How accurate are models of complex systems?
Consider that if highly accurate models were really possible someone could make one that would accurately predict NFL game outcomes. When that happens we will know it because Sports Books would no longer take bets on NFL games. Sports Books profit because they get a slight statistical advantage which pays off due to bettor volume.
So we know they can't make an accurately predicting model for football. How accurate can they possibly get with climate change modeling? It's orders of magnitude more difficult. Football is formally rule bounded and tons of data on a long history of games is readily available. In contrast, the vast majority of climate data is based on theoretical methods. This isn't bad, it's just far less precise because the variance of all those theory based estimators accumulates across the model, widening the confidence interval of its outputs to a point where it may not be useful.
Finally, climate scientists are people. They have careers, families, financial responsibilities. What happens if they say the climate is fine? Some other crisis gets the funding and their careers are destroyed. They can't just switch to a different science, they've invested a lot of time and money becoming a specific thing.
I found this article very wordy and skimmed parts, but it looks to me like a goalpost moving exercise to explain why their aggressive predictions for the present time didn't come true.