Indeed. When you're model goes from predicting THE WORLD WILL END to predicting "no biggie, equivalent to a minor flu season" within a week, it means the model is garbage and we stop paying attention to your predictions
That said people are taking the wrong lesson
I'm seeing many people go "See! Now the model predicts a small # of deaths! We clearly overreacted!"
Wrong
The lesson is "these models are garbage and we can't rely on them in EITHER direction."
I'm seeing many people go "See! Now the model predicts a small # of deaths! We clearly overreacted!"
Wrong
The lesson is "these models are garbage and we can't rely on them in EITHER direction."
Then we're back to the original position
Which is: enormous uncertainty about transmissibility, lethality, etc
Some guess that huge %s of the population have already been infected and the virus is less lethal than people think
Great! Hope for the best, prepare for the worst.
Which is: enormous uncertainty about transmissibility, lethality, etc
Some guess that huge %s of the population have already been infected and the virus is less lethal than people think
Great! Hope for the best, prepare for the worst.
I really hope that the naysayers are right, and that the virus has infected huge numbers of people already, that the current fatality rates are wildly overstated, etc.
But that hasn't been *demonstrated* yet
It could be with widespread antibody testing
Until then, precaution
But that hasn't been *demonstrated* yet
It could be with widespread antibody testing
Until then, precaution
You'll notice that I've almost *never* discussed the Imperial model, or used it as an argument for drastic action
If you were around in 2007-08, you should be inoculated against relying on complicated predictive models—especially when model error can have massive consequences
If you were around in 2007-08, you should be inoculated against relying on complicated predictive models—especially when model error can have massive consequences
