Printable View
Geography & Public Safety, March 2011
Quote:
Contents
- Police Legitimacy and Predictive Policing
- The Predictive Policing Symposium: A Strategic Discussion
- Proactive Policing: Using Geographic Analysis to Fight Crime
- Experimenting with Future-Oriented Analysis at Crime Hot Spots in Minneapolis
- Geospatial Technology Working Group (TWG): Meeting Report on Predictive Policing
- Technical Tips
- News Briefs
- Geography and Public Safety Events
Jed, thanks for posting this. Really interesting concept "Predictive Policing."
Second city cop has some rather pungent comments about the "crystal ball" unit the CPD has been using.
Added to help:http://www.secondcitycop.blogspot.com/ and is currently the third item 'Spring is Still Here'.
The Crystal Ball Unit had their legs knocked out from under them yesterday.Quote:
Originally Posted by Steve Blair
I have watched the developing debate around attempts to predict for a long time, which has become part of the so-called intelligence-led policing model and can be cited as a gain from using analysis / data-mining.
Taken from a paper by a US data-mining presentation:Link:http://www.detecter.eu/index.php?opt...&id=7&Itemid=9 within summary of the Zurich meeting on data mining.Quote:
X PD used data mining as a tool for determining how best to position police assets in anticipation of crime. Allocating assets so as to increase police presence where a particular incident is expected, for example, might help to prevent crime. She provided two examples of what she considered to be effective data analysis. One involved the application of supervised learning to the problem of random gun fire on New Year’s Eve. Data analysis was used to identify the times and places where the most incidents occurred. This information permitted local police to deploy officers strategically, resulting in a 47% reduction in the number of reported incidents and a reduction in personnel costs.
One of the biggest issues around prediction and analysis is the data available, there is a considerable difference between actual / reported / recorded incidents and crimes. In the UK for example to officers dismay a large proportion of house burglaries are not reported. We have learnt, sometimes painfully, that low-level quality of life issues are far more important to the public than what the police want to do, such as "fighting crime".
From your quoted article it seems that if the data mining predicts a crime and an arrest is made a crystal ball is successful. If nothing happens the crystal ball prevented a crime. Another success.
I have seen New Years Eve shooting cut down significantly from one year to the next. This reduction has to do with putting 300 police in a complex. This deployment does not happen until the swells move into the area.
I've done a considerable amount of work on intelligence-led policing and I find this new thrust for predictive policing a real reach. I've yet to see any law enforcement agency conduct intelligence operations on the same scale or as continuously as the military-intel community. I think its great that police departments are finally reverting to intel processes and smarter employment, but I think most of their so called predictive policing techniques have been done by smart cops for years but not documented. Those that come from data are strictly extrapolations that don't really anticipate change as much as hope that future crime patterns match old crime patterns.
The "think tank" Policy Exchange, who have considerable impact on UK government policies, have a seminar next week on 'Pre-Crime and Predictive Policing'.
From the summary:Link:http://www.policyexchange.org.uk/eve...ent.cgi?id=405Quote:
Predicting where offences will occur and deploying police before crime happens has been an inexact science until recently, but that may soon change. Two pilots of the experimental ‘predictive policing’ method are underway in California and this new approach could have important lessons for UK policing in how forces deploy their resources to prevent crime....
The most robust predictive policing pilot, in Los Angeles, has just begun and shows some promising early results. The lead officer for the LAPD pilot, and a pioneer of predictive policing, Sean Malinowski, will be a speaker alongside George Tita, an expert on predictive models, from the University of California at Irvine.
(Added later in 2016) Link to video of the event:http://www.policyexchange.org.uk/mod...ctive-policing
A lengthy article in The Guardian, sub-titled:He starts with:Quote:
Smart technology and the sort of big data available to social networking sites are helping police target crime before it happens. But is this ethical?
Link no longer works due to copyright and a search on Evgeny Morozov, the author's blog failed.Quote:
The police have a very bright future ahead of them – and not just because they can now look up potential suspects on Google. As they embrace the latest technologies, their work is bound to become easier and more effective, raising thorny questions about privacy, civil liberties, and due process.
The author correctly draws attention to the leviathans of public use IT, for example Facebook & Amazon and asks who reviews their algorithms, for their ethical basis and effectiveness.
Given the clear failure to win the so called 'war on drugs', which has had massive funding and much hi-tech - why would this predictive policing be effective?
The Economist weighs in on the topic.
I get a little queasy when we start handing decision making over to algorithms. As we used to say, you can't quantify the bad things that don't happen.Quote:
Predicting and forestalling crime does not solve its root causes. Positioning police in hotspots discourages opportunistic wrongdoing, but may encourage other criminals to move to less likely areas. And while data-crunching may make it easier to identify high-risk offenders—about half of American states use some form of statistical analysis to decide when to parole prisoners—there is little that it can do to change their motivation.
Found this JHU author's paper whilst looking for something else, on a quick skim read it is a useful summary, but I still have my doubts over this approach.
Link:
The (US) NIJ has commissioned a RAND study and the linked piece quickly summarises the position, form a critical viewpoint:http://nextcity.org/daily/entry/predictive-policing-crime-stats-data-measure
A short NYT article reviews the situation, with a focus on Kansas City, but these two paragraphs struck me - is this option really working?
Link:http://www.nytimes.com/2015/09/25/us...t-crimes.html?Quote:
The Memphis police force, a pioneer in predictive policing, has worked with the University of Memphis for about a decade to forecast crime by noting time and location of episodes and information about victims. Officers then flood those areas with marked and undercover police cars, and also increase traffic stops, the department said.
But violent crime has proved stubborn in Memphis, and the city continues to be one of the most dangerous places in the nation, according to F.B.I. data.
A short article and no links that explain more alas:Link:http://www.telegraph.co.uk/news/worl...dia-posts.htmlQuote:
Police departments in many American cities are using high-tech databases to determine how dangerous individuals might be when officers arrive at a crime scene.The systems take into account criminal history, social media profiles, property records and other factors to produce a “threat score”. Before reporting to a crime scene, police can use the databases to retrieve scores on those inside.
A better explanation was found in WaPo, thanks to a "lurker" and is based on Fresno PD, California and the full title is 'The new way police are surveilling you: Calculating your threat ‘score’:https://www.washingtonpost.com/local...a0c_story.html
The scoring system is briefly described:Interested in more ask via:http://www.intrado.com/bewareQuote:
But perhaps the most controversial and revealing technology is the threat-scoring software Beware. Fresno is one of the first departments in the nation to test the program. As officers respond to calls, Beware automatically runs the address. The searches return the names of residents and scans them against a range of publicly available data to generate a color-coded threat level for each person or address: green, yellow or red.
Exactly how Beware calculates threat scores is something that its maker, Intrado, considers a trade secret, so it is unclear how much weight is given to a misdemeanor, felony or threatening comment on Facebook. However, the program flags issues and provides a report to the user.
A US academic intelligence SME has responded by proving, free, his 2004 article on 'Homeland Security Intelligence: Just the Beginning' which explains a lot, almost predictive? Link:https://www.academia.edu/3695760/Hom..._the_Beginning
Now The Atlantic weighs in with a long article and concludes that Fresno PD's use of 'Beware' is:Link:http://www.theatlantic.com/politics/...danger/423642/Quote:
Beware of this product and proceed only with great caution.
In a somewhat strange IMHO article 'Defence One' has an article on how China is exploiting fusion and the new capacity of IT to predict dissent, if not protest:http://www.defenseone.com/technology/2016/03/thanks-america-china-aims-tech-dissent/126491/?
It starts with:Quote:
What if the Communist Party could havepredicted Tiananmen Square? The Chinese government is deploying a new tool to keep the population from uprising. Beijing is building software to predict instability before it arises, based on volumes of data mined from Chinese citizens about their jobs, pastimes, and habits. It’s the latest advancement of what goes by the name “predictive policing,” where data is used to deploy law enforcement or even military units to places where crime (or, say, an anti-government political protest) is likely to occur. Don’t cringe: Predictive policing was born in the United States. But China is poised to emerge as a leader in the field.
More "cold water" on predictive policing:Link:https://mic.com/articles/156286/crim...ows#.3IhFXDIIhQuote:
But according to a study to be published later this month in the academic journal Significance, PredPol may merely be reinforcing bad police habits. When researchers from the Human Rights Data Analysis Group — a nonprofit dedicated to using science to analyze human-rights violations around the world — applied the tool to crime data in Oakland, the algorithm recommended that police deploy officers to neighborhoods with mostly black residents. As it happens, police in Oakland were already sending officers into these areas.
The cited journal Significance is an Anglo-US publication of the two national statistical groups. The article is behind a pay-wall alas, here is a summary:https://www.statslife.org.uk/signifi...-issue-preview
CIA claims of predicting some unrest up to 5 day ahead:
https://www.engadget.com/2016/10/05/...-5-days-ahead/
Recently RUSI, a Whitehall "think tank" published a report 'Big Data and Policing: An Assessment of Law Enforcement Requirements, Expectations and Priorities', with 54 pgs. and there is a comprehensive summary on this link:https://rusi.org/publication/occasio...t-requirements
It is very UK-centric report, so little mention is made of the various US experiments and schemes.
Today's The Independent on Sunday has an article, based on the report, but has some other comments:http://www.independent.co.uk/news/uk...-a7963706.html
A podcast (33 mns) via WNYC with:Link:https://www.wnyc.org/story/data-driven-policing/Quote:
Andrew Ferguson Professor of Law at the University of the District of Columbia's David A. Clarke School of Law, discusses his book The Rise of Big Data Policing: Surveillance, Race, and the Future of Law Enforcement. He examines big data and algorithm-driven policing and its impact on law enforcement. He also looks at how new technologies will alter the who, where, when and how we police, and why data-driven methods could actually improve police accountability.
Part of his argument is that this approach has grown out of Compstat.
I have heard hints about a large Home Office (central government) project worth £15m on 'big data' and it is actually run locally. So started a look around and found the project is National Data Analytics Solution (NDAS).
Last month there was this 'exclusive' and within an explanation by the police 'lead':Link:https://www.newscientist.com/article...re-it-happens/Quote:
This the first such project of its kind in the world, pooling multiple data sets from a number of police forces for crime prediction, says Donnelly. In the early phases, the team gathered more than a terabyte of data from local and national police databases, including records of people being stopped and searched and logs of crimes committed. Around 5 million individuals were identifiable from the data. Looking at this data, the software found nearly 1400 indicators that could help predict crime, including around 30 that were particularly powerful. These included the number of crimes an individual had committed with the help of others and the number of crimes committed by people in that individual’s social group.
The machine learning component of NDAS will use these indicators to predict which individuals known to the police may be on a trajectory of violence similar to that observed in past cases, but who haven’t yet escalated their activity. Such people will be assigned a risk score indicating the likelihood of future offending.
A short article and the sub-title is: A UK police force uses an algorithm to choose which crimes to investigate. It has led to half as many assaults and public order offences being pursued.
It ends with a salutary warning:Link:https://www.newscientist.com/article/2189986-a-uk-police-force-is-dropping-tricky-cases-on-advice-of-an-algorithm/?Quote:
Police forces only ever know about crimes they detect or have reported to them, but plenty of crime goes unreported, especially in communities that have less trust in the police. This means the algorithms are making predictions based on a partial picture. While this sort of bias is hard to avoid, baking it into an algorithm may make its decisions harder to hold to account compared with an officer’s. John Phillips, superintendent at Kent Police, says that for the types of crimes that EBIT is being used for, under-reporting isn’t an issue and so shouldn’t affect the tool’s effectiveness.
Predictive policing is getting rather than more media coverage in the UK, just why and now is unclear.
Much of the criticism concerns the use of AI in decision-making, even down to "hot spots" where police activity should concentrate. Ethics has also become an issue, with some ethicists now sitting on advisory groups and others "throwing stones from outside".
Today The Guardian has an editorial, sub-titled:Link:https://www.theguardian.com/commentisfree/2019/feb/04/the-guardian-view-on-and-algorithms-big-data-makes-bigger-problems?Quote:
Machines can make human misjudgments very much worse. And should never be trusted with criminal justice
This highlights the civil liberties group Liberty have just issued a critical report:https://www.libertyhumanrights.org.u...inatory-data-0
The BBC News had a short item yesterday and this referred to a critical local report which I had missed:https://www.bbc.co.uk/news/technology-47118229 and https://www.turing.ac.uk/sites/defau...ort_to_wmp.pdf
Two articles, the first I missed is quite lengthy and very critical.
Link:https://motherboard.vice.com/en_us/a...icing-software
The second makes available the FOI documents gathered for the first article.
Link:https://motherboard.vice.com/en_us/a...ing-in-america