Thread/ 1 One of the risks of game theoretic models of policy dilemmas is translation back to real world. Firstly, model set-ups claim to be 'simplification' of a target system but this assumes we know the system's basic form already. Catch-22... https://twitter.com/GernotWagner/status/1273347886932205574

2/ This model of #climate negotiations with #geoengineering it is claimed 'simplifies to the basic form' of climate diplomacy. But how climate diplomacy works is precisely what is in question in this new simulation by @GernotWagner and @adrien_fabre . https://www.nature.com/articles/s41599-020-0492-6#Fn1

3/ Second, assuming this leap of faith (abstraction=simplification of actual real world) exacerbates already substantial risk of slippage IMO, from 'this model world works like this' to 'the target real world works like this'. I think this is an example of such slippage

4/ without providing evidence - historical or contemporary - why is it 'likely' these conditions apply? The paper does not say.

Models can be useful, but 'cherry-picking by modelling' is also a danger: curate an artificial model system, assume the model represents 'basic form',

5/ emphasise a particular model-result, claim 'likely' in real world.Not accusing of dishonesty - models are very seductive. But target worlds cannot be assumed to have 'basic form' of the curated model, and model results initially demonstrate the model world, not the real one

6/ With geoengineering already often conceptualised, studied & evaluated as abstracted devices outside society & world politics, inserting that into abstracted hyper-stylised models of world politics is asking for trouble, IMO. See also Contraption Fallacy http://ceassessment.org/guest-post-olaf-corry-open-university-climate-engineering-and-the-contraption-fallacy/

7/ The standard Game Theory modeller response is a tactical epistemic retreat: oh but it's just a model outcome, absolutely not a claim about reality. See responses here for an example: https://twitter.com/GernotWagner/status/1273347886932205574?s=20

8/ Models are useful. The problem is what Paul Pfleiderer calls 'Chameleon models' that shape-shift from 'hey look guys, I'm policy relevant' to 'oh but I'm just a model', when challenged: https://www.gsb.stanford.edu/faculty-research/working-papers/chameleons-misuse-theoretical-models-finance-economics

8/ Models have a strange alure, especially where uncertainty of the future and complexity reign. But they are fraught with danger when (inevitable) translation to real world happens (either by authors, peers, policy-makers). Esp, in a field with highest stakes & vested interests