In Minnesota, there's only been one day since the start date of the protests (May 26) that have had more infections. Infections are very much down since the protests. Whereas in Florida, May 30 (earliest day on this

chart) saw 870 infections and now there's been a day with nearly 10,000 infections and probably (I don't really feel like doing the math) a 2-week span with 5,000 infections per day. Also, Florida currently has a higher % of the population that has been infected than Minnesota. I don't think any place besides NYC could have an infection rate that would be even close to 10% of the population (adjusted via antibody surveys). So we're talking single digit percentages of the population that has been infected pretty much anywhere. Even a state that has twice the immunity than another state (like 2% vs 4%) is not that much in actuality to slow the spread down significantly just due to immunity. How is Minnesota's infection rate lower after than protests while Florida's rate is over 5x its pre-protest rate if the protests have caused the spike? Because it's really hard to spread the virus outside and the protests or any outdoor activities are very low risk for spreading the virus. The increase in infections is from the indoor congregating of public groups of people without masks.

On May 26th, Minnesota was coming down quickly from a major wave, and that downward trend was promptly stopped in its tracks a week later. That is as significant a change in trajectory as a slow rise turning up into a spike.

And the difference in those small percentages is huge, if you operate under my understanding. So like, where is my 10% coming from. I originally made that estimate based on the Diamond Princess. A bunch of people were trapped on a ship with a brand new virus that can spread asymptomatically and 20% of the ship was infected. If you treat that as a test tube and say roughly everyone on that ship was exposed to the virus, 19.2% of the people on board got infected. I round up to 20%. (Brief aside, I'm making assumptions, I know I'm making assumptions, but making predictions is always about finding the best assumptions and then checking them against reality.) So, imagine for a moment this pans out, and 80% of the population is naturally immune to the virus, that 80% of people can be exposed without being infected. What does that mean? It would mean people are much more easily exposed to the virus but most aren't infected.

Normal assumptions of pandemic spread and heard immunity center around an R value, that is meant to describe how many people an average carrier will pass the infection onto. Herd immunity is then calculated as the necessary percentage of acquired immunity required to make it so that 1 or fewer of those people are susceptible. If each person passes it to multiple people, the pandemic grows. If fewer, the pandemic shrinks. That's calculated as 1-1/R0, where R0 is the R value if nobody has been exposed. Early R0 estimates were around 2.5, which calculates to a necessary 60% infection to reach the point where R=1 based on herd immunity. But that's based on the idea that nobody has been exposed to this virus so everyone is susceptible. One of the "1"s in that equation is the current R value, it just happens to be 1 at the point of herd immunity stopping exponential spread. So the broader equation is %immune = 1 - R/R0. If we make a faux R0* for the theoretical world where everyone is susceptible, call that 2.5 R at the beginning of the pandemic, and calculate based on 80% immunity, we get 80% = 1 - 2.5/R0*, therefore R0* = 12.5. Put that back into the equation as R0 to find herd immunity, 1-1/12.5 = 92%. Pull out the 80%, you're left with 12% of people getting actually infected (and then I round off to 10 because these obviously aren't precise figures, and 12% specifically communicates an inappropriate degree of precision; 10 is also just a really easy number to visualize).

That was all based on assumptions from an early test case of sorts. Let's check that against reality. Well, while other places are currently increasing in cases, New York's trend is staying down. The R value isn't going back above 1. What do serology tests suggest about New York? The city in places got as high as 20%, which would be equivalent to the Diamond Princess, and the state has about a 12% infection rate. We look at Sweden, where it's not right to say they did nothing, but they decided to do medium social distancing measures and hold the course until immunity kicks in. Well, the number of cases did come back down as expected. Based on regular assumptions, they expected like a 50% infection rate to have caused that trend, but then serology tests were coming back with single digit infection rates (and people kind of freaked out). If you go with my assumptions instead, that single digit number makes perfect sense, you don't expect more than 10% infection until the end of the pandemic. Like, it checks out pretty well. Not even I expected my back of the envelope math to pan out this accurately.

So circling way back to the 2%-4% comparison. If I'm accurate, the difference between 2% infection and 4% infection is a 20% decrease in how much people spread to each other. That is a big difference, and can easily be the difference between a state going from R < 1 to R = ~1 like Minnesota corresponding to Florida going from R = ~1 to R > 1 with cases growing exponentially as any R great than 1 does. If those simultaneous flexes correspond to the same timeline of activity, and that activity has largely ended, we should see an unnatural looking return to Florida only slowly increasing and Minnesota should go back to dropping slowly over the next week or two.