The huge flaw in your arguments; "Gun violence". There is no such thing. Violence is violence and it will continue as long as there are bad people wanting to do bad things. Guns are inanimate objects. Just like knives, bats, rocks, etc.
1
We have a habit in this country when confronted with a terrible social reality--the murder rate in Chicago--in seeking out some form of social science research to look for an answer. But the statistics we dig up are always after the fact data rather than looking at before the fact data. Instead of attempting to profile a potential murderer, why not look at the mountains of data we already have that profile the causes of poverty in Chicago. We are the most segregated city in the nation, and have been for decades. Our public services---schools for example---are the most segregated in the country. We have vast areas in the cities where the jobless rates exceed 80%. I could go on, but the Mayors in our city, except for Mr. Washington, have either ignored or been purposeful about maintaining a deeply segregated city with no hope of climbing out of that poverty--those are the numbers we should be attending to.
25
I am a crimnal defense atorney working in several of the poorest cities in NJ, and the USA; Camden Bridgeton, Trenton, and Atlantic City. I remember a politician who said, "It takes a village to raise a child!" and the self help, personal responsibility right wing shot that argument to shreds!!
I am looking at the "biggest risk factors" and i do not see the following red flags in my mind:
1) single family home
2) father in jail for
3) high school drop out
4) siblings in jail for ..
5) economic opportunities job,
6) high school diploma
7) parent(s) addicts ????
8) participation on religious activities .....
9) (hope) for economic advancement out side of criminal enterprise?
I have observed a vast majority of gun deaths are territorial gang and drug violence used to enforce economic advantage to put food on the table for the 2,3,4,5,6 mouths in a family. Baby mom's who were not able to get or didn't trust the clinic for care. Planned parenthood and access to birth control is a key component of this issue!!!!! Why anyone would argue against that for any mom trying to become upwardly mobile in this environment (village) is incomprehensible and downright ignorant of the realities that actually do exist! Another aspect of gun violence is shooters killing those that come into the cities with wades o money to buy drugs who become victims of opportunistic economic killings. Drugs!!!! Then there are the inner city overdose deaths from opiate dependence. Big Pharma!!!!!
I am looking at the "biggest risk factors" and i do not see the following red flags in my mind:
1) single family home
2) father in jail for
3) high school drop out
4) siblings in jail for ..
5) economic opportunities job,
6) high school diploma
7) parent(s) addicts ????
8) participation on religious activities .....
9) (hope) for economic advancement out side of criminal enterprise?
I have observed a vast majority of gun deaths are territorial gang and drug violence used to enforce economic advantage to put food on the table for the 2,3,4,5,6 mouths in a family. Baby mom's who were not able to get or didn't trust the clinic for care. Planned parenthood and access to birth control is a key component of this issue!!!!! Why anyone would argue against that for any mom trying to become upwardly mobile in this environment (village) is incomprehensible and downright ignorant of the realities that actually do exist! Another aspect of gun violence is shooters killing those that come into the cities with wades o money to buy drugs who become victims of opportunistic economic killings. Drugs!!!! Then there are the inner city overdose deaths from opiate dependence. Big Pharma!!!!!
11
A gang member is NOT more likely to be shot than a non-gang member. Sure. I believe that. Anyone got a bridge to sell me? If this Strategic Subject List has not been politicized, I'll even eat that bridge. Good Luck Chicago.
17
One great thing about using computers to predict crime is that they take politics out of crime prevention, and reestablish the importance of results.
2
While risk assessments are valuable, any type of automated assessments that can influence lives need heavy human intervention and review from a wide audience. Algorithms, like children, carry the biases of their creators. If we can't stamp out racism and prejudice in our daily lives, how will we keep it away from our machines?
2
is skin color used in this algorithm? If so the Chicago police dept needs to immediately suspend it use, whether or not it is effective. Period.
1
Read the article please, the SSL "stays away from variables that could discriminate, like race, gender and geography." Which is politically correct but ignoring facts, as other statistic show most of the victims and perpetrators are young black men in a small geographic area.
15
Yoda, I guess that you really are in another galaxy. In my galaxy, this comes directly from the Chicago website, "Based on this time frame’s version of the Strategic Subject Algorithm, individuals with criminal records are ranked using eight attributes, not including race or sex."
Why do people look for race in every event occurring in the world?
Why do people look for race in every event occurring in the world?
7
Why? This kind of thinking is what causes even more problems.
And in this very journal:
https://www.nytimes.com/2017/06/13/opinion/how-computers-are-harming-cri...
I'll note that the algorithm in question is being hidden under claims of "proprietary technology". The public has no way of knowing what is going on with a program that may have life altering consequences. GIGO.
"Just trust us, the Police are your friends."
https://www.nytimes.com/2017/06/13/opinion/how-computers-are-harming-cri...
I'll note that the algorithm in question is being hidden under claims of "proprietary technology". The public has no way of knowing what is going on with a program that may have life altering consequences. GIGO.
"Just trust us, the Police are your friends."
3
Regardless of how targeting is done, despite whatever sophisticated models are contrived, humans will remain violent and unpredictable. Chicago is a city that has long and perversely prided itself on its ugly reputation for violence, a place where certain demographics are known to have no respect for any other human's life. Chicago will remain emblematic of the saying that it was Cain who built the first city, and his descendants who live there.
3
One suspects this is a version of multiple regressions whose predictive value is sometimes based on how the model is built, that is, what is the INCREMENTAL value of x knowing y,z, a,b,,c and so forth
=
also from what I have read, the model predicts to specific persons, who are then approached by PD in Chicago and elsewhere, of the form we are watching YOU b/c we think YOU are likely to commit more crimes
i see no illegality here,
and it is reported to work, to predict and to deter
=
the sequencing of the variables in the creation of the model almost controls what the model tells us, if all factors are correlated and co-present
=
also from what I have read, the model predicts to specific persons, who are then approached by PD in Chicago and elsewhere, of the form we are watching YOU b/c we think YOU are likely to commit more crimes
i see no illegality here,
and it is reported to work, to predict and to deter
=
the sequencing of the variables in the creation of the model almost controls what the model tells us, if all factors are correlated and co-present
It is not that gang membership is necessarily unimportant. It could also be the case that gang membership is so strongly correlated with the other variables like being shot, being arrested etc. that those tend to be better variables for prediction purposes. What would be interesting to see is the correlation coefficient between the variables themselves.
12
I also don't wonder if gang membership has an interactive effect. That is, are gang members involved in previous shootings more likely to be involved in future violence than non-gang members?
What are the algorithms and tea leaves worth against the three missing words in the 2nd Amendment: the right of the people to keep and bear Arms, EVERYWHERE AND ALWAYS, shall not be infringed ?.
1
don't forget the other three important words: WELL REGULATED MILITIA.
12
This may work if properly verified, but will be a civil rights disaster if not properly validated. It's Risk Management 101.
Verify: Continually test the accuracy of the algorithm against carefully compiled metrics on crime as time goes on. The metrics are based upon the assumptions used to create the algorithm and therefore verify it. That is the obvious task.
Less obvious is validation: See if the community is better off with the verified algorithm in place even with less crime. Look at the unexpected consequences and re-examine the choice of categories and criteria upon which the algorithm is based. Validation is vital because the act of making those choices at the very beginning always introduces bias and errors.
Verify: Continually test the accuracy of the algorithm against carefully compiled metrics on crime as time goes on. The metrics are based upon the assumptions used to create the algorithm and therefore verify it. That is the obvious task.
Less obvious is validation: See if the community is better off with the verified algorithm in place even with less crime. Look at the unexpected consequences and re-examine the choice of categories and criteria upon which the algorithm is based. Validation is vital because the act of making those choices at the very beginning always introduces bias and errors.
6
The article should clarify a bit what is meant by "No of X's". I.e. when the article states that "No. of arrests for violent offences" has a score of +15, does it mean that the person with three violent arrests gets +45?
In this case, I think that the clarification is needed, since I think that the upper bound of the number of shooting incidents (as victim) that a person can have is much smaller than the number of narcotic arrests - which means that while an individual narcotic arrest is not a great predictor, having 20 of them is a very, very good one.
In this case, I think that the clarification is needed, since I think that the upper bound of the number of shooting incidents (as victim) that a person can have is much smaller than the number of narcotic arrests - which means that while an individual narcotic arrest is not a great predictor, having 20 of them is a very, very good one.
4
If you got away with something once, you wouldn’t be in any database but you’d be more likely to do it again.
9
Membership in a gang may also appear low in the risk score because it is likely highly correlated with other factors. For example if only gang members were committing murders, but only those who were "experienced" and had prior arrests for assault committed murders and people were far more likely to be arrested for assault if they were gang members, then the gang member contribution to the score would be reduced.
6
This algorithm is probably smudgy because it tries to do two things - predict victims, and predict shooters; the antecedents must be enormously different between the two.
As for predicting who will try to kill somebody with a gun, the problem with statistical modelling of this type of action is that it's an extreme behavior (regardless of how large the number of occurrences seems to be getting). Human behavior distributes itself along normal curves, and the people you're trying to pre-identify are by definition at the tail end of the curve, outliers, the hardest to predict. Any model will over- or underestimate the number of people who might shoot, and the final yes/no determinant are chance factors you can't anticipate in a model.
In NYC, the Administration for Children's Services analyzes its tens of thousands of reports of child abuse every year by criteria that are not publicized, trying to identify households most at risk of producing dead children. In the main, it is quite successful at preventing these events - where it knows the household beforehand; but every year there are reports of dead child victims in households previously known to ACS. (The number of these cases is also kept secret.) The media is perfectly right to criticize faulty procedures in cases where these can be identified, but the fact is that zero fatalities (in child abuse, or traffic safety, or any other setting) are unobtainable. It's just the way large numbers work.
As for predicting who will try to kill somebody with a gun, the problem with statistical modelling of this type of action is that it's an extreme behavior (regardless of how large the number of occurrences seems to be getting). Human behavior distributes itself along normal curves, and the people you're trying to pre-identify are by definition at the tail end of the curve, outliers, the hardest to predict. Any model will over- or underestimate the number of people who might shoot, and the final yes/no determinant are chance factors you can't anticipate in a model.
In NYC, the Administration for Children's Services analyzes its tens of thousands of reports of child abuse every year by criteria that are not publicized, trying to identify households most at risk of producing dead children. In the main, it is quite successful at preventing these events - where it knows the household beforehand; but every year there are reports of dead child victims in households previously known to ACS. (The number of these cases is also kept secret.) The media is perfectly right to criticize faulty procedures in cases where these can be identified, but the fact is that zero fatalities (in child abuse, or traffic safety, or any other setting) are unobtainable. It's just the way large numbers work.
2
This is interesting, but focuses on WHO is committing crimes, not on WHY. I was a prosecutor in Chicago (Assistant State's Attorney) back in the 1980's when we were running about 900 murders/year. High now, but not as high. The main driver of violence then and now is the drug trade. When a commodity is illegal and buyers and sellers cannot make use of the courts (as Jeff Sessions helpfully pointed out), the only alternative is private enforcement, which is inevitably violent. Chicago, if anyplace, should understand the connection between illegality and violence. All the violence during Prohibition (1920-1933) was due to illegality; Al Capone provided enforcement services, among other things. The day Prohibition was repealed, beer distributors with franchise disputes went back to suing each other in Circuit Court and the mayhem stopped. After two generations of drug prosecutions in Chicago (as elsewhere), thousands of young men are precluded from any legitimate employment. Prosecution is just recycling them back into the drug trade. The result is failed futures, overincarceration, fewer marriageable males, broken homes, and schools challenged by fatherless children. See www.lawenforcementactionpartnership.org. We are a group of current and former law enforcement (police, prosecutors, judges) who have seen the damage caused by the War on Drugs. Legalize and regulate. Expunge past convictions and integrate these inner city areas into the legitimate economy.
59
Agreed Inge, the war on drugs has been a failure. However, also Hollywood and parents of youth especially parents of minority youth also are at fault re gun violence.
Without our cultural gun sickness, these offenders would not be dead or wounded or serving long prison terms.
Hollywood is a big driver of this with grat. violent films targeting youth especially youth of color and all adults should condemn it especially parents of youth of color and come up with a policy of legality, regulation, responsibility and non promotion of guns...a policy that worked with other dangerous objects, vices like alcohol, drunk driving, cigs. etc.
Without our cultural gun sickness, these offenders would not be dead or wounded or serving long prison terms.
Hollywood is a big driver of this with grat. violent films targeting youth especially youth of color and all adults should condemn it especially parents of youth of color and come up with a policy of legality, regulation, responsibility and non promotion of guns...a policy that worked with other dangerous objects, vices like alcohol, drunk driving, cigs. etc.
13
Excuse me, but you've missed the main point -- illegal use of guns, resulting in death and injury. Football star Plaxico Burress had an illegal gun in Manhattan and got a mandatory 2-year imprisonment.
What happens in Chicago? A lot less enforcement. And a lot more problems. There's the issue, IMHO.
Too punishing? More like a message -- "hey, knock it off. Now."
What happens in Chicago? A lot less enforcement. And a lot more problems. There's the issue, IMHO.
Too punishing? More like a message -- "hey, knock it off. Now."
5
Legalize and regulate just like the opioid pain killers that have entered the market in the last 15 years. Deaths from this legalized drug have now past auto accidents as the number one killer of young adults in the country.
1
I think there needs to be a better presentation of the data for the Figure shown. As shown visually, it's misleading, suggesting that the risk of gun violence is not proportional to the SSL score.
But the proportional risk of violence adjusted for the population in each stratum shows very large differences. 1% of the gun violence is done by people with scores between 480-499 (0.03%). That's an adjusted risk of 33.3 for committing gun violence. For someone with a score of 360-379, the adjusted risk is 7.5 (15%/2%). Indeed, the adjusted risk score tracks fairly well with the PTV list breakdown.
The article would have benefited greatly from showing this adjusted risk per stratum. As it stands, the data risks being seen as presented to fit a certain narrative.
But the proportional risk of violence adjusted for the population in each stratum shows very large differences. 1% of the gun violence is done by people with scores between 480-499 (0.03%). That's an adjusted risk of 33.3 for committing gun violence. For someone with a score of 360-379, the adjusted risk is 7.5 (15%/2%). Indeed, the adjusted risk score tracks fairly well with the PTV list breakdown.
The article would have benefited greatly from showing this adjusted risk per stratum. As it stands, the data risks being seen as presented to fit a certain narrative.
20
It appears similar to the Operations Ceasefire findings in the 1990s, where a small number of people appeared to be responsible for most of the firearm violence. They also found that a number of people in the neighborhoods were arming themselves for protection, people who otherwise weren’t involved in a lot of violent crime. The program was successful if reducing gun violence but it was later dropped, and the gun violence increased again.
2
The outcome measure, "party to violence," is pretty meaningless. It is largely a measure of lack of intelligence and experience. Smart criminals are more able to evade notice of the police. Older criminals have learned not to get caught committing crimes and not to be vulnerable to assaults. So one would not expect SSL to have much impact on reducing crime.
1
Worthy doctoral thesis, but task is to seize & destroy hand-guns [12-15 in slammer for 3-D printing]; if there's a better solution in principle, this nanagenarian has yet to hear of it.
4
If being the victim of a violent crime is the most important predictive factor, it seems logical that increased victim assistance (instead of including them in 'round-ups') would help. It also seems logical that even extreme fiscal conservatives would support victim assistance programs based on this.
8
Age is always a leading factor in crime rates. It is the main reason for the dramatic drop in murders from 1992 to present in most American cities, ie the aging of the baby boomers.
In today's America gentrification can have a dramatic effect too in major cities. NYC is at or near the top in major American cities re gentrification and the type of gentrification and has enjoyed and continue to enjoy a dramatic drop in the murder rate. Minority areas like Bushwich, Williamsburg have shown a dramatic drop because of the influx of young adults coming from many western nations. They do not suffer from America's cultural gun sickness, ie use the gun as the first resort to solve all problems instead of the last resort.
Chicago has not had as much gentrification and also not the same type where minority areas have completely turned into young "hipster" areas.
Also app. half of all murders are committed by minorities but the other half are committed by whites. Whites in conservative states with a gun culture have generally not gentrified and in general do not enjoy the lower murder rates that gentrification has brought to many (not all), more liberal cities like NYC.
In today's America gentrification can have a dramatic effect too in major cities. NYC is at or near the top in major American cities re gentrification and the type of gentrification and has enjoyed and continue to enjoy a dramatic drop in the murder rate. Minority areas like Bushwich, Williamsburg have shown a dramatic drop because of the influx of young adults coming from many western nations. They do not suffer from America's cultural gun sickness, ie use the gun as the first resort to solve all problems instead of the last resort.
Chicago has not had as much gentrification and also not the same type where minority areas have completely turned into young "hipster" areas.
Also app. half of all murders are committed by minorities but the other half are committed by whites. Whites in conservative states with a gun culture have generally not gentrified and in general do not enjoy the lower murder rates that gentrification has brought to many (not all), more liberal cities like NYC.
10
Let me reply to my post....like the old ad....you don't have to be Jewish to enjoy levi bread (spelled wrong).....also you dont have to not believe in the right for every american to own a gun to dramatically lower our gun death total, the two are not exclusive of each other...
4
So, the algorithm works but can be improved. Any surprise in that?
1