Every COVID-19 Model Is A Waste Of Time

You don’t need to model a tsunami. Just run

Don’t try to understand this graph, will make your head explode

I read a lot of South Korean news, to see what a functioning country is like. One thing I’ve noticed is that they never talk about COVID-19 models. They never talk about projections.

The closest I could find were comments like this:

“This is an acute emergency of an infectious disease that we know little about — it’s best advised to play it safe.” — Shin Hyoung-shik, director of the National Medical Center’s Center for Infectious Diseases, told The Korea Herald.

In Western media, however, it’s all about projections and models, hypothetical deaths, and cases, trying to optimize disaster. These models are endlessly debated until reality catches up and obliterates them. Then they run the models again. It’s a complete waste of time.

You don’t need to model a tsunami. You’re not going to surf the damn thing, just get out of the way.

The Best Model Is Experience

There’s a joke in my country. Someone called the President and said a tsunami was coming. She sent someone to the airport with a sign, looking for a Japanese dignitary named T. SUNAMI.

Before the 2004 Indian Ocean tsunami, we didn’t know what a tsunami was. After the thing killed 30,000 of us, however, we understood. If you say tsunami now people will just run. Nobody’s going to ask for a model, or if they can finish their pina colada. We’ll just run.

Nobody had to tell Koreans, Taiwanese or Vietnamese what an epidemic was. They had experience with SARS and MERS. They clearly saw what happened in China. They didn't model or debate something which is fundamentally unpredictable. They just moved their asses. And, knock on wood, they won.

The Worst Model Is Any Model

Every model is an explanation. If you’re taking time to explain stuff, you’re already dead. The worst model is any model, because a model isn’t action.

In the tsunami example, it took that wave an hour to wrap around our island. At any point, someone could have picked up the phone and called the west coast, but it would have taken 30 minutes to explain WTF a tsunami was, by which time the wave was already there.

The fact is that when faced with a novel, rapidly approaching threat you do not want to gather more data. That data will come soon enough, and it will be dead people and a completely unmanageable situation. You need to leap before you look.

I think this is why I can’t find any mentions of models in the Korean media I follow. They could see reality in China, and they had experience with MERS, and they knew that this thing was going to be bad. And that’s really all you need to know.

The Tsunami

A tsunami is a complex but physically constrained process. There is a displacement of water somewhere, and that water moves as a ripple for hundreds of kilometers. My father-in-law was out fishing when it happened. The tsunami just passed under his boat. When the water hits the shallows, however, it is forced up, into a giant wave. This is all a mathematical process. You can model the height of the wave, when it will hit and how far it will go.

The experience of a tsunami, however, is very different. It’s just chaos. The water recedes, exposing flopping fish and the sea bed. People walk out to look. Then the wave comes, smashing everything — bodies, buildings, overturning trains. There’s elegant math behind it, but who cares. It’s a waste of time.

I often think about that half-hour between the wave hitting Arugam Bay and then Galle. You could have easily modeled the spread in half-an-hour, but by that point, the wave would have already arrived.

A pandemic is the same thing, except even more unpredictable. Any good epidemiological prediction will always be wrong. In a way, that’s the point. You show people what will happen if they do X. So they do Y. The model could be right, but the result will be wrong.

Any good epidemiological prediction is a complicated version of just shouting “RUN!”, which makes you wonder if it needs to be complicated at all.

A Model Is Always Useless

In the beginning, a novel infectious disease is just that. Novel. You have to react aggressively because you don’t know how bad it could get, not because you do. At this point, a model is useless because you don’t have enough data.

In the middle, once the thing spreads, you do have some data. You can make a model, but that model is still useless. The data is dead people, and you don’t want that. You still need to just move.

If you wait till the very end, when you have your own epidemic, your model has reached the crowning heights of uselessness. It’s like debating fluid dynamics with your head underwater. Who cares? Grab a tree. If you can find some dry land, RUN!

Stop Debating Models

Every country debating models is just debating how precisely they’re fucked. It’s pointless, and it’s getting people killed. I don’t know about the other successful countries, but my observation of Korean media is that this never comes up at all. They just acted. They had experience, they had plans, and they could clearly see the carnage in China. They didn’t debate simulations, they fought reality.

Even today, now that they have time, you never see modeling come up. Indeed, where I live in Sri Lanka we don’t seem to have modeled the spread at all. No one is like ‘this many people will get infected, therefore do something.’ We all find it somewhat self-evident that this is bad.

Modeling COVID-19 is just a waste of time, a parlor trick for pundits, a black swan song on the Titanic. They don’t matter. None of them matter. This thing is bad. You need to act fast. That’s all you need to know.