat “getting the plan adopted” as a later communication or change-management problem. Analysis produces a recommendation; then the organisation fails to implement it; then people invent explanations (resistance, politics, lack of buy-in, poor messaging). The Grammar lets us treat the adoption failure as its own live situation and run the same diagnosis on it.

The body that is supposed to adopt the plan is still under the same three simultaneous pressures:

- Protect its current form, status, methods, and story about itself.

- Stay in workable relationship with the people and groups who matter.

- Stay coherent with constraints it cannot escape.

A recommendation that is analytically correct can still be structurally unavailable if implementing it would require a transformation the current form is not yet willing or able to make. When that is the case, the non-adoption is not mysterious. It is a predictable navigation under those pressures. The protect move of the current form reasserts; the plan is delayed, diluted, or quietly killed while everyone continues to speak as if implementation is still the goal.

Because the method can be turned directly onto that instance, the practical questions become:

- What is this decision body most invested in protecting right now that the recommended plan would disturb?

- Who or what is currently carrying the cost of not adopting it?

- Is the present arrangement (plan on paper, no real change) expanding or contracting real capacity?

- Which moves are still actually available for this body in relation to this specific recommendation?

- If we follow the stated desire for “successful implementation” one layer deeper — what would actual adoption give them, and what would that give them — does a clearer orientation appear that makes a workable navigation visible?

This turns adoption from a vague “resistance to change” problem into a mappable situation with a finite set of remaining moves. Intervention can then target the actual geometry that is blocking implementation instead of adding more persuasion, more process, or more incentives that leave the underlying bind untouched.

In short: the same instrument that improves the quality of the decision also supplies a way to diagnose and work with the specific pattern that stops good decisions from being lived. That is a distinct practical gain for decision science, not only for the content of plans but for the conditions under which plans become real.

AI decisions can also be tracked: Evidence that the deVinery Grammar of Desire sits upstream of Decision Science

Because the Grammar is structural rather than psychological or cultural, the same three paradoxes shape any decision-making body — a person, a team, an organisation, an institution, a nation, or an AI system. Once the finite set of moves is legible, the pattern of navigation (the desire signature) can in principle be read in silicon the same way it is read in flesh. That scope is itself evidence that the instrument sits upstream of decision science as it is currently practised, which still largely assumes a human psychological or organisational substrate. The Grammar does not yet claim a fully operational AI implementation; it claims that the geometry does not change when the decision-making body is artificial. The same diagnostic and predictive questions therefore remain available wherever relational decisions must be made under real constraints, human or machine.

Comparing Decision Science and the deVinery Grammar of Desire

—

	Current decision science (at its best)	With the deVinery Grammar of Desire
Core tensions	Competing factors / tension points to balance, harmonise, or prioritise	Three locked simultaneous paradoxes that cannot be solved, only navigated
Option set	Open-ended; new programmes continually invented	Finite set of moves still actually available under the paradoxes
Why good plans fail	Resistance, politics, messaging, incentives	Mappable navigation of the decision body itself; can be diagnosed and worked with directly
Cost / transformation	Treated as optional or as process failure	Made visible so high-damage versions can be refused and lower-damage, higher-capacity versions preferred
Prediction	Forecasts of external outcomes	Constrained predictions of the decision body’s own next navigations and their likely shape
Orientation	Stated goals and revealed preferences	Stated goals plus a practical method (following desire one or more layers deeper) for locating a clearer orientation from inside
Scope	Usually the problem and the options	The problem, the options, and the live geometry of the body that must adopt and live the decision

How to Implement the deVinery Grammar of Desire to Decision Science

The deVinery Grammar of Desire and the production of deVinery Desire Signatures are the copyright work and IP of Artemisia de Vine, founder of the deVinery Institute. Reading the public core documents establishes the claim and the reasoning. Implementing the instrument in real decision environments — policy, corporate, AI, community, or other relational systems — requires training and licensing through the Institute.

Contact the deVinery Institute / Artemisia de Vine for implementation, training, consultation, licensing and collaboration terms.