Will the US see a massive riot in 2020?

Last updated: July 18, 2020

Metaculus question

This question asks: Will the United States experience an incident of a massive riot in 2020?

For the purposes of this question, ‘massive riot’ is defined as an event of rioting or civil unrest in which any of the following conditions is met:

  1. At least 50 people die due to violence, either due to the actions of rioters or other civilians, or police, military, national guard, or other law enforcement or government agents.
  2. At least 10,000 people are arrested by police, military, national guard, or other law enforcement or government agents.
  3. Property damages are credibly estimated at $1 billion or more. This estimate may come from either the local government of the locality in which the incident takes place, the state government, the federal government, or a major US news publication such as the New York Times, Washington Post, or major broadcast news networks.

Updates

August 30th


Current Prediction

Spreadsheet

July 18: 15%

August 30: 14%

Resolution: 0%, there was not a massive riot.


Base Rate

Incidents of civil unrest are very common, and usually occur several times per year, but rarely do they rise to the level of a “massive riot” as defined in this question. In the last ~100 years, I have only found 8 incidents that qualify as a massive riot.

  1. Elaine massacre (1919): 100-300 dead
  2. Ocoee massacre (1920): 30-50 dead
  3. Tulsa race massacre (1921): > 50 dead
  4. Battle of Blair Mountain: 63-133 dead
  5. Rosewood massacre (1923): 8-150 dead
  6. May Day protests (1971): 12,000 arrested
  7. New York City blackout (1977): $300 million in damages, which is $1.2 billion in 2017 dollars
  8. Los Angeles riots (1992): 63 dead, 12,111 arrests, and > $1 billion in damages

This means there are, on average, 8/100 massive riots per year, or an 8% chance of a massive riot each year.

Because there are are only 166 days left in the year, the base rate for the remainder of the year is (123/365) * 8% or 2.70%.

Limitations:

For and Against

The killing of George Floyd

The killing of George Floyd has sparked widespread protests, many incurring tens, perhaps hundreds, of millions in property damage and widespread violence. Despite this, no protest to my knowledge has met the definition of massive riot.

Although it seemed that the intensity of the protests was beginning to diminish, recent events in Kenosha show that destructive unrest is still possible. How many more instances of this should we expect? According to a WaPo database, there have been on average 25 killings of unarmed black men per years at the hands of police between 2015 and 2019, inclusive. This works out to about 2 per month and there are 4 months left in the year. Suppose that 20% are captured on camera and go viral. That means I’d expect there to be 1 or 2 more videos before the end of the year. As we’ve seen, a single instance can cause unrest for weeks.

The George Floyd trial itself could cause unrest if Derek Chauvin in acquitted, but the trial is set for next year.

Likelihood ratio: 2:1

Economic turmoil

Unemployment is high and some are warning of an evictions crisis. This could lead to civil unrest. Although the federal level moratorium is still on track to expire, some states are extending them at the local level.

Likelihood ratio: 3:2

COVID

I see little evidence that COVID is directly causing unrest that isn’t already accounted for by the Economic Turmoil update. That said, I update slightly because I believe that COVID causes unknown unknowns that are hard to anticipate in advance.

Likelihood ratio: 5:4

Presidential election

The 2016 election triggered riots. If Trump wins again, I anticipate some unrest, but I think it’s unlikely to rise to the level of a massive riot. That said, the election is more polarized than ever, and coronavirus has destabilized it further. If people feel like the GOP has used the virus as cover for disenfranchising them by shutting down polling stations or making mail in ballots hard to get, this could lead to unrest.

Likelihood ratio: 3:2