4 ms·

The title seems like clickbait, and the last part of the article does explain that the scientists actually know that new cloud models are mostly responsible for

The title seems like clickbait, and the last part of the article does explain that the scientists actually know that new cloud models are mostly responsible for the change. The big question is then of course whether the new cloud model is wrong and the temperature increase therefore as well, or whether the new cloud model is more accurate than previous ones and the temperature increase is a real effect. I found another article that looks at this in a bit more detail: https://news.ucar.edu/132678/ncars-new-climate-model-running-hot https://news.ucar.edu/132678/ncars-new-climate-model-running... The linked paper this is based on has the following as the last sentence: > What scares us is not that the CESM2ECS is wrong (all models are wrong, [Box, 1976]) but that it might be right.

klodolph 7y ago
Or a third, frustrating possibility—the new cloud model is more accurate, but some other inaccuracy in the model used to be balanced out by the inaccurate cloud model.
- Accujack 7y ago
  Or that there are multiple inaccuracies in the model which interact, so finding and correcting each one makes progress but still doesn't improve the model overall.
  - k33n 7y ago
    Software is fun :)
  - cryptica 7y ago
    Having built some models to simulate very basic things myself, I've seen first-hand how the slightest deviation of the model from reality can completely invalidate the entire output of the model. For this reason, I'm skeptical of models that try to simulate very complex systems with a lot of variables and feedback loops. Models that have a large number of variables are particularly difficult to get right because different tiny mistakes in various parts of the model can compound and yield completely arbitrary and contradictory results. Approximating the reality in software is almost impossible.
    - shusson 7y ago
      I believe you are talking about Chaos theory [1] [1] https://en.wikipedia.org/wiki/Chaos_theory https://en.wikipedia.org/wiki/Chaos_theory
    - mistrial9 7y ago
      you are not wrong, but you neglect to include converging aspects from disparate inputs and processes
    - willyt 7y ago
      It's not impossible. Listen to this podcast, it's a discussion about weather forecasting models with the scientists that create them. They discuss model verification for weather models and they touch how this also applies to climate modelling. http://omegataupodcast.net/326-weather-forecasting-at-the-ecmwf/ http://omegataupodcast.net/326-weather-forecasting-at-the-ec...
madaxe_again 7y ago
Cloud modelling is only part of it - advances in computing have allowed the reduction of cell size, allowing more detailed and therefore hopefully more accurate physical simulations.
- JackFr 7y ago
  > advances in computing have allowed the reduction of cell size, allowing more detailed and therefore hopefully more accurate physical simulations. 'Hopefully' is doing a lot of work there. In modeling a highly dimensional dynamic system it's not obvious that a smaller cell size will lead to more accurate predictions.
  - TimTheTinker 7y ago
    ... beyond a certain (perhaps trivial or intuitive) threshold.
    - Igelau 7y ago
      I guess that depends on the nature of the models. If they actually capture the inherent chaos of the system they claim to model, then minor tweaks will have wildly unpredictable side effects.
cat199 7y ago
> The title seems like clickbait Not to diverge, but I'd be extremely wary of editorial bias/spin from this particular publication until November or so.. which is not to say that the topic itself is suspect - as you point out, the big question is if the updated model is actually worse than previous models or if it is better
zyxzevn 7y ago
Shock: Climate science is not settled. Currently studying solar energy transfers via hot solar particles, latest hot topic.

klodolph 7y ago

Or a third, frustrating possibility—the new cloud model is more accurate, but some other inaccuracy in the model used to be balanced out by the inaccurate cloud model.

Accujack 7y ago
Or that there are multiple inaccuracies in the model which interact, so finding and correcting each one makes progress but still doesn't improve the model overall.
- k33n 7y ago
  Software is fun :)
- cryptica 7y ago
  Having built some models to simulate very basic things myself, I've seen first-hand how the slightest deviation of the model from reality can completely invalidate the entire output of the model. For this reason, I'm skeptical of models that try to simulate very complex systems with a lot of variables and feedback loops. Models that have a large number of variables are particularly difficult to get right because different tiny mistakes in various parts of the model can compound and yield completely arbitrary and contradictory results. Approximating the reality in software is almost impossible.
  - shusson 7y ago
    I believe you are talking about Chaos theory [1] [1] https://en.wikipedia.org/wiki/Chaos_theory https://en.wikipedia.org/wiki/Chaos_theory
  - mistrial9 7y ago
    you are not wrong, but you neglect to include converging aspects from disparate inputs and processes
  - willyt 7y ago
    It's not impossible. Listen to this podcast, it's a discussion about weather forecasting models with the scientists that create them. They discuss model verification for weather models and they touch how this also applies to climate modelling. http://omegataupodcast.net/326-weather-forecasting-at-the-ecmwf/ http://omegataupodcast.net/326-weather-forecasting-at-the-ec...

madaxe_again 7y ago

Cloud modelling is only part of it - advances in computing have allowed the reduction of cell size, allowing more detailed and therefore hopefully more accurate physical simulations.

JackFr 7y ago
> advances in computing have allowed the reduction of cell size, allowing more detailed and therefore hopefully more accurate physical simulations. 'Hopefully' is doing a lot of work there. In modeling a highly dimensional dynamic system it's not obvious that a smaller cell size will lead to more accurate predictions.
- TimTheTinker 7y ago
  ... beyond a certain (perhaps trivial or intuitive) threshold.
  - Igelau 7y ago
    I guess that depends on the nature of the models. If they actually capture the inherent chaos of the system they claim to model, then minor tweaks will have wildly unpredictable side effects.

cat199 7y ago

> The title seems like clickbait Not to diverge, but I'd be extremely wary of editorial bias/spin from this particular publication until November or so.. which is not to say that the topic itself is suspect - as you point out, the big question is if the updated model is actually worse than previous models or if it is better

zyxzevn 7y ago

Shock: Climate science is not settled. Currently studying solar energy transfers via hot solar particles, latest hot topic.