What the GA models really say…

I decided to follow up on my own post and spend some time with the simulator and charts that are making so much news. https://analytics-tools.shinyapps.io/covid19simulator07/

First, if you see their actual charts you will see a third graph that no one is sharing. Lockdown to a full reopening if even better.

So what do these really say? What are these models. Here are their definitions that you are unlikely to see anywhere.

We simulate model outcomes until August 31, 2020 under different levels and duration of interventions, as defined below:
1: Minimal restrictions: There is no intervention in place to reduce the spread of COVID-19. However, there is an assumed level of learned social awareness (handwashing, avoid close contact when sick, etc.) that is included in the model. Please see the methodology for more details.
2: Current intervention: For most states this is a stay at home order, where people are advised to stay at home except for essential needs such as grocery shopping and picking up prescriptions. The New York Times provides a list of the current interventions in each state.
3: Lockdown: There is a complete ban on travel, including canceling flights and closing inter-state travel and local travel, as has been done in some countries such as Italy, China, and India. Exceptions are made for essential needs such as grocery shopping and picking up prescriptions.

To summarize, the grey (minimal) is a return to a very normal open environment. Current is a bit open since each state is very different on “current”. And then Lockdown is no travel with some open for essentials.

Working with those definitions where are you today? What is re-opening under the three phases? I would argue that most of the country is currently in lockdown and the reopening is not minimal as shown in this graph but a minor nudge beyond stay at home. What is missing the the stages of current of the three phases of opening before we return to minimal.

The third and ignored chart is the most fascinating to me. It seems to say that if you lockdown for four weeks you can reopen widely with little impact. Isn’t that exactly what we just did? I worked a few other models through the simulator that do not have the same hype factor and “GA is blowing the doors wide open”. But I hope they are reality.

These two charts show sticking with current tight lock down for two more weeks, then 4 weeks with “current” restrictions and then wide open (mirroring as best possible phase 1, 2 and 3). Or a second strategy of just sticking with the “current” for 12 more weeks. What these graphs show is that tighter restrictions and a slow opening is slightly better than “current”. However, the basic fact is that if you open too far, too early you can get a spike. Is that news?

But what is shows even more is the uncertainty. Notice that the shaded background says that we “may” have a spike in deaths and cases, but the predictive line line says “or we may not”.

My basic thesis here is that the models still do not reflect reality. “Current” is a wide variety of options across the 50 states and the world. I would argue we lean closer to lock down. However, the overall impact of those definitions does not seem affect the curves greatly. The biggest impact is a return to “minimal”. A metric that no one is advocating in the short term. These models also ignore the economic impact that a lock down and/or current policies have on businesses, people and neighborhoods. Fact is this is a hard problem with a lot of variables. I appreciate the efforts to try and get a handle on it from every front. But this as with every other model is being intentionally misused.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: