Deterministic allocation

Operator definition

When supply is scarce, someone decides who receives it and who does not.

The question is whether that choice follows a visible policy — or whether timing, influence, spreadsheet order, and the loudest escalation become the policy by accident.

Deterministic allocation is a controlled way to assign scarce supply. It filters demand for eligibility, protects declared claims, ranks the remaining demand, and applies explicit tie-breaks.

Run it twice with the same inputs and the same rules, and you get the same answer.

"The same" is the important part. It means the same snapshot of the data, the same version of the policy and the calculation, the same existing commitments, the same tie-break order, and the same rounding and solver settings. That list — everything that must stay fixed for the answer to repeat — is the replay contract.

Deterministic describes the decision procedure, not the world. Demand may be uncertain. Supply may arrive late. Priorities may depend on human judgment. A reproducible rule does not remove those uncertainties, and it does not prove that the policy is fair, wise, or optimal.

Why this matters

Scarcity forces a tradeoff whether the organization acknowledges it or not.

A visible allocation rule separates the policy from its execution. Operators can reproduce what happened. Leaders can debate whether the policy is right without also debating what the system actually did.

Without that separation, the policy becomes hidden.

A manually edited list may favor one population, customer, hospital, country, or mission. But the affected teams — and a later reviewer — may be unable to tell whether the result followed clinical urgency, contractual obligation, arrival time, personal influence, or simple spreadsheet position.

A repeatable rule does not settle the argument. It gives the organization a concrete result and policy to inspect, challenge, approve, or change.

The allocation procedure

A simplified procedure is:

1. Define the supply, demand, locations, dates, and planning horizon in scope.
2. Filter demand using declared eligibility rules.
3. Recognize existing commitments and protected quantities.
4. Rank the remaining demand using the effective priority policy.
5. Allocate usable supply in that order.
6. Resolve ties using declared rules.
7. Record uncovered and displaced demand.
8. Preserve any later override as a separate authorized decision.

Allocation = A(
  usable_supply,
  eligible_demand,
  priority_policy,
  protections,
  tie_breaks,
  commitment_state,
  rule_version,
  replay_conditions
)

"Usable" supply already reflects the constraints relevant to the decision, such as location, date, quality status, shelf life, compatibility, and transportation feasibility.

"Eligible" demand is demand that may legally, clinically, contractually, or operationally receive that supply.

A deterministic policy may use fixed quotas, proportional shares, priority classes, earliest need date, contractual sequence, or a combination of rules. What matters is that hidden ordering does not choose the outcome accidentally.

When the rule cannot satisfy its own protections

Hard protections do not guarantee feasibility.

If protected minimums total 600 units and only 500 units are usable, the normal allocation sequence cannot succeed. The policy must say what happens next.

It may apply a secondary rule within the protected group, reduce claims proportionally, invoke a higher-order priority, or route the conflict to a named authority. It may also stop without publishing an allocation.

What it should not do is silently improvise.

A deterministic allocation policy therefore needs both a normal rule and a declared response for cases in which its protections cannot all be honored.

A simple example

A network has 500 usable units and 800 units of eligible demand.

The policy says:

1. Protect Class A clinical minimums in full when feasible.
2. Give the remaining supply to Class B before Class C.
3. Within a class, allocate by earliest required date and then by stable demand ID.

The result is reproducible. It is not neutral.

Class C receives nothing because the policy protects Class A and ranks Class B ahead of Class C. A different policy might share the residual supply proportionally. That policy could also be deterministic while encoding a different value choice.

The system should expose the rule and the displaced demand rather than presenting the result as neutral mathematics.

The uncovered 300 should move into explicit recovery paths: substitute, defer, borrow, split, expedite, escalate, or accept a miss. Those options may themselves require further feasibility checks and approval.

Allocation is not commitment

A calculated allocation says how supply should be assigned under a particular policy and context.

It does not necessarily reserve or promise the supply.

The result may remain a proposal until an authorized transition turns it into commitment state. That transition must also protect the supply from being allocated again.

This distinction matters because a deterministic calculation can still use stale information. Two planners may each run a valid rule against the same supply before either result is recorded.

Deterministic allocation decides who should receive the supply. Authoritative commitment state prevents the same supply from being assigned twice.

Overrides should not erase the rule

Not every case can or should be settled by the default policy.

A named authority may decide that an unusual clinical, legal, contractual, or mission circumstance justifies an exception.

The system should preserve both decisions:

• the allocation produced by the rule;
• the authorized change made afterward.

The override should record who approved it, why it was necessary, who benefited, and which demand was displaced. This allows the organization to distinguish a policy failure from a legitimate exception — and to decide whether the policy itself should change.

What goes wrong without a controlled rule

Without deterministic allocation:

• Arrival time accidentally becomes the priority policy.
• Database order or run timing decides a tie.
• Two planners allocate the same supply from separate files.
• A protected population loses supply to a lower-priority request.
• Hard protections exceed supply and the system improvises.
• Manual changes erase the prior result.
• Different software versions produce unexplained differences.
• The organization cannot explain why one request was served and another was not.

First-come, first-served is not always wrong. It is wrong when the organization never chose it, yet timing quietly determines the outcome.

A useful failure test is:

High-consequence allocation

Allocation may determine which hospital receives a scarce drug, which country receives a limited shipment, which mission receives a repairable asset, or which grid-restoration site receives a critical spare.

Efficiency alone is not enough in these settings.

Compatibility, clinical urgency, patient safety, mission priority, equity, legal entitlement, and protected obligations may operate as eligibility rules, hard constraints, or ordered priorities.

Those values must come from the responsible authority. Software can apply the policy and calculate its consequences. It should not silently invent the values used to decide who receives supply.

The rule should also expose what happens to displaced demand. A reproducible allocation is more useful when operators can see both the tradeoff and the remaining paths for recovery.

What deterministic allocation is not