I told myself after reporting on the West African #Ebola outbreak to stay away as much as possible from stories about models in an outbreak. Now I've written a story about #covid19 models (with my colleague and editor @martinenserink). Quick thread. sciencemag.org/news/2020/03/m…
@martinenserink First of all what’s the problem with models?
For science journalists:
1. Most of us are more familiar with reading (and writing) about cell cultures than computer models. We tend to ignore or gloss over the assumptions that go into a model (if we even understand them).
For science journalists:
1. Most of us are more familiar with reading (and writing) about cell cultures than computer models. We tend to ignore or gloss over the assumptions that go into a model (if we even understand them).
@martinenserink 2. Models appeal to the worst instincts in journalists. We end up presenting the most extreme scenario often neglecting the uncertainty around it or the spread of possible outcomes. One of the #Ebola models I wrote about was this CDC projection: 
@martinenserink Below are some of the headlines. One of the terrible side effects of throwing around such projections is that it dwarfs the current tragedy and makes thousands of human lives seem like small change: 550,000 cases? More than a million? 1,4 million? Take your pick. Same difference.
@martinenserink To be clear that CDC model also included a best-case scenario, in which #Ebola would be contained and there would be far fewer cases (ultimately what happened). But that did not make it into many stories. And take-away was duh: without containment it will get bad.
@martinenserink Given all this, why am I writing about models? Because they have played a hugely important role in #covid19 pandemic. Entire cities and countries have been locked down based on hastily done forecasts (often not peer reviewed). “It’s a huge responsibility,” as @cmyeaton told us.
@martinenserink @cmyeaton Given my experience with how we journalists respond to models you do have to wonder whether politicians will be any better. Most academics "sort of pivot into a new world when an emergency hits”, says @cmyeaton. She suggests a "National Infectious Disease Forecasting Center”.
@martinenserink @cmyeaton One example: The UK going from “take it on the chin” via some talk about herd immunity (amid loud opposition from @devisridhar @richardhorton1 and others) to basically doing what most European countries are doing was clearly influenced by the models politicians were getting.
@martinenserink @cmyeaton @devisridhar @richardhorton1 The reason that models are so important at the moment is that we have so little experience with #SARSCoV2 and no drugs or vaccines. But precisely because the virus and the disease are so new, models also have more uncertainty baked into them than say flu models.
@martinenserink @cmyeaton @devisridhar @richardhorton1 For instance: Early plans on mitigating #covid19 in UK were based on demand for ICU beds being similar to flu (what models are based on). When data from Italy suggested more beds are needed, it showed that mitigation was not feasible without overwhelming NHS.
@martinenserink @cmyeaton @devisridhar @richardhorton1 The UK debate was partly about what a middle path between locking down the entire country and letting the virus burn through the population could look like. Given all the uncertainty about the new virus charting such a path based on these models is a big gamble.
@martinenserink @cmyeaton @devisridhar @richardhorton1 Or as @BillHanage put it to me: “It’s like, you’ve decided you’ve got to ride a tiger, except you don’t know where the tiger is, how big it is, or how many tigers there actually are.”
