Within weeks of the rock-wall collapse that cascaded into a debris flood down the Bhotekoshi-Trishuli on 26 August, the Ministry of Agriculture, Forests and Environment did something that deserves notice. It took stock. Nepal's Position for COP31, now close to final, names the kind of hazard that came down the valley, treats it as one of a class of cascading risks that cross district and national boundaries, and says clearly that the losses which matter most to a family – a place, a shrine, a livelihood tied to a mountain slope – are the ones assessments do not capture. For a ministry that in the same weeks was preparing Nepal's appeal to the Fund for responding to Loss and Damage, that is quick learning, and the right thing to carry to Antalya. What Antalya can give in return is limited by what Antalya is. Negotiations run on sympathy, evidence and strategy. None of these reaches a household in Rasuwa and the districts downstream. Justice, if it is served at all, is served at home, in the decisions that settle where people may rebuild, what they are owed and who decides. Those decisions will be governed by another document from the same Ministry. The draft National Climate Change Policy 2083 was published for comment on 22 September, four weeks after the disaster, with ten days to respond. It runs to 15 policy areas, 55 strategies and some 280 working policies. The Ministry reports that it consulted in every province and has promised a revised draft once the comments have been vetted. Governments change; ministries carry on; a draft begun under one administration is finished under the next. That is how institutions keep their memory, and much of what the last climate policy began, seven years ago, deserves to be carried forward. An overhaul does not mean a blank page; it means selecting, with a view of where the country is going, what to keep, what to drop and what to add. This draft exists because the 2019 policy provided for a review every five years; the calendar called for this revision, not the disaster and not the election ballot. Its description of the problem would have served seven years ago. It concedes that the targets set then were not met, but does not say which, or why, or what it will now do differently. Its answer to the question 'what changes?' is six new funds, five new platforms, a new research centre and a new civil-service group. A government that has said what it wants to be known for – transparency, accountability, a state that uses its own scientists and innovators – deserves a climate policy that shows it. This Policy does not yet. Two aspects of this Policy draft has fallen behind the country's COP31 Position. The first is the category. The Position recognizes ice-rock avalanches and cascading hazards as a class of risk in their own right. The draft Policy has no such category: its risk framework is a list of hazard types, and what came down the Bhotekoshi in August is not on it. Budgets, monitoring, early warning and insurance are all organised by category. A hazard that has no category is not monitored, not warned against and not covered; it is missed, as this disaster was missed by the National Adaptation Plan. The revised Policy should organize risk by what an event severs: roads, water, schooling, health posts, the practices that hold a community together. It should allow a new class of hazard to be recognised by impact, without waiting for the next five-year review, and widen monitoring from the glacial lakes to the unstable terrain above them. The second is the loss. The Position says that non-economic loss goes uncounted by conventional assessment; the draft offers it only as psychosocial services, a response to a loss that has not been assessed. The revision should direct a method for assessing loss of place, ritual and burial sites, practice and livelihoods tied to terrain, used alongside the economic assessment, and require every assessment to state what it could not capture. Neither the COP31 Position nor the draft Policy says how resettlement and reconstruction will be carried out. Two months on and a month before winter, about a thousand people are still in holding centres. In Nuwakot, families are moving back into flood-damaged houses even though the National Disaster Risk Reduction and Management Authority's rule is that no one should live below the flood mark. They move back because the rent assistance approved on 11 September has not reached them. Some 3,000 parcels of land have been identified in Rasuwa and Nuwakot, but not assessed for suitability. Eligibility for transitional shelter, and the process for land allocation and tenure, are described by the United Nations as unclear. Khalte, in Rasuwa, shows where that leads. At least 250 of its 300 households had moved there after the 2015 earthquake. A geo-hazard assessment in 2018 found the site at risk; most residents, asked, chose to stay; a government report flagged it again last year; officials were preparing to move people when 26 August took every house. One resident has now moved three times in eleven years. Nepal's COP31 Position speaks of community-centered recovery. Strategy 26 of the draft Policy has two working policies. A displaced household has six questions that those two working policies cannot answer. Where may I rebuild: a published site standard, and a no-build category under the Land Use Act 2076, drawn on the mapped extent of what happened. On what terms: land for exchange and pooling, dorest and grazing rights carried over, and a livelihood designed into the site, because people walk back to fields that were not provided. Can I go back: a right to remain and a right to return, with a published timeline, because temporary becomes permanent by default. Who decides, and on what evidence: independent scientific advice sought before the siting decision and published with it. Do I know enough about the risk to my family to choose: the hazard assessment of the old site and the new one, given to the household in a language and a form it can be understood before it is asked to decide, because a right to science begins around the kitchen hearth. Will the record follow us: the survey of my house, my land, my losses and the decisions taken about my household, kept by a national custodian, not lost when a project closes, and open to me, so that the next assessment starts from what the last one found. The disaster authority will lead reconstruction under a bill currently being drafted; another bill would oblige the state to resettle landless households from risky land. Standards written into those bills while they are bills will bind. Dialogue Earth asked last week how many times Nepal will resettle its disaster-affected. So far the answer has been given by events' speeches, not by policy. The revised Policy can give an answer – once – by writing down the six questions above and answering them before the next site is chosen. The Ministry has shown, in its COP31 Position, that it can learn in weeks. What Nepal says in Antalya should already be written in Kathmandu. Truong Huyen-Chi (Shachi) is Director of the Asia Coordination Office of the Asian Mountain Academic Alliance. On 2 October the Alliance submitted 39 recommendations and 139 clause-by-clause suggestions on the draft Policy to the Ministry.