Upthread there’s a climatologist openly admitting to lying to make their point seem more dire. The whole field has a credibility problem. I don’t blame laypeople for their skepticism.

Models are constantly tweaked to make them more accurate. How do you do that without looking at output and then going back to tweak it?

I will treat this as a genuine question. This is a complex topic. You could start by reading "The Art and Science of Climate Model Tuning".

https://journals.ametsoc.org/view/journals/bams/98/3/bams-d-...

You add more physical parameters. Climate models are not parameter fitting from the past data, they are physical models. Of course you test them with old data too, but they are mainly for verification and controlling for variables that are still missing.

They test these models with other planets too. Set in parameters for Mars or Venus and they precinct pretty well how those climates behave.

One way is to verify them. Run multiple variations on data from the 1950's and see how they compare to measured values up to today. Throw away the models that don't fit the measured values. Then you have some confidence that the remaining variations may be representative when run into the future.

This is also done in weather forecasting, though maybe they run the model in reverse, I can't recall. Same idea though, verify with existing data before predicting.

Yes this happens, but it is very controversial. To quote from the article I linked above:

> The question of whether the twentieth-century warming should be considered a target of model development or an emergent property is polarizing the climate modeling community, with 35% of modelers stating that twentieth-century warming was rated very important to decisive, whereas 30% would not consider it at all during development. Some view the temperature record as an independent evaluation dataset not to be used, while others view it as a valuable observational constraint on the model development. Likewise, opinions diverge as to which measures, either forcing or ECS, are legitimate means for improving the model match to observed warming. The question of developing toward the twentieth-century warming therefore is an area of vigorous debate within the community.

Not going to lie but it sounds like there is significant disagreement among scientists on evaluating climate models?

And do I read it right that 35% of scientists deem it important that a model indicates future warming? Science doesn’t work that way, you don’t get to choose a conclusion then model for it.

Yes, there is (and continues to be) significant disagreement among scientists on evaluating climate models. This is normal and has always been the case from the very beginning of climate models.

No, you read it wrong. 35% thinks it appropriate to fit to past warming. Of course nobody is suggesting to fit to future warming! But past warming is observed.

If you are familiar with training/test split in machine learning, some thinks past warming is in training set (and test with things like reproducing ENSO), and some thinks past warming is in test set. I think it boils down to disagreement over how determined was twentieth-century warming.