Hypothesis, not conclusion: “In the US, switching to EVs would save lives and be worth billions”

With a confidence interval between zero and infinity:

A team led by Northwestern’s Daniel Peters decided to have a particularly detailed look at this issue, examining several scenarios of grid generation and EV adoption in the US. The results show that even with today’s grid, switching to EVs produces significant benefits.

The researchers used simulated hourly air pollution data from vehicles around the country, along with emissions data for power plants. This went into a model of weather over the course of a year (2014, as it happens), which also simulated important chemical reactions and natural emissions of compounds that interact with pollutants. The resulting air quality simulations were applied to an EPA population health model to show the expected impact on human health.

Source: In the US, switching to EVs would save lives and be worth billions | Ars Technica

And this was pushed through climate models afterwards.

No matter how you slice it, when your model is based on assumptions, simulated values, multiple models, all applied on top of one another, you have created an interesting video game simulation.

Perhaps you can use it to produce multiple hypotheses. But one thing you cannot do – in any way, shape or form – is produce a useful forecast of anything. Claiming this pile of models produces definitive conclusions is scientific fraud.

After dealing with confidence intervals infinitely wide, they conclude EVs will save 100 to 370 lives per year – in a country where almost 3 million people die every year. Commenters to the article saw right through this scientific bullshit.

This is an hypothesis generator. Sadly, the easy availability of computing and modeling has turned much of science into fake science, no longer based on observations and data. This model’s output is an hypothesis – turning it into a conclusion is scientific malpractice.