Small enough to
name everyone

现在,这里有四个人

Adaptation Lab is a project team inside SEESALT, not a legal entity, and everyone here works on it alongside a degree. What follows is not a roster of credentials. It is a division of the problem: three parts, and who holds each one. A list this short is not modesty. It is the state of the project.

01The argument, and the cases that test it

Zhijun He · Luoxianer Peng

Zhijun He — MA candidate in African Studies, MacMillan Center, Yale University. Previously a member of the World Bank Youth Advisory Group and a research fellow to the UN Office of the Special Representative on Sexual Violence in Conflict. His undergraduate honors thesis on Rwanda's post-genocide institutional transformation is where the ideas on the premises page came from.

Luoxianer Peng — undergraduate in International Studies and Anthropology at Macalester College, working on disaster anthropology and humanitarian governance. Secretary and interpreter with Blue Sky Rescue in Thailand and China, ICS-300 certified. She coordinated cross-regional relief logistics during the 2024 Chiang Mai floods and interpreted between Thai USAR teams, government agencies and Blue Sky command during the 2025 Bangkok earthquake response. Her own IRB-approved study of the Chiang Mai flood measured psychological outcomes using standard diagnostic instruments.

This part holds the theoretical half: the claim that states anticipate attribution and build accordingly, the four settlement traces used to assess it, and the coding rules that the diagnostic will inherit. Attributive Ambiguity is accepted for presentation at the AAG Annual Meeting in New York, February 2027, and both of us will be there. It now has an empirical companion in part 03, which exists to test it rather than to illustrate it.

Everyone else on this page studies response capacity from the outside. Luoxianer has stood inside it — moving supplies during a flood, interpreting between rescue commands during an earthquake.

The paper argues that visible response is what gets built, because it is what gets credited. It matters to have a co-author who has watched that happen at the operational level rather than in a document, and who can say when the argument does not match what the week after a disaster actually looks like.

02How the indicator actually gets made

Xinyue Wen

Doctoral student, School of Social Work, Columbia University. Previously an MA in International Policy and Development at the Middlebury Institute and an MSW at Renmin University. At UNDP's Oslo Governance Centre she tracked SDG 16 indicator gaps across countries and sorted national governance reporting into eight dimensions; at UNEP she followed intergovernmental process at the High-Level Political Forum. At Harvard's Spatial Data Lab she works on urban shrinkage and older-adult mental health in China, using spatial analysis and regression discontinuity in R.

This part holds what is easiest to get wrong from the outside. We claim that composite capacity indices construct a distinction and then remove it in the last step. Anyone can say that. Saying where in the process it happens, and whether the people doing it have any choice, requires having been in the room.

She has assembled a country-level governance measure from inside a UN system — which is the thing this project is arguing against. Knowing how the aggregation is really made is what keeps the critique testable rather than rhetorical, and it is also the fastest route to finding out that we are wrong.

03The interval between record and action

Guqing Yu

MSc student in Environmental Science and Engineering, EPFL. Previously a BSc in Liberal Arts and Sciences at Amsterdam University College, with an exchange semester in mathematics at NUS. Geospatial work on least-cost path modelling of wildlife corridors in Sri Lanka, deep-learning vectorization of historical maps, and a long-term water quality monitoring database.

This part holds the empirical floor, and it is now a paper of its own. The Lag takes the Msimbazi basin in Dar es Salaam, where flood hazard has been on the official record for decades and prevention eventually arrived at very large scale and very late, and measures how much exposure accumulated in between. Guqing is first author and will present it in the 2027 virtual programme.

The choice of case was hers to make and she took the harder one. Nobody in Dar es Salaam was ever able to call the flooding an act of nature, which removes the alibi our companion argument depends on. What remains is a cleaner question: not whether prevention happened, but how long the interval was, and what filled it.

If prevention leaves no observable proxy at all, the diagnostic cannot be built regardless of whether the argument is right. This is our largest technical risk, and it is now a paper designed to fail loudly rather than a caveat on a page.

The imagery half is deliberately conventional — land-cover change in this city has been mapped many times, and we reuse those methods because they work. The new object is the documentary timeline: which corridor entered the official record in which year, on whose authority, at what resolution.

04The parts nobody holds yet

Open, and honestly open

Global health security. The JEE, IHR core capacity assessments, the GHS Index literature. The same attribution logic appears in pandemic preparedness, and that line of argument is currently an outline.
Development finance and procurement. Which capacity indicators the GCF, the Adaptation Fund and the MDBs actually cite in eligibility assessment. This decides how large the gap is, or whether it exists at all.
In-country researchers. Attribution contests happen in local languages and local politics. Remote coding misses the part that matters.

You do not need to be at Yale, or a student anywhere. We recruit from SEESALT's 135-person consultant pool and from outside it on the same terms. If you do none of these but know someone who does, an introduction is the most useful thing you can send.

05What joining actually involves

Said before a first conversation, not after

Co-authorship where the contribution is authorship, an open-source contribution record, and real design authority over the method — including the authority to overturn a premise on this site.
No employer documentation for CPT, OPT or STEM OPT. SEESALT is an unincorporated volunteer initiative and cannot issue it.
No salary, and no budget for conference travel or fieldwork. Costs are absorbed personally. Anything covered so far was covered by one person, out of pocket, for one occasion — that is not a fund and should not be planned around.

We ask elsewhere on this site who benefits from a confusion, and we said we would point that question at ourselves. Money is where it gets sharpest, so: there is none. What has been paid was paid privately, once. Anyone joining should assume nothing follows from it.

Think one of these parts is held by the wrong person?

That is a useful message too. Zhijun He — zhijun.he@yale.edu

How we see the problem: Premises · The protocol at v0.1: Method · The papers: Attributive Ambiguity · The Lag

播种光明，收获永恒