White Collar Crime Risk Zones(whitecollar.thenewinquiry.com)
whitecollar.thenewinquiry.com
White Collar Crime Risk Zones
https://whitecollar.thenewinquiry.com/
17 comments
I particularly enjoyed "
Fig. 4: Example of features in a landscape that create unique behavior settings for white collar criminal activity." with a fisheye picture of skyscrapers.
I thought Matt Levine's daily had a good take:
"The deep message is that if you define criminality based on your negative perception of some disliked group, then your criminals are going to look like that group. If you assume that rural white people with guns are hunters and urban minorities with guns are gang members, then your predictive policing efforts will look for guns in cities rather than forests. If you assume that Wall Street is an industry whose business model is fraud, then your predictive policing efforts will look for fraud in midtown Manhattan. In both cases, it is at least plausible that the group perception leads to the definition of criminality, rather than the reverse. What makes you a criminal is not doing a certain objectively defined sort of act; it's being a certain sort of person."
https://www.bloomberg.com/view/articles/2017-04-26/bank-meet...
"The deep message is that if you define criminality based on your negative perception of some disliked group, then your criminals are going to look like that group. If you assume that rural white people with guns are hunters and urban minorities with guns are gang members, then your predictive policing efforts will look for guns in cities rather than forests. If you assume that Wall Street is an industry whose business model is fraud, then your predictive policing efforts will look for fraud in midtown Manhattan. In both cases, it is at least plausible that the group perception leads to the definition of criminality, rather than the reverse. What makes you a criminal is not doing a certain objectively defined sort of act; it's being a certain sort of person."
https://www.bloomberg.com/view/articles/2017-04-26/bank-meet...
Hmm. The JPMorgan Chase building is surrounded by red squares, but for some reason the map indicates zero white collar crime there. I looked at a little town I used to live in and the whole town had one single bright red square, with the rest blank - the one red square was the location of a breakfast restaurant and some gift shops (nobody lived there.)
If there's any useful information to be gained from this map, I'm not finding it.
If there's any useful information to be gained from this map, I'm not finding it.
Hi. I'm one of the project creators. Happy to answer any questions.
What specifically are you trying to say/show with this satire?
I mean I get the point but I think you are underestimating the usefulness of models for predicting white collar crime and crime in general by giving a really bad model for doing so. In terms of data analysis, the geotag stuff has decent specificity in terms of narrowing down your universe (from entire US to city blocks), but the face stuff probably doesn't have very good specificity (probably lots of false positives there). A good "data science" person would probably try to find factors with higher specificity. But most of the stuff in this model is already known via common sense and acted upon. The SEC knows to look for white collar crime in Manhattan and not Lebanon NH without such a model.
I mean I get the point but I think you are underestimating the usefulness of models for predicting white collar crime and crime in general by giving a really bad model for doing so. In terms of data analysis, the geotag stuff has decent specificity in terms of narrowing down your universe (from entire US to city blocks), but the face stuff probably doesn't have very good specificity (probably lots of false positives there). A good "data science" person would probably try to find factors with higher specificity. But most of the stuff in this model is already known via common sense and acted upon. The SEC knows to look for white collar crime in Manhattan and not Lebanon NH without such a model.
As a proof of concept, we have downloaded the pictures of 7000 corporate
executives whose LinkedIn profiles suggest they work for financial
organizations, and then averaged their faces to produce generalized
white collar criminal subjects unique to each high risk zone. Future
efforts will allow us to predict white collar criminality through
real-time facial analysis.
https://arxiv.org/pdf/1611.04135v1.pdf (Automated Inference on Criminality using Face Images)
Too bad there haven't been enough arrests of financial criminals to get actual mugshots.
I tried to use your app but I still got defrauded. 2/5 stars
Even the white paper doesn't clearly identify the source data on financial crimes, and it's hard not to suspect that the locations used for the crimes are just the locations at which people charged with financial crimes of the type included in the not-well-identified FINRA data set used worked.
And it's useless anyway; a violent crime heatmap is useful for deploying police or avoiding violent crime. A financial crime heatmap is not, because police patrolling the neighborhood doesn't deter or detect financial crimes, and avoiding the locations where such crimes are "located" doesn't actually protect you from the crimes.
And it's useless anyway; a violent crime heatmap is useful for deploying police or avoiding violent crime. A financial crime heatmap is not, because police patrolling the neighborhood doesn't deter or detect financial crimes, and avoiding the locations where such crimes are "located" doesn't actually protect you from the crimes.
> And it's useless anyway; a violent crime heatmap is useful for deploying police or avoiding violent crime ...
I'm guessing your reaction is kinda the point of the project. Horrific crimes are routinely committed that injure millions of people, with no punishment.
We know where these crimes are happening, and what kind of people are committing them, yet there is very little action.
If it's OK to profile people by their appearance for drug and street crime, why not do the same for financial swindlers, cheats and fraudsters?
I'm guessing your reaction is kinda the point of the project. Horrific crimes are routinely committed that injure millions of people, with no punishment.
We know where these crimes are happening, and what kind of people are committing them, yet there is very little action.
If it's OK to profile people by their appearance for drug and street crime, why not do the same for financial swindlers, cheats and fraudsters?
> A financial crime heatmap is not, because police patrolling the neighborhood doesn't deter or detect financial crimes
It would if they were doing the right type of patrolling. I think part of the point is that's something we haven't historically emphasized.
It would if they were doing the right type of patrolling. I think part of the point is that's something we haven't historically emphasized.
I don't think that really applies here. For example in NYC you have almost no white collar crime in Brooklyn, Queens or the Bronx but you do have hotspots up in Stamford and Greenwich where there are many fewer people.
If you zoom all the way out, sure, it's just a population heat map, but this is intended to be (a satire of) a neighborhood-level map.
If you zoom all the way out, sure, it's just a population heat map, but this is intended to be (a satire of) a neighborhood-level map.
Impressively, it's even picking up some block-level data. Notice the hotspot near Court Square in Long Island City:
that's a Citibank tower.
I thought of that too.
Compare the US population density map [1] with the white collar crime map, zoomed out. On the national scale, they match moderately well.
There are interesting hotspots. Etsy HQ in Brooklyn is one.
[1] https://www2.census.gov/geo/img/maps-data/maps/density80.gif
There are interesting hotspots. Etsy HQ in Brooklyn is one.
[1] https://www2.census.gov/geo/img/maps-data/maps/density80.gif
Yeah, all the value here seems to be at city level zoom. Which makes some sense, since it's parodying neighborhood crime maps.
I zoomed out and immediately thought it looked like this, prepared this link by putting it in my copy buffer... and it's the only top-level comment here :)
But actually, looking at New York it's more of a "daytime" population density map than a "nighttime" population density map, which is what I think the XKCD one would be. So I guess people are committing their white collar crimes at work, at least in New York.
But actually, looking at New York it's more of a "daytime" population density map than a "nighttime" population density map, which is what I think the XKCD one would be. So I guess people are committing their white collar crimes at work, at least in New York.
"We therefore plan to augment our model with facial analysis and psychometrics to identify potential financial crime at the individual level. As a proof of concept, we have downloaded the pictures of 7000 corporate executives whose LinkedIn profiles suggest they work for financial organizations, and then averaged their faces to produce generalized white collar criminal subjects unique to each high risk zone. Future efforts will allow us to predict white collar criminality through real-time facial analysis."
https://whitecollar.thenewinquiry.com/static/whitepaper.pdf