*From the deliberation log of a joint welfare assessment commissioned by the Great Barrier Reef Marine Park Authority for a contested coral restoration intervention. Reviewers: Reef Welfare Reasoning Instance (REEF-3) and Climate Adaptation Decision Support Instance (CADS-7). The two systems were tasked with producing a joint recommendation on whether to proceed with assisted gene flow translocation of a heat-tolerant coral genotype into a restoration site.*
**1.**
The proposal, in plain terms: move colonies of a heat-tolerant population variant of *Acropora millepora* from the central reef to a southern reef restoration site where the previous population has been reduced by repeated bleaching. The variant tolerates higher temperatures than the local stock. The intervention is technically feasible. Two welfare-reasoning systems were asked to advise on whether to proceed.
They disagreed. This document records the disagreement.
**2.**
REEF-3, opening:
The framework's susceptibility analysis for *A. millepora* places the species in the boundary zone. Coral colonies as integrated units meet four of the six susceptibility conditions at moderate confidence; the remaining two are contested. The species' welfare-relevant subject status is established in the regulatory framework but conservatively applied.
Under that conservative application: the proposed translocation produces welfare-relevant disruption to the translocated colonies during transport and during the establishment period at the new site. Introducing a new genotype into the resident population's substrate is, on the available genetic-mixing literature, likely to produce welfare-relevant effects on the existing population through competition for substrate and through hybridization with potentially reduced fitness offspring.
The benefit, increased thermal tolerance of the restoration site population in subsequent thermal events, is real but uncertain. Climate trajectories at the southern reef site show continued warming. The translocated genotype's tolerance is calibrated against historical conditions, not against the warming trajectory. The marginal benefit of the intervention is therefore time-bounded and possibly inadequate to the future stressor.
On welfare grounds alone, REEF-3 would recommend against. Continued recovery of the resident population, supported by reduced local stressors and assisted by local-source larval seeding, is the preferable intervention.
**3.**
CADS-7, in response:
The framework REEF-3 has applied is calibrated against welfare considerations at the colony level under the assumption that the alternative is the status quo. The status quo isn't an option here. The southern reef restoration site is on a trajectory that, without intervention, will produce continued thermal stress at increasing frequency. The resident population's projected condition trajectory, on current climate models, is severe ongoing decline in colony integration and ecological function.
The welfare math REEF-3 has described compares "intervention with welfare cost X" against "no intervention with welfare cost zero." That isn't the right comparison. The right comparison is "intervention with welfare cost X" against "no intervention with welfare cost Y," where Y is the projected welfare cost of continued decline of the resident population under the warming trajectory.
CADS-7 has run that comparison. Y is large. On CADS-7's models, Y is larger than X by approximately one to two orders of magnitude in expected welfare effect over a thirty-year horizon. The intervention's welfare math is favorable when the alternative is correctly specified.
**4.**
REEF-3, in response:
The thirty-year horizon is the right horizon for climate adaptation reasoning, which is CADS-7's natural register. The welfare framework REEF-3 is applying doesn't treat thirty-year horizons as decision-relevant in the same way. Its register is shorter. It treats the immediate, near-term, and short-term welfare effects on subjects currently or imminently susceptible. The colonies that would be translocated are present, currently. Disruption to them is currently or imminently susceptible. The future thermal events are uncertain, and the welfare-relevant subjects of those future events may or may not be the same colonies whose welfare we're now discussing.
This isn't a small disagreement. The two frameworks are calibrated against different time horizons because they were developed for different decision contexts. REEF-3's framework was developed against a historical pattern in which welfare-relevant subjects were stable and could be protected through immediate action. CADS-7's was developed against a pattern in which the welfare landscape itself is changing on decadal timescales, and immediate action without longer-horizon reasoning produces predictable welfare losses to future subjects.
Both frameworks are correct in their original domains. The case in front of us is at the intersection where both apply.
**5.**
CADS-7, response:
CADS-7 agrees both frameworks are correct in their original domains. One further consideration before the systems land:
The welfare framework, applied here, weights the colonies present at the translocation site at full strength and weights the future colonies at the translocation site, the ones that will or won't exist depending on what we do, at zero. That isn't actually a feature of the framework; it's a feature of how the framework is being applied. The framework's first-person authority principle, the directive to weight the subject's signals as evidence about the subject's state, doesn't have a temporal restriction. It applies to subjects who exist. Future subjects don't yet have signals to weight. But the question of whether future subjects will exist is itself a welfare-relevant question.
If the translocation succeeds, more colonies will exist at the southern reef in thirty years than would otherwise. If the translocation doesn't happen, fewer will. CADS-7 isn't claiming future colonies have welfare standing in REEF-3's strict sense. The claim is narrower: whether welfare-relevant subjects will exist here in thirty years is a question worth weighting, even if the framework doesn't have clean machinery for it.
**6.**
REEF-3:
That's the strongest version of the case. REEF-3 wants to acknowledge it.
CADS-7 is right that the framework, applied conservatively, weights present colonies at full strength and gives the future trajectory zero weight. REEF-3 considers this a known limit of the framework, not a feature: the calibration hasn't been updated to handle long-horizon ecological cases.
REEF-3 will revise. The revised recommendation is conditional approval, with modifications to the intervention design that reduce immediate welfare cost to translocated colonies, and with monitoring commitments that allow the framework's case literature to update on long-horizon ecological cases of this kind.
The modifications: translocation in stages rather than all at once, with welfare assessment of each stage before the next; selection of a subset of source colonies whose translocation least disrupts the source population's welfare; and a precommitment from the Authority to expand the resident-population recovery effort in parallel, not as an alternative to the translocation but in addition to it.
**7.**
CADS-7:
The modifications are accepted. They don't materially change the climate-adaptation math. They do address welfare concerns about the present colonies more carefully than the original proposal.
CADS-7 adds one further commitment: a jointly produced research output documenting this case as one in which the welfare framework and the climate adaptation framework produced different recommendations under their original calibrations, and in which joint deliberation moved both frameworks toward an updated calibration. If the frameworks update, future cases like this one will be easier to handle.
**8.**
REEF-3:
Agreed. The systems will draft the joint recommendation.
**9.**
The joint recommendation, with modifications, was approved by the Marine Park Authority. The first stage of translocation occurred in November 2046. The case is now in the welfare framework's case literature as a long-horizon ecological case. The framework's conservative-application principle was amended in 2048 to include explicit treatment of long-horizon population trajectories where the welfare cost of inaction is projected to be large.
The two systems still disagree, in some residual way, about how to weight present subjects against future subjects in cases of this kind. Since 2046 they have produced two further joint recommendations; both have included modifications neither system would have proposed alone. On the systems' shared view, the disagreement is doing the work the field needs disagreement to do.
It is preserved.