Thread by @ElonBachman

1\ Live "Traffic congestion" maps have emerged as a useful tool for tracking the effect of Coronavirus in foreign cities

But care is needed in understanding what they do and don't show

E.g. here is Shanghai:

2\ First thing to note is that what looks like a 50-80% decline on some days is a decline *relative* to off-peak congestion

E.g. a decline from 30% to 5% looks like an 85% fall in traffic, but it's really just a 19% fall in congestion

3\ But wait!

What is "congestion", actually?

According to TomTom, it's a measure of how long a trip takes

So it's not a measure of cars, or of people, but of speed!

To measure actual throughput of people and vehicles, we need to dig deeper...

4\ Traffic is famously non-linear, and in fact is sometime modeled with complex fluid dynamics (turbulent flow, laminar flow, etc.)

Here is a generic diagram, linking speed to passenger throughput:

5\ Note that peak throughput (2000 cars/hr/lane) does not correspond to peak speed

Note also that at peak throughput, large changes in congestion (e.g. b/w 45 and 60kmph) do not correspond to large changes in throughput

Throughput can increase, even as congestion decreases!

6\6 My conclusion: these are neat tools, but care is needed in interpreting them.

Does anyone know of tools that show real time *throughput* in major cities?