Computer models are so woefully inadequate in predicting weather patterns, simply because they have not taken into consideration the far greater number of variables that may exist and are as yet insufficiently studied, and the interactions of those variables that may deflect a whole other set of hypotheses and assumptions.
You could come up with a huge hairy-butted equation that nobody could solve, even with the quickest mathematical computer in the universe. Certainly not with the level of technology we now have at our disposal.
Comes from overthinking the problem.
Models used for predicting weather are reasonably accurate and have improved greatly in the past decade, mostly because they have to be accurate. If you tell me it’s going to snow tomorrow, I’m going to plan my day accordingly. I’ll not forget you were wrong when it’s a sunny day, and I’m going to start watching a different weather guy on a different channel.
Climate models don’t have the same incentive for accuracy, and thus can be made to say what you want them too without the same immediate public feedback.