Our current best prediction is that 596 million people will get infected (247M-1.4B 50% CI) before the end of the year and that only 16 million (8.0M-36M 50% CI) will be reported to WHO. We expect also 1.74 million deaths (532k-6.0M 50% CI) world wide before the end of 2020. Overall infection fatality rate is likely to be 0.8% (0.5-1.2 50% CI).
And, I agree with you - we should have had massive projects for track and tracing already in February, but politicians almost everywhere dropped a ball on this besides Taiwan, South Korea maybe Germany and some other countries. In many places there are still no good exit plans.
Also, scientific community is slow to do anything. Masks are good example of that - making a masks at home costs nothing, so if widespread use of homemade masks have a chance to save lives and reduce length of lock downs even by days then the risk-cost-benefit analysis is just overwhelming in favor of widespread mask use. But medical community wants controlled trials and tip-top evidence before trying anything even if it's as simple as a mask.
Overall there seems to be a sort of fog of war going on.
From my perspective the models never had any credibility due to the fact that small errors in model parameters lead to exponentially big errors in predictions. Also, soon after China managed to contain the outbreak it became more or less obvious that human behavior is the biggest unknown in all epidemiological models. No model is able to predict how politicians and society will react.
Here is my comment from 3rd of March [2]: > I think very important part of a model that intends to give real world prediction would be ability to model what happened in China. There needs to be something in the model that allows to slowdown the growth of infection.
Thankfully, big political decisions were never only based on models. They were based on 1. empirical data from SARS outbreaks in China, later on based on China and Italy experiences 2. simple common sense computations like around 20-80% of population can get it if we do nothing (Swine Flu or Spanish Flu) and 0.5-2% of infected can die based on Grand Princess and Chinese data. So the best case scenario assuming no reaction and wide spread like with Swine Flu or Spanish Flu is 0.1% of population dieing and the worst case scenario is 1.6% population dieing. It was obvious that reaction was needed given the severity of the best case scenario with no reaction.
I think what got missed is the fact the people across the world were already starting to react by themselves in early March soon after Lombardy went into lock down. You can see that in Google/Apple Mobility data and in Open Table restaurant reservations. People miss that piece of data even now as lock downs are lifted.
Regarding 50% - this is what shows up on CT scans. The paper does not provide much of interpretation of that scans. But seems like we can indeed expect reduced lung capacity at least soon after recovery in some nontrivial percent of the population. That much we know, the long terms effects are just a guess. I will probably try to make a question to quantify the actual risks, but operationalizating it so that resolution is clear is difficult.
[0] https://pandemic.metaculus.com/COVID-19/
[1] https://www.metaculus.com/questions/247/pandemic-series-a-ma...
[2] https://www.metaculus.com/accounts/profile/103304/#comment-2...