3 ms·
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 overa
by 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 agoSoftware is fun :)
- cryptica 7y agoHaving 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 agoI 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 agoyou are not wrong, but you neglect to include converging aspects from disparate inputs and processes
- willyt 7y agoIt'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...