I went to Glacier National Park a while ago and stood before a sign that said "The glaciers you see will be gone by year 20xx" or something like that. The year was 20xx and I was looking right at the glaciers and they were decidedly not gone. I understand the motivation. The reality is that an action of strength A is required to create outcome X, but if one expresses this need, it's likely one achieves only an action of strength A- < A and consequently the outcome X- != X. This encourages us to say "we need an action of strength A+ or we will get the outcome X-". The hope is that the discount that people apply on the prediction results in an action of strength A' ~ A and an outcome X' ~= X.

But public exhortation is a repeated game and unless you're Heaven's Gate the success of your popularized predictions has a significant effect on the discount that the public applies to your predictions. Failed predictions have an effect on the discount rate asymmetric to the direction. It is likely that a successful public authority allows some number of people to experience catastrophe pour encourager les autres. Attempting a zero error rate by biasing to zero catastrophe is likely to cause repeated discounting until true catastrophe cannot be averted.

I'm reminded me of the short story The Wandering Earth. Spoilers to follow so don't read if you care. In the story, a group of scientists determine that the Sun is going to flare enough to kill everyone, so all of humanity mobilizes to install massive engines to move the Earth far enough to survive while creating technology that survives the dark. After years of traveling to past Pluto, no Sun catastrophe materializes, resulting in a massive Earthwide revolt that eventually takes control of the control center and the scientists and government officials are sentenced to death by walking into the cold waste. Just as the last of them dies, the Sun flares as predicted, just a little later.

In a democracy, credibility is an important resource. To avert actual existential catastrophe, it is better to err on the side of local under-predicted disaster. It is probably best to let Rancho Palos Verdes fall into the ocean without compensating the homeowners. It is probably best to let the Texan flood plains flood and people to lose their homes entirely. It is probably best to let the Californians lose their homes to the wildfires.