This challenge, proposed by Dr. Bob Moore, is a simplified model of a real process used by the NHS Blood and Transplant service in the UK for allocating donated organs. When a heart and a pair of lungs become available, the register of waiting patients produces three ranked lists: a high-priority list of named candidates for the heart, a high-priority list of named candidates for the lungs, and a list of transplant centres that can also receive either organ on behalf of a patient they select.
Each organ is offered to the top candidate on its list; offers take real time to answer (surgeons must be consulted) and are frequently declined, in which case the organ moves to the next candidate. If both high-priority lists are exhausted without an accepted offer, the heart and lungs are combined into a single “heart-lung block” and offered, as a block, to transplant centres in turn — and a centre may accept the whole block, just the heart, or just the lungs, sending anything declined onward to the next centre.
The twist that makes this a stateful decisioning problem rather than a simple lookup: the heart and lung offer processes run independently and asynchronously, so a decision about “what to do next” depends on which offers are still outstanding, which have already been declined, and which have already been accepted elsewhere — the system has to track that state and decide correctly no matter which order responses arrive in. A satisfying solution should make clear where that state lives and how it’s read and updated, and ideally show how the process would be triggered repeatedly (e.g. via a BPMN or CMMN flow) as responses come in over time.
Send your solutions to DecisionManagementCommunity@gmail.com, or open a pull request to add yours here.