Federal Times: “Agencies would stop producing a variety of unnecessary reports, under legislation passed by the House April 28.
The Government Reports Elimination Act would cut reports from across government and save agencies about $1 million over the next five years. The legislation is sponsored by House Oversight and Government Reform Committee chairman Darrell Issa, R-Calif, and by Reps. Gerry Connolly, D-VA., and Rob Woodall, R-Ga. Senators Mark Warner, D-Va., and Sen. Kelly Ayotte, R-N.H., have introduced a companion bill in the Senate.
“Congress relies on accurate, timely reports to inform its spending and policy decisions, but outdated or duplicative reports are simply a waste of government resources,” Issa said in a press release.
Connolly said it is important that Congress leverage every opportunity to streamline or eliminate antiquated reporting requirements in a bipartisan way.
“Enacting our bipartisan legislation will free up precious agency resources, allowing taxpayer dollars to be devoted to operations that are truly mission-critical, high priority functions,” Connolly said.”
Bill at: http://www.cbo.gov/publication/45303
Crowdsourcing the future: predictions made with a social network
New Paper by Clifton Forlines et al: “Researchers have long known that aggregate estimations built from the collected opinions of a large group of people often outperform the estimations of individual experts. This phenomenon is generally described as the “Wisdom of Crowds”. This approach has shown promise with respect to the task of accurately forecasting future events. Previous research has demonstrated the value of utilizing meta-forecasts (forecasts about what others in the group will predict) when aggregating group predictions. In this paper, we describe an extension to meta-forecasting and demonstrate the value of modeling the familiarity among a population’s members (its social network) and applying this model to forecast aggregation. A pair of studies demonstrates the value of taking this model into account, and the described technique produces aggregate forecasts for future events that are significantly better than the standard Wisdom of Crowds approach as well as previous meta-forecasting techniques.”
VIDEO:
Mapping the Intersection Between Social Media and Open Spaces in California
Stamen Design: “Last month, Stamen launched parks.stamen.com, a project we created in partnership with the Electric Roadrunner Lab, with the goal of revealing the diversity of social media activity that happens inside parks and other open spaces in California. If you haven’t already looked at the site, please go visit it now! Find your favorite park, or the parks that are nearest to you, or just stroll between random parks using the wander button. For more background about the goals of the project, read Eric’s blog post: A Conversation About California Parks.
In this post I’d like to describe some of the algorithms we use to collect the social media data that feeds the park pages. Currently we collect data from four social media platforms: Twitter, Foursquare, Flickr, and Instagram. We chose these because they all have public APIs (Application Programming Interfaces) that are easy to work with, and we expect they will provide a view into the different facets of each park, and the diverse communities who enjoy these parks. Each social media service creates its own unique geographies, and its own way of representing these parks. For example, the kinds of photos you upload to Instagram might be different from the photos you upload to Flickr. The way you describe experiences using Twitter might be different from the moments you document by checking into Foursquare. In the future we may add more feeds, but for now there’s a lot we can learn from these four.
Through the course of collecting data from these social network services, I also found that each service’s public API imposes certain constraints on our queries, producing their own intricate patterns. Thus, the quirks of how each API was written results in distinct and fascinating geometries. Also, since we are only interested in parks for this project, the process of culling non-park-related content further produces unusual and interesting patterns. Rural areas have large parks that cover huge areas, while cities have lots of (relatively) tiny parks, which creates its own challenges for how we query the APIs.
Broadly, we followed a similar approach for all the social media services. First, we grab the geocoded data from the APIs. This ignores any media that don’t have a latitude and longitude associated with them. In Foursquare, almost all checkins have a latitude and longitude, and for Flickr and Instagram most photos have a location associated with them. However, for Twitter, only around 1% of all tweets have geographic coordinates. But as we will see, even 1% still results in a whole lot of tweets!
After grabbing the social media data, we intersect it with the outlines of parks and open spaces in California, using polygons from the California Protected Areas Database maintained by GreenInfo Network. Everything that doesn’t intersect one of these parks, we throw away. The following maps represent the data as it looks before the filtering process.
But enough talking, let’s look at some maps!”
This is what happens when you give social networking to doctors
CArmel DeAmicis in PandoDaily: “Dr. Gregory Kurio will never forget the time he was called to the ER because a epileptic girl was brought in suffering a cardiac arrest of sorts (HIPAA mandates he doesn’t give out the specific details of the situation). In the briefing, he learned the name of her cardiac physician who he happened to know through the industry. He subsequently called the other doctor and asked him to send over any available information on the patient — latest meds, EKGs, recent checkups, etc.
The scene in the ER was, to be expected, one of chaos, with trainees and respiratory nurses running around grabbing machinery and meds. Crucial seconds were ticking past, and Dr. Kurio quickly realized the fax machine was not the best approach for receiving the records he needed. ER fax machines are often on the opposite of the emergency room, take awhile to print lengthy of records, frequently run out of paper, and aren’t always reliable – not exactly the sort of technology you want when a patient’s life or death hangs in the midst.
Email wasn’t an option either, because HIPAA mandates that sensitive patient files are only sent through secure channels. With precious little time to waste, Dr. Kurio decided to take a chance on a new technology service he had just signed up for — Doximity.
Doximity is a LinkedIn for Doctors of sorts. It has, as one feature, a secure e-fax system that turns faxes into digital messages and sends them to a user’s mobile device. Dr. Kurio gave the other physician his e-fax number, and a little bit of techno-magic happened.
….
With a third of the nation’s doctors on the platform, today Doximity announced a $54 million Series C from DFJ, T. Rowe Price Associates, Morgan Stanley, and existing investors. The funding news isn’t particularly important, in and of itself, aside from the fact that the company is attracting the attention of private market investors very early in its growth trajectory. But it’s a good opportunity to take a look at Doximity’s business model, how it mirrors the upwards growth of other vertical professional social networks (say that five times fast), and the way it’s transforming our healthcare providers’ jobs.
Doximity works, in many ways, just like LinkedIn. Doctors have profiles with pictures and their resume, and recruiters pay the company to message medical professionals. “If you think it’s hard to find a Ruby developer in San Francisco, try to find an emergency room physician in Indiana,” Doximity CEO Jeff Tangney says. One recruiter’s pain is a smart entrepreneur’s pleasure — a simple, straightforward monetization strategy.
But unlike LinkedIn, Doximity can dive much deeper on meeting doctors’ needs through specialized features like the e-fax system. It’s part of the reason Konstantin Guericke, one of LinkedIn’s “forgotten” co-founders, was attracted to the company and decided to join the board as an advisor. “In some ways, it’s a lot like LinkedIn,” Guericke says, when asked why he decided to help out. “But for me it’s the pleasure of focusing on a more narrow audience and making more of an impact on their life.”
In another such high-impact, specialized feature, doctors can access Doximity’s Google Alerts-like system for academic articles. They can sign up to receive notifications when stories are published about their obscure specialties. That means time-strapped physicians gain a more efficient way to stay up to date on all the latest research and information in their field. You can imagine that might impact the quality of the care they provide.
Lastly, Doximity offers a secure messaging system, allowing doctors to email one another regarding a fellow patient. Such communication is a thorny issue for doctors given HIPPA-related privacy requirements. There are limited ways to legally update say, a primary care physician when a specialist learns one of their patients has colon cancer. It turns into a big game of phone tag to relay what should be relatively straightforward information. Furthermore, leaving voicemails and sending faxes can result in details getting lost in what its an searchable system.
The platform is free for doctors, and it has attracted them quickly join in droves. Doximity co-founder and CEO Jeff Tangney estimates that last year the platform had added 15 to 16 percent of US doctors. But this year, the company claims it’s “on track to have half of US physicians as members by this summer.” Fairly impressive growth rate and market penetration.
With great market penetration comes great power. And dollars. Although the company is only monetizing through recruitment at the moment, the real money to be made with this service is through targeted advertising. Think about how much big pharma and medtech companies would be willing to cough up to to communicate at scale with the doctors who make purchasing decisions. Plus, this is an easy way for them to target industry thought leaders or professionals with certain specialties.
Doximity’s founders’ and investors’ eyes might be seeing dollar signs, but they haven’t rolled anything out yet on the advertising front. They’re wary and want to do so in a way that ads value to all parties while avoiding pissing off medical professionals. When they finally pul lthe trigger, however, it’s has the potential to be a Gold Rush.
Doximity isn’t the only company to have discovered there’s big money to be made in vertical professional social networks. As Pando has written, there’s a big trend in this regard. Spiceworks, the social network for IT professionals which claims to have a third of the world’s IT professionals on the site, just raised $57 million in a round led by none other than Goldman Sachs. Why does the firm have such faith in a free social network for IT pros — seemingly the most mundane and unprofitable of endeavors? Well, just like with doctor and pharma corps, IT companies are willing to shell out big to market their wares directly to such IT pros.
Although the monetization strategies differ from business to business, ResearchGate is building a similar community with a social network of scientists around the world, Edmodo is doing it with educators, GitHub with developers, GrabCAD for mechanical engineers. I’ve argued that such vertical professional social networks are a threat to LinkedIn, stealing business out from under it in large industry swaths. LinkedIn cofounder Konstantin Guericke disagrees.
“I don’t think it’s stealing revenue from them. Would it make sense for LinkedIn to add a profile subset about what insurance someone takes? That would just be clutter,” Guericke says. “It’s more going after an opportunity LinkedIn isn’t well positioned to capitalize on. They could do everything Doximity does, but they’d have to give up something else.”
All businesses come with their own challenges, and Doximity will certainly face its share of them as it scales. It has overcome the initial hurdle of achieving the network effects that come with penetrating the a large segment of the market. Next will come monetizing sensitively and continuing to protecting users — and patients’ — privacy.
There are plenty of data minefields to be had in a sector as closely regulated as healthcare, as fellow medical startup Practice Fusion recently found out. Doximity has to make sure its system for onboarding and verifying new doctors is airtight. The company has already encountered some instances of individuals trying to pose as medical professionals to get access to another’s records — specifically a former lover trying to chase down their ex-spouse’s STI tests. One blowup where the company approves someone they shouldn’t or hackers break into the system, and doctors could lose trust in the safety of the technology….”
Looking for the Needle in a Stack of Needles: Tracking Shadow Economic Activities in the Age of Big Data
Manju Bansal in MIT Technology Review: “The undocumented guys hanging out in the home-improvement-store parking lot looking for day labor, the neighborhood kids running a lemonade stand, and Al Qaeda terrorists plotting to do harm all have one thing in common: They operate in the underground economy, a shadowy zone where businesses, both legitimate and less so, transact in the currency of opportunity, away from traditional institutions and their watchful eyes.
One might think that this alternative economy is limited to markets that are low on the Transparency International rankings (such as sub-Saharan Africa and South Asia, for instance). However, a recent University of Wisconsin report estimates the value of the underground economy in the United States at about $2 trillion, about 15% of the total U.S. GDP. And a 2013 study coauthored by Friedrich Schneider, a noted authority on global shadow economies, estimated the European Union’s underground economy at more than 18% of GDP, or a whopping 2.1 trillion euros. More than two-thirds of the underground activity came from the most developed countries, including Germany, France, Italy, Spain, and the United Kingdom.
Underground economic activity is a multifaceted phenomenon, with implications across the board for national security, tax collections, public-sector services, and more. It includes the activity of any business that relies primarily on old-fashioned cash for most transactions — ranging from legitimate businesses (including lemonade stands) to drug cartels and organized crime.
Though it’s often soiled, heavy to lug around, and easy to lose to theft, cash is still king simply because it is so easy to hide from the authorities. With the help of the right bank or financial institution, “dirty” money can easily be laundered and come out looking fresh and clean, or at least legitimate. Case in point is the global bank HSBC, which agreed to pay U.S. regulators $1.9 billion in fines to settle charges of money laundering on behalf of Mexican drug cartels. According to a U.S. Senate subcommittee report, that process involved transferring $7 billion in cash from the bank’s branches in Mexico to those in the United States. Just for reference, each $100 bill weighs one gram, so to transfer $7 billion, HSBC had to physically transport 70 metric tons of cash across the U.S.-Mexican border.
The Financial Action Task Force, an intergovernmental body established in 1989, has estimated the total amount of money laundered worldwide to be around 2% to 5% of global GDP. Many of these transactions seem, at first glance, to be perfectly legitimate. Therein lies the conundrum for a banker or a government official: How do you identify, track, control, and, one hopes, prosecute money launderers, when they are hiding in plain sight and their business is couched in networked layers of perfectly defensible legitimacy?
Enter big-data tools, such as those provided by SynerScope, a Holland-based startup that is a member of the SAP Startup Focus program. This company’s solutions help unravel the complex networks hidden behind the layers of transactions and interactions.
Networks, good or bad, are near omnipresent in almost any form of organized human activity and particularly in banking and insurance. SynerScope takes data from both structured and unstructured data fields and transforms these into interactive computer visuals that display graphic patterns that humans can use to quickly make sense of information. Spotting of deviations in complex networked processes can easily be put to use in fraud detection for insurance, banking, e-commerce, and forensic accounting.
SynerScope’s approach to big-data business intelligence is centered on data-intense compute and visualization that extend the human “sense-making” capacity in much the same way that a telescope or microscope extends human vision.
To understand how SynerScope helps authorities track and halt money laundering, it’s important to understand how the networked laundering process works. It typically involves three stages.
1. In the initial, or placement, stage, launderers introduce their illegal profits into the financial system. This might be done by breaking up large amounts of cash into less-conspicuous smaller sums that are then deposited directly into a bank account, or by purchasing a series of monetary instruments (checks, money orders) that are then collected and deposited into accounts at other locations.
2. After the funds have entered the financial system, the launderer commences the second stage, called layering, which uses a series of conversions or transfers to distance the funds from their sources. The funds might be channeled through the purchase and sales of investment instruments, or the launderer might simply wire the funds through a series of accounts at various banks worldwide.
Such use of widely scattered accounts for laundering is especially prevalent in those jurisdictions that do not cooperate in anti-money-laundering investigations. Sometimes the launderer disguises the transfers as payments for goods or services.
3. Having successfully processed the criminal profits through the first two phases, the launderer then proceeds to the third stage, integration, in which the funds re-enter the legitimate economy. The launderer might invest the funds in real estate, luxury assets, or business ventures.
Current detection tools compare individual transactions against preset profiles and rules. Sophisticated criminals quickly learn how to make their illicit transactions look normal for such systems. As a result, rules and profiles need constant and costly updating.
But SynerScope’s flexible visual analysis uses a network angle to detect money laundering. It shows the structure of the entire network with data coming in from millions of transactions, a structure that launderers cannot control. With just a few mouse clicks, SynerScope’s relation and sequence views reveal structural interrelationships and interdependencies. When those patterns are mapped on a time scale, it becomes virtually impossible to hide abnormal flows.
SynerScope’s relation and sequence views reveal structural and temporal transaction patterns which make it virtually impossible to hide abnormal money flows.”
Using data to treat the sickest and most expensive patients
Dan Gorenstein for Marketplace (radio): “Driving to a big data conference a few weeks back, Dr. Jeffrey Brenner brought his compact SUV to a full stop – in the middle of a short highway entrance ramp in downtown Philadelphia…
Here’s what you need to know about Dr. Jeffrey Brenner: He really likes to figure out how things work. And he’s willing to go to extremes to do it – so far that he’s risking his health policy celebrity status.
Perhaps it’s not the smartest move from a guy who just last fall was named a MacArthur Genius, but this month, Brenner began to test his theory for treating some of the sickest and most expensive patients.
“We can actually take the sickest and most complicated patients, go to their bedside, go to their home, go with them to their appointments and help them for about 90 days and dramatically improve outcomes and reduce cost,” he says.
That’s the theory anyway. Like many ideas when it comes to treating the sickest patients, there’s little data to back up that it works.
Brenner’s willing to risk his reputation precisely because he’s not positive his approach for treating folks who cycle in and out of the healthcare system — “super-utilizers” — actually works.
“It’s really easy for me at this point having gotten a MacArthur award to simply declare what we do works and to drive this work forward without rigorously testing it,” Brenner said. “We are not going to do that,” he said. “We don’t think that’s the right thing to do. So we are going to do a randomized controlled trial on our work and prove whether it works and how well it works.”
Helping lower costs and improve care for the super-utilizers is one of the most pressing policy questions in healthcare today. And given its importance, there is a striking lack of data in the field.
People like to call randomized controlled trials (RCTs) the gold standard of scientific testing because two groups are randomly assigned – one gets the treatment, while the other doesn’t – and researchers closely monitor differences.
But a 2012 British Medical Journal article found over the last 25 years, a total of six RCTs have focused on care delivery for super-utilizers.
Perhaps it’s not the smartest move from a guy who just last fall was named a MacArthur Genius, but this month, Brenner began to test his theory for treating some of the sickest and most expensive patients.
“We can actually take the sickest and most complicated patients, go to their bedside, go to their home, go with them to their appointments and help them for about 90 days and dramatically improve outcomes and reduce cost,” he says.
That’s the theory anyway. Like many ideas when it comes to treating the sickest patients, there’s little data to back up that it works.
Brenner’s willing to risk his reputation precisely because he’s not positive his approach for treating folks who cycle in and out of the healthcare system — “super-utilizers” — actually works.
“It’s really easy for me at this point having gotten a MacArthur award to simply declare what we do works and to drive this work forward without rigorously testing it,” Brenner said. “We are not going to do that,” he said. “We don’t think that’s the right thing to do. So we are going to do a randomized controlled trial on our work and prove whether it works and how well it works.”
Helping lower costs and improve care for the super-utilizers is one of the most pressing policy questions in healthcare today. And given its importance, there is a striking lack of data in the field.
People like to call randomized controlled trials (RCTs) the gold standard of scientific testing because two groups are randomly assigned – one gets the treatment, while the other doesn’t – and researchers closely monitor differences.
But a 2012 British Medical Journal article found over the last 25 years, a total of six RCTs have focused on care delivery for super-utilizers.
…Every major health insurance company – Medicare and Medicaid, too – has spent billions on programs for super-utilizers. The absence of rigorous evidence raises the question: Is all this effort built on health policy quicksand?
Not being 100 percent sure can be dangerous, says Duke behavioral scientist Peter Ubel, particularly in healthcare.
Ubel said back in the 1980s and 90s doctors prescribed certain drugs for irregular heartbeats. The medication, he said, made those weird rhythms go away, leaving beautiful-looking EKGs.
“But no one had tested whether people receiving these drugs actually lived longer, and many people thought, ‘Why would you do that? We can look at their cardiogram and see that they’re getting better,’” Ubel said. “Finally when somebody put that evidence to the test of a randomized trial, it turned out that these drugs killed people.”
WellPoint’s Nussbaum said he hoped Brenner’s project would inspire others to follow his lead and insert data into the discussion.
“I believe more people should be bold in challenging the status quo of our delivery system,” Nussbaum said. “The Jeff Brenners of the world should be embraced. We should be advocating for them to take on these studies.”
So why aren’t more healthcare luminaries putting their brilliance to the test? There are a couple of reasons.
Harvard economist Kate Baicker said until now there have been few personal incentives pushing people.
“If you’re focused on branding and spreading your brand, you have no incentive to say, ‘How good is my brand after all?’” she said.
And Venrock healthcare venture capitalist Bob Kocher said no one would fault Brenner if he put his brand before science, an age-old practice in this business.
“Healthcare has benefitted from the fact that you don’t understand it. It’s a bit of an art, and it hasn’t been a science,” he said. “You made money in healthcare by putting a banner outside your building saying you are a top something without having to justify whether you really are top at whatever you do.”
Duke’s Ubel said it’s too easy – and frankly, wrong – to say the main reason doctors avoid these rigorous studies is because they’re afraid to lose money and status. He said doctors aren’t immune from the very human trap of being sure their own ideas are right.
He says psychologists call it confirmation bias.
“Everything you see is filtered through your hopes, your expectations and your pre-existing beliefs,” Ubel said. “And that’s why I might look at a grilled cheese sandwich and see a grilled cheese sandwich and you might see an image of Jesus,” he says.
Even with all these hurdles, MIT economist Amy Finkelstein – who is running the RCT with Brenner – sees change coming.
“Providers have a lot more incentive now than they use to,” she said. “They have much more skin in the game.”
Finkelstein said hospital readmission penalties and new ways to pay doctors are bringing market incentives that have long been missing.
Brenner said he accepts that the truth of what he’s doing in Camden may be messier than the myth.
Collective intelligence in crises
Buscher, Monika and Liegl, Michael in: Social collective intelligence. Computational Social Sciences Series: “New practices of social media use in emergency response seem to enable broader ‘situation awareness’ and new forms of crisis management. The scale and speed of innovation in this field engenders disruptive innovation or a reordering of social, political, economic practices of emergency response. By examining these dynamics with the concept of social collective intelligence, important opportunities and challenges can be examined. In this chapter we focus on socio-technical aspects of social collective intelligence in crises to discuss positive and negative frictions and avenues for innovation. Of particular interest are ways of bridging between collective intelligence in crises and official emergency response efforts.”
Cyberlibertarians’ Digital Deletion of the Left
David Golumbia in Jacobin: “The digital revolution, we are told everywhere today, produces democracy. It gives “power to the people” and dethrones authoritarians; it levels the playing field for distribution of information critical to political engagement; it destabilizes hierarchies, decentralizes what had been centralized, democratizes what was the domain of elites.
Most on the Left would endorse these ends. The widespread availability of tools whose uses are harmonious with leftist goals would, one might think, accompany broad advancement of those goals in some form. Yet the Left today is scattered, nearly toothless in most advanced democracies. If digital communication technology promotes leftist values, why has its spread coincided with such a stark decline in the Left’s political fortunes?
Part of this disconnect between advancing technology and a retreating left can be explained by the advent of cyberlibertarianism, a view that widespread computerization naturally produces democracy and freedom.
In the 1990s, UK media theorists Richard Barbrook and Andy Cameron, US journalist Paulina Borsook, and US philosopher of technology Langdon Winner introduced the term to describe a prominent worldview in Silicon Valley and digital culture generally; a related analysis can be found more recently in Stanford communication scholar Fred Turner’s work. While cyberlibertarianism can be defined as a general digital utopianism, summed up by a simple slogan like “computerization will set us free” or “computers provide the solution to any and all problems,” these writers note a specific political formation — one Winner describes as “ecstatic enthusiasm for electronically mediated forms of living with radical, right-wing libertarian ideas about the proper definition of freedom, social life, economics, and politics.”
There are overt libertarians who are also digital utopians — figures like Jimmy Wales, Eric Raymond, John Perry Barlow, Kevin Kelly, Peter Thiel, Elon Musk, Julian Assange, Dread Pirate Roberts, and Sergey Brin, and the members of the Technology Liberation Front who explicitly describe themselves as cyberlibertarians. But the term also describes a wider ideological formation in which people embrace digital utopianism as compatible or even identical with leftist politics opposed to neoliberalism.
In perhaps the most pointed form of cyberlibertarianism, computer expertise is seen as directly applicable to social questions. In The Cultural Logic of Computation, I argue that computational practices are intrinsically hierarchical and shaped by identification with power. To the extent that algorithmic forms of reason and social organization can be said to have an inherent politics, these have long been understood as compatible with political formations on the Right rather than the Left.
Yet today, “hacktivists” and other promoters of the liberatory nature of mass computerization are prominent political voices, despite their overall political commitments remaining quite unclear. They are championed by partisans of both the Right and the Left as if they obviously serve the political ends of each. One need only reflect on the leftist support for a project like Open Source software to notice the strange and under-examined convergence of the Right and Left around specifically digital practices whose underlying motivations are often explicitly libertarian. Open Source is a deliberate commercialization of Richard Stallman’s largely noncommercial notion ofFree Software (see Stallman himself on the distinction). Open Source is widely celebrated by libertarians and corporations, and was started by libertarian Eric Raymond and programmer Bruce Perens, with support from businessman and corporate sympathizer Tim O’Reilly. Today the term Open Source has wide currency as a political imperative outside the software development community, despite its place on the Right-Left spectrum being at best ambiguous, and at worst explicitly libertarian and pro-corporate.
When computers are involved, otherwise brilliant leftists who carefully examine the political commitments of most everyone they side with suddenly throw their lot in with libertarians — even when those libertarians explicitly disavow Left principles in their work…”
Handbook Of The International Political Economy Of Governance
New book edited by Anthony Payne, and Nicola Phillips: “Since the 1990s many of the assumptions that anchored the study of governance in international political economy (IPE) have been shaken loose. Reflecting on the intriguing and important processes of change that have occurred, and are occurring, Professors Anthony Payne and Nicola Phillips bring together the best research currently being undertaken in the field. They explore the complex ways that the global political economy is presently being governed, and indeed misgoverned. Covering all themes central to the field of politics, this extensive and detailed Handbook will be of great value to students of governance, political economy, international relations and development studies.”
Twitter Can Now Predict Crime, and This Raises Serious Questions
Motherboard: “Police departments in New York City may soon be using geo-tagged tweets to predict crime. It sounds like a far-fetched sci-fi scenario a la Minority Report, but when I contacted Dr. Matthew Greber, the University of Virginia researcher behind the technology, he explained that the system is far more mathematical than metaphysical.
The system Greber has devised is an amalgam of both old and new techniques. Currently, many police departments target hot spots for criminal activity based on actual occurrences of crime. This approach, called kernel density estimation (KDE), involves pairing a historical crime record with a geographic location and using a probability function to calculate the possibility of future crimes occurring in that area. While KDE is a serviceable approach to anticipating crime, it pales in comparison to the dynamism of Twitter’s real-time data stream, according to Dr. Gerber’s research paper “Predicting Crime Using Twitter and Kernel Density Estimation”.
Dr. Greber’s approach is similar to KDE, but deals in the ethereal realm of data and language, not paperwork. The system involves mapping the Twitter environment; much like how police currently map the physical environment with KDE. The big difference is that Greber is looking at what people are talking about in real time, as well as what they do after the fact, and seeing how well they match up. The algorithms look for certain language that is likely to indicate the imminent occurrence of a crime in the area, Greber says. “We might observe people talking about going out, getting drunk, going to bars, sporting events, and so on—we know that these sort of events correlate with crime, and that’s what the models are picking up on.”
Once this data is collected, the GPS tags in tweets allows Greber and his team to pin them to a virtual map and outline hot spots for potential crime. However, everyone who tweets about hitting the club later isn’t necessarily going to commit a crime. Greber tests the accuracy of his approach by comparing Twitter-based KDE predictions with traditional KDE predictions based on police data alone. The big question is, does it work? For Greber, the answer is a firm “sometimes.” “It helps for some, and it hurts for others,” he says.
According to the study’s results, Twitter-based KDE analysis yielded improvements in predictive accuracy over traditional KDE for stalking, criminal damage, and gambling. Arson, kidnapping, and intimidation, on the other hand, showed a decrease in accuracy from traditional KDE analysis. It’s not clear why these crimes are harder to predict using Twitter, but the study notes that the issue may lie with the kind of language used on Twitter, which is characterized by shorthand and informal language that can be difficult for algorithms to parse.
This kind of approach to high-tech crime prevention brings up the familiar debate over privacy and the use of users’ date for purposes they didn’t explicitly agree to. The case becomes especially sensitive when data will be used by police to track down criminals. On this point, though he acknowledges post-Snowden societal skepticism regarding data harvesting for state purposes, Greber is indifferent. “People sign up to have their tweets GPS tagged. It’s an opt-in thing, and if you don’t do it, your tweets won’t be collected in this way,” he says. “Twitter is a public service, and I think people are pretty aware of that.”…
The system Greber has devised is an amalgam of both old and new techniques. Currently, many police departments target hot spots for criminal activity based on actual occurrences of crime. This approach, called kernel density estimation (KDE), involves pairing a historical crime record with a geographic location and using a probability function to calculate the possibility of future crimes occurring in that area. While KDE is a serviceable approach to anticipating crime, it pales in comparison to the dynamism of Twitter’s real-time data stream, according to Dr. Gerber’s research paper “Predicting Crime Using Twitter and Kernel Density Estimation”.
Dr. Greber’s approach is similar to KDE, but deals in the ethereal realm of data and language, not paperwork. The system involves mapping the Twitter environment; much like how police currently map the physical environment with KDE. The big difference is that Greber is looking at what people are talking about in real time, as well as what they do after the fact, and seeing how well they match up. The algorithms look for certain language that is likely to indicate the imminent occurrence of a crime in the area, Greber says. “We might observe people talking about going out, getting drunk, going to bars, sporting events, and so on—we know that these sort of events correlate with crime, and that’s what the models are picking up on.”
Once this data is collected, the GPS tags in tweets allows Greber and his team to pin them to a virtual map and outline hot spots for potential crime. However, everyone who tweets about hitting the club later isn’t necessarily going to commit a crime. Greber tests the accuracy of his approach by comparing Twitter-based KDE predictions with traditional KDE predictions based on police data alone. The big question is, does it work? For Greber, the answer is a firm “sometimes.” “It helps for some, and it hurts for others,” he says.
According to the study’s results, Twitter-based KDE analysis yielded improvements in predictive accuracy over traditional KDE for stalking, criminal damage, and gambling. Arson, kidnapping, and intimidation, on the other hand, showed a decrease in accuracy from traditional KDE analysis. It’s not clear why these crimes are harder to predict using Twitter, but the study notes that the issue may lie with the kind of language used on Twitter, which is characterized by shorthand and informal language that can be difficult for algorithms to parse.
This kind of approach to high-tech crime prevention brings up the familiar debate over privacy and the use of users’ date for purposes they didn’t explicitly agree to. The case becomes especially sensitive when data will be used by police to track down criminals. On this point, though he acknowledges post-Snowden societal skepticism regarding data harvesting for state purposes, Greber is indifferent. “People sign up to have their tweets GPS tagged. It’s an opt-in thing, and if you don’t do it, your tweets won’t be collected in this way,” he says. “Twitter is a public service, and I think people are pretty aware of that.”…