The talk by Spiegehalter really underlined the relationship between uncertainty and predictive futures. In regard to his articles on COVID statistics, I enjoyed his discussion on how he limited his own word count to try and make sure he was not relying on unnecessary jargon, and delivering clear and digestible insights to the public.
I think this is very insightful for two reasons. First, these advanced predictive systems do deal with much unknown, and there is much politicization of the reliability of these predictions. I recall initial comments made by the Bush administration that refuted the reliability of climate change predictions because they weren't exactly right. The compressibility and 'straightforward' nature of these shorter articles generates the risk that they can be easily refuted.
The second area is in relation to the fact COVID and other health issues are wildly unpredictable, and requires constant updates. A long term model doesn't encapsulate the moving data components. This raises questions about whether our current mechanisms for tracking health epidemics are adequate.