Read on Twitter
Just wanted to make a quick point of agent-based versus more standard epidemiological models. I claimed that agent-based models have a smooth landscape. That is, of course to start. They evolve into something more rugged. https://theconversation.com/the-modelling-behind-melbournes-extended-city-wide-lockdown-is-problematic-145681
The issue is that to properly validate those models you need to see if what they evolve to is something like the environment you are examining. It is near impossible for that to be the case. But it is a critical issue if you want to understand whether the virus might jump from ..
.. one half of the city to the other. The alternative way is to use mobility data (and yes it exists) to model the network of movements in a city. You can then amend the standard epidemiological models to take that into account. I discussed that here https://joshuagans.substack.com/p/cities-and-covid-response
... but the paragraph didn't make it into the article on Melbourne. Now Victoria may not have that data. That means you have to consider more than one model to inform on your reopening. That didn't seem to be done which worries me.
