4 ms·

Models show what you want them to show.

by the-dude 5y ago

cinntaile 5y ago
What is it that you are implying here?
- the-dude 5y ago
  That models are inherently biased. A model does not contain or produce 'truth'. A model will produce what one is comfortable with.
  - loopz 5y ago
    While getting to truth may be very helpful, most models just simulate observed facts and effects, to see if it is useful and can be validated. The point is not to predict, but to see how much variance different models can display, what one might learn from it and to evolve new models from that. What is comfortable or not shouldn't be a factor. In fact most climate scientists would search for any and all redeeming factors. It is concerning how insignificant they all seem to be. Major contrary findings would make great news, get prizes and grants. Timelines are indeed less useful and accurate, than levels of temperature and descriptions of accumulated and latent effects. Prediction is too hard and a sidetrack from all that is established peer-reviewed science so far.
  - cinntaile 5y ago
    It still feels like I have to read between the lines. What are you trying to say when you claim that (climate) models produce what you are comfortable with?
    - the-dude 5y ago
      When models don't produce what one is comfortable with, it gets tweaked until it does.
      - cinntaile 5y ago
        Do you think this is the case here? That the models have been tweaked for 50+ years to claim something that is not based on what actually happens?
        the-dude 5y ago
        > Do you think this is the case here? No reason to believe these models are different ( it is not about the models, it is about the people building them ). > that is not based on what actually happens? That is not what I said. These models are used to say things about the future, so by definition they are an assumption about what actually happens. I will try to explain my point by a fictitious example : Researcher A is building model for global temperature supposedly correlated with compound X. Model shows that within 100 yrs the temperature will plateua and then drop off to pre-industrial levels. The model will be considered faulty and will be tweaked until it does feel comfortable : that is, aligns with one's assumptions. Models are crystalizations of one's assumptions, there is no truth.
        cinntaile 5y ago
        Of course you use the intuition of experts within that field when building models, that's a given. I don't see what is wrong with that, that's how all models are built. But what is it that you're trying to say about climate models? State your own assumptions please, I don't want to keep playing this guessing game.

Models show what you want them to show.

cinntaile 5y ago

What is it that you are implying here?

the-dude 5y ago
That models are inherently biased. A model does not contain or produce 'truth'. A model will produce what one is comfortable with.
- loopz 5y ago
  While getting to truth may be very helpful, most models just simulate observed facts and effects, to see if it is useful and can be validated. The point is not to predict, but to see how much variance different models can display, what one might learn from it and to evolve new models from that. What is comfortable or not shouldn't be a factor. In fact most climate scientists would search for any and all redeeming factors. It is concerning how insignificant they all seem to be. Major contrary findings would make great news, get prizes and grants. Timelines are indeed less useful and accurate, than levels of temperature and descriptions of accumulated and latent effects. Prediction is too hard and a sidetrack from all that is established peer-reviewed science so far.
- cinntaile 5y ago
  It still feels like I have to read between the lines. What are you trying to say when you claim that (climate) models produce what you are comfortable with?
  - the-dude 5y ago
    When models don't produce what one is comfortable with, it gets tweaked until it does.
    - cinntaile 5y ago
      Do you think this is the case here? That the models have been tweaked for 50+ years to claim something that is not based on what actually happens?
      - the-dude 5y ago
        > Do you think this is the case here? No reason to believe these models are different ( it is not about the models, it is about the people building them ). > that is not based on what actually happens? That is not what I said. These models are used to say things about the future, so by definition they are an assumption about what actually happens. I will try to explain my point by a fictitious example : Researcher A is building model for global temperature supposedly correlated with compound X. Model shows that within 100 yrs the temperature will plateua and then drop off to pre-industrial levels. The model will be considered faulty and will be tweaked until it does feel comfortable : that is, aligns with one's assumptions. Models are crystalizations of one's assumptions, there is no truth.
        cinntaile 5y ago
        Of course you use the intuition of experts within that field when building models, that's a given. I don't see what is wrong with that, that's how all models are built. But what is it that you're trying to say about climate models? State your own assumptions please, I don't want to keep playing this guessing game.