3 ms·

You know what actually needs an investment? Climate models which are completely shit right now. It's actually preposterous that there are trillions of dollars w

You know what actually needs an investment? Climate models which are completely shit right now. It's actually preposterous that there are trillions of dollars worth of decisions in ESG and climate credits and government incentives while the science behind this is supported by old fortran code that has terrible quality and can't give reasonable predictions. We have trillions of dollars chasing predictions of code barely worth a million dollar investment. That's absurd, it's a recipe for disaster. Instead of useless government beaurocrats shuffling hypothetical CO2, should invest in climate models that aren't patched together by grad students in universities and old as hell code and a few NASA engineers maintaining an untested unreadable and unverifiable monstrosities. Maybe there's no crisis at all. Maybe it's all doom and gloom, but the certainty that everyone's pretending that science gives is just not reflective of the reality of how climate models look right now. They are absolutely shit and I don't blame them given the resources they have, they did their best, but it's absolutely insane how much money is moving around on predictions coming out of that mess.

tito 3y ago
Say more about this! Let's say you had $100B to build better climate models over the next 5 years. Where would you spend it? My general sense is the models are good enough to tell us that a) things are headed in the wrong direction and b) we're really far off from a solution. Generally in agreement with you but I haven't researched deeply about what's needed on more modeling and how to prioritize.
- machina_ex_deus 3y ago
  You need at least two completely independent teams who don't see each other's code so you'll have a rough idea how much developer bias is responsible for the signal. Right now they are all copying code from each other, a single error in these and all the climate models are effected. Replication here should mean different scientists writing different code arriving at the same estimates. It needs to be open source, and it needs to be way more organized, no physical constants which aren't constant or which have bad precision or which have different values in different parts of the code. Actual error modeling built in or at least a way to measure and quantify the accuracy. Some models only use CO2 input for radiation and not for many other parameters that depend on CO2. Because the code is a mess and constants that shouldn't be constants are all over the place. And getting "constants" right is even more important than having high resolution to the simulation. Getting real physical data to replace all the ad hoc constants is extremely important. And validating the correctness of the cloud modeling with real world data is also important. And you can't write huge spaghetti code, it needs to be readable and comprehensible to someone other than original programmer so that people can validate the assumptions.
  - tito 3y ago
    I want this to happen, how can I help?
    - machina_ex_deus 3y ago
      Look up IPCC's CMIP6 project. I think you need an expert team of software engineers, physicists and climate experts that will either build good climate models or decipher the mess that is the current models variability. Honestly, looking at the code of some models I don't think there's much hope in understanding them to the point where you know where the deviations are coming from, it's probably easier putting together new models that are subject to much more strict testing, unit testing, evaluations against real world data than salvaging the current mess. And just throw out the models which fail the testing.

tito 3y ago

Say more about this! Let's say you had $100B to build better climate models over the next 5 years. Where would you spend it? My general sense is the models are good enough to tell us that a) things are headed in the wrong direction and b) we're really far off from a solution. Generally in agreement with you but I haven't researched deeply about what's needed on more modeling and how to prioritize.

machina_ex_deus 3y ago
You need at least two completely independent teams who don't see each other's code so you'll have a rough idea how much developer bias is responsible for the signal. Right now they are all copying code from each other, a single error in these and all the climate models are effected. Replication here should mean different scientists writing different code arriving at the same estimates. It needs to be open source, and it needs to be way more organized, no physical constants which aren't constant or which have bad precision or which have different values in different parts of the code. Actual error modeling built in or at least a way to measure and quantify the accuracy. Some models only use CO2 input for radiation and not for many other parameters that depend on CO2. Because the code is a mess and constants that shouldn't be constants are all over the place. And getting "constants" right is even more important than having high resolution to the simulation. Getting real physical data to replace all the ad hoc constants is extremely important. And validating the correctness of the cloud modeling with real world data is also important. And you can't write huge spaghetti code, it needs to be readable and comprehensible to someone other than original programmer so that people can validate the assumptions.
- tito 3y ago
  I want this to happen, how can I help?
  - machina_ex_deus 3y ago
    Look up IPCC's CMIP6 project. I think you need an expert team of software engineers, physicists and climate experts that will either build good climate models or decipher the mess that is the current models variability. Honestly, looking at the code of some models I don't think there's much hope in understanding them to the point where you know where the deviations are coming from, it's probably easier putting together new models that are subject to much more strict testing, unit testing, evaluations against real world data than salvaging the current mess. And just throw out the models which fail the testing.