Dashcam images reveal where police are deployed

> Before describing the details of the framework, the high-level intuition is that we are reweighting the Nexar data sample, which is sampled from a non-representative set of locations, so that it matches the locations where different demographic groups actually live. For example, to calculate the police deployment levels that Asian residents of New York City experience, we reweight the original data sample to upsample neighborhoods with larger Asian populations.

> Overall, our estimation procedure compensates for two types of potential bias. Equation 2 compensates for a data bias, reweighting the Nexar dataset (which is sampled from a set of locations which does not necessarily match the population distribution; Figure 1) to match the population distribution of demographic subgroups. This is conceptually similar to inverse propensity weighting procedures [4] which are used to compensate for non-representative data in other settings. Equation 3 compensates for imperfect model performance, and allows us to check that model performance is unbiased (i.e., calibrated) across demographic subgroups.

Section 4.1 goes into the mathematical functions they use to address the data set.

Section 3.2 describes the data set and how it is geographically distributed.

> Data was provided to us by Nexar in two phases. Phase 1 consists of 3,987,835 images sampled prior to September 1 2020, and is extremely geographically and temporally skewed. Geographically, it is concentrated within the boroughs of Manhattan and Brooklyn, and does not contain data from the boroughs of Staten Island, Queens, and the Bronx at all; temporally, it overrepresents data from Thursday nights. Phase 2, which constitutes the majority of the dataset, consists of 20,816,019 images sampled after October 4 2020, and is much more geographically and temporally representative: it is sampled at all times of the day, on all days of the week, and also covers the entire geographic area of New York City.

> Because Phase 2 is much more representative than Phase 1, we conduct our primary analysis of disparities using only data from Phase 2. We additionally conduct numerous validations and bias corrections, described in §4.1, to compensate for non-representative sampling in the dataset. Geographic and temporal coverage during the Phase 2 period is very good. Specifically, 100% of hours during the Phase 2 period are covered; 99.6% of Census Block Groups (CBGs)3 have at least one image, with a mean of 168.2 images per CBG; 88% of roads contained within the borders of New York City are covered by at least one image, using data from OSMNX [6]. Figure 1 summarizes geographic data availability; Figure S1 summarizes temporal data availability.

You are saying that for thousands of years of civilization nobody got paid for law enforcement before capitalism? Yeah, that doesn't sound very plausible.

I don't know what to tell you.

People were paid to protect someone's stuff, those people were called guards, but they weren't enforcing law. Some people were given ownership of a place and the people in that place, they were called lords or knights, but they weren't enforcing laws, they were enforcing their will.

Some communities had expectations and rules, and the community was self policing - if you did theft, the mob would come punish you. They weren't enforcing laws and certainly weren't paid. Their justice was uneven and often brutal.

Sometimes soldiers would be pressed into enforcing laws, but they were being paid to be soldiers.

Eventually we got to some folks who were paid to enforce laws. Some of the earliest were "Shire Reeves" (from which we derive the word sheriff). A shire reeve was a single man who was charged with keeping order in a shire. That's not exactly enforcing laws, but it's close enough in spirit. A reeve would hire temporary folks in a posse if needed to achieve a temporary goal, but did not have a police department.

Temporary posses were not uniformed, and were typically paid for short term labor - arrest this guy or get that property back. A reeve may have guards for his safety or to intimidate local folks, but those guards weren't really law enforcement.

Around the fourteenth century, governments started relying on formal roles known as justices of peace, or conservators of peace, who were explicitly charged with binding people to laws and enforcing laws. They were relatively large in number and the practice continued through to the industrial revolution. Except JPs weren't paid, they were typically gentry who enjoyed the social status it granted them.

The history of US police is really fascinating, and not something most people really every dive into. They believe it was always this way, but it very much wasn't. Police as a concept like we know them is very, very modern.

Let’s also quote why

“We report racial and socioeconomic disparities without attempting to control for other covariates. First, if New York City residents of different races face different levels of police deployments, that disparate impact is itself important; it can also bias algorithms trained on downstream policing data irrespective of the true causal mechanism. Second, controlling for other factors in policing data can be difficult to interpret, introducing concerns about omitted variable bias and model misspecification which make it difficult to identify which factor is truly the “cause” of higher police deployments. For the sake of transparency and simplicity, therefore, we report results by stratifying each variable separately, noting that these disparities are themselves important but that multiple causal mechanisms may underlie them.”

Inciting race hate is not healthy human behavior. Getting mad at someone who incites race hate is perfectly healthy IMO.