January 18, 2017

GIS Analysis as a Research Communication Tool

The power of geospatial analysis lies in the new ways it provides to look at datasets and the relations among them. It allows you to explore more nuanced questions and discover correlations previously hidden. Used properly, geographic information system (GIS) tools can increase the saliency of a policy issue by expressing your argument visually and often much more effectively. Below is my recent experience in using GIS tools to broaden the audience for my research.

Property Assessment Disparities

Municipalities across the country are under fiscal duress due to cuts in state/federal aid, property tax levy limits, and rising employee fringe benefit costs. Often limited in their ability to generate new revenue streams, municipalities have become overly dependent on property taxes to “keep the lights on”.

Taxes are always a contentious issue and nobody wants to pay more than their fair share. To get a sense of how equitable the property tax burden was in Milwaukee, a city wrestling with all of the challenges noted above, I analyzed 33,000 property sales transactions over a 10-year period and compared them with their corresponding assessment values. By regressing the assessment value/sales price ratio on a host of predictors including building condition, lot size, geographic location, etc., I was able to get a sense of how equitable the city’s property taxation system is. While the findings presented an interesting disparity in who was paying their fair share, the results were neither accessible to the average citizen nor actionable for the policy maker. They required an understanding of my model specification and an ability to interpret coefficients expressed in terms of log odds. [Read more…]

NYC to Collect GPS Data on Car Service Passengers—Good Intentions Gone Awry or Something Else?

During the holiday season, New York City through its Taxi & Limousine Commission (the “TLC”) proposed a new rule expanding data reporting obligations for car service platform companies including Uber and Lyft. If the rule is adopted, car services will now have to report the GPS coordinates of both passenger pick-up and drop-off locations to the city government. Under NY’s Freedom of Information Law, that data in bulk will also be subject to full public release.

This proposal is either a classic case of good intentions gone awry or a clandestine effort to track millions of car service riders while riding roughshod over passenger privacy.

The stated justification for the new rule is to combat “driver fatigue” and improve car service safety. While the goal is laudable and important, the proposed data collection does not match the purpose and makes no sense. Does anyone really think GPS data measures a driver’s hours on the job or is relevant for the calculation of a trip’s duration? If the data collection were really designed to address driver fatigue, then the relevant data would be shift length (driver start/stop times, ride durations, possibly trip origination), not pick up/drop off locations.

The reporting, though, of this GPS data to the city government poses a real and serious threat to passenger privacy. The ride patterns can be mined to identify specific individuals and where they travel. In 2014, for instance, The Guardian reported that the TLC released anonymized taxi ride data that was readily reverse engineered to identify drivers. A 2015 paper shows that mobility patterns can also be used to identify gender and ethnicity. Numerous examples—from the Netflix release of subscriber film ratings  that were reverse engineered to identify subscribers to the re-identification of patients from supposedly anonymous health records—show that bulk data can often be identified to specific individuals. Disturbingly, the TLC proposal only makes one innocuous reference to protecting “privacy and confidentiality” and yet includes neither any privacy safeguards against identification of individual passengers from ride patterns nor any exemption from the NY State Freedom of Information Law.

If this weren’t worrisome enough for privacy, here’s the flashing red light. The TLC proposal mentions in passing that the data might be useful for “other enforcement actions.” But, the examples given for “other enforcement actions” do not map to the data being collected. For instance, the proposal says the GPS data “will facilitate investigating passenger complaints or complaints from a pedestrian or other motorist about unsafe driving, including for incidents alleged to have occurred during or between trips, by allowing TLC to determine the location of a vehicle at a particular time.” The pick-up and drop-off locations will not work for this goal. Likewise, the proposal says that “[b]y understanding when for-hire trips to and from the airports occur TLC can better target resources to ensure that passengers are picked up at the airport only by drivers authorized to do so.” This too is a strange justification to collect individual passenger records for every ride throughout the city! This goal would be satisfied much more effectively by seeking aggregate drop-off data for the particular areas of concern to the TLC.

This vague enforcement language and the mismatch between the proposal and the articulated goals strongly suggests that the rule may be a smokescreen for a new mass surveillance program of individuals traveling within New York City. Only two years ago, the NY Police Department was caught deploying a controversial program to track cars throughout the city using EZ Pass readers on traffic lights. This proposed new rule looks like a surreptitious expansion of that program to car service passengers. The TLC rule, if adopted, would provide a surveillance data trove that makes an end run around judicial oversight, subpoenas, and warrants.

It’s time to put the brakes on the city’s collection of trip location data for car service rides.

Disrupting The Business Model of the Fake News Industry

By Katherine Haenschen & Paul Ellenbogen 

In the aftermath of the 2016 election, researchers and media professionals alike seized on the vast proliferation of so-called “Fake News” on Facebook as a cause for concern. An informed citizenry is a necessary condition for democracy, so it is far from ideal to have millions of people consuming intentionally misleading information masquerading as hard news. Now that Facebook has admitted that it has a problem with Fake News, Mark Zuckerberg and Co. need to do even more to prevent its spread on the platform. We propose one solution: Facebook should block advertising links to Fake News websites and Fake News pages on the Facebook platform itself.

When we talk about Fake News, we’re referring to websites that intentionally and knowingly publish factually untrue content intended to masquerade as traditional “hard news.”  Individuals may choose to publish Fake News for political reasons, such as seeking to impact voting decisions. However, there is also a profit motive behind Fake News: publishers can make big money from advertising revenue that results from traffic to their sites. The Fake News business model utilizes Facebook’s paid features to gain readers and build audiences. Facebook offers a variety of advertising options for individuals looking to reach its 156 million users in the United States, including newsfeed and sidebar ads and promoted posts from pages. It is Facebook’s advertising features that should be rendered unavailable for the paid promotion of Fake News to users.

There is precedent for calling on Facebook to block Fake News from being advertised directly to its users: Facebook already bans certain kinds of ads on the platform such as those promoting dietary supplements and “controversial content.” Additionally, Facebook announced that it will stop placing ads on third-party Fake News websites. Now, we are calling on Facebook to ban Fake News from being advertised and promoted on the Facebook platform itself. Facebook should apply this type of ban even if it hurts Facebook’s own revenue.

There’s Big Money In Fake News

Media stories about Fake News producers emphasize the tremendous profits to be made in publishing knowingly false information and helping it to “go viral” on Facebook. Display ad revenue generated by Fake News sites can reach $10,000 to $30,000 per month. One Macedonian teen who publishes Fake News sites told Buzzfeed that advertising revenue could reach thousands of dollars per day or week. And though Google and Facebook have blocked the sites from their third-party advertising platforms, the Fake News publishers also note that there is no shortage of advertising networks willing to display ads on their websites. The reach of these articles amounts to millions of Facebook shares and clicks to the website, in turn generating millions of ad impressions, according to Buzzfeed.

The same articles about Fake News sites indicate that publishers are not bothered by the potential impact of sharing incorrect information. A Macedonian teen who operated multiple sites admitted that his content was “bad, false, and misleading,” and that he was motivated by the advertising revenue generated by his Fake News site. Therefore, if Facebook wants to curtail the proliferation of Fake News on its website, it should disrupt its business model using tools that are already at the platform’s disposal.

The Fake News Business Model

Multiple news articles have referenced Fake News producers’ use of Facebook advertising features to promulgate their posts. Here’s how the business model for Fake News works, with each step in the process illustrated by the diagram below:

  1. An individual publishes false information on a Fake News website, then pays to advertise a link to the post in Facebook users’ newsfeeds.
  2. Facebook profits from advertising on its platform, earning money for every person who clicks the link or every 1,000 users who see the ads.
  3. Facebook users click on the advertised links and go to the Fake News website, generating an impression for each display ad on the website.
  4. The Fake News site earns revenue from the resulting advertising impressions, which amount to millions of page views and tens of thousands of dollars per month.

Publishers must have a Facebook page to run newsfeed ads. As such, an ancillary cycle exists in which Fake News Publishers can promote this page to gain fans and organic traffic.

  1. Fake News producers advertise their Page to fans, growing an organic Facebook audience to whom they can share links at no cost.
  2. Fans can share these links to their own Facebook networks, furthering the organic reach of Fake News. This is how something “goes viral.”

As long as the cost of the Facebook ads that promote the posts is lower than the display ad revenue from the resulting clicks, the business model above will generate net income for the Fake News producer.

Cut Off Paid Features for Fake News

Our solution is simple: Facebook needs to deny the use of paid features by pages that promote Fake News. This means Fake News pages should not be able to run newsfeed and sidebar ads, promote page posts, or market their Facebook page to gain fans. Furthermore, Facebook should block any third-party attempts to advertise links to Fake News sites. Currently, any individual with a Facebook account can create a public page and use it to run ads for Fake News stories in Facebook users’ newsfeeds. Banning all advertising links to Fake News sites would prevent publishers from setting up new and deceptive Facebook pages for the purpose of advertising.

Facebook has already taken action to limit its role in directly funding Fake News. The platform cut off advertising on — but not leading to — Fake News websites, as has Google’s AdSense network. However, the Facebook advertising platform can still be used to drive traffic to these sites and fuel the cycle detailed above. Even if it does ban the use of advertising outbound links to Fake News sites, Facebook will still need to grapple with the size of Fake News pages, some of which surpass 700,000 fans and have a tremendous potential for organic reach. Our purpose is not to weigh in on that argument, but simply to point out a simple step Facebook can take that is consistent with its external ad placement on third-party sites.

Facebook Already Bans Certain Advertisers

Furthermore, a ban on the use of Facebook’s advertising features by Fake News sites would be in keeping with existing rules pertaining to the kinds of advertisers that can use the platform to reach users. For example, Facebook restricts the advertising of unsafe supplements at its “sole discretion,” such as various diet aids and performance-enhancing substances. Other prohibited content includes “controversial content,” which is defined as “…content that exploits controversial political or social issues for commercial purposes.”

Given that Fake News producers are open about their profit motivations, their use of Facebook advertisements to drive traffic should be considered a commercial purpose rather than a political purpose. As such, Facebook should use its existing rules to draw a line between political content and commercial content. If it fails to do so, unscrupulous individuals could start dressing up their questionable advertisements as political speech — Donald Trump Diet Pills, anyone? They’ll make your waistline great again!

Distinguishing between Fake News and news from reputable outlets is something that Facebook is already committed to doing now that it has pledged to pull advertisements from Fake News sites in the Facebook Audience Network program. Facebook could use the same criteria that it uses in the Audience Network on its advertising platform. While we are not proposing a heuristic to determine what is Fake News and what is merely an opinion piece devoid of factual content, we suggest that Facebook apply the same rules for banned third-party sites to advertisements on the platform for those very same sites.

Fight Fake News, Or Else Everyone Gets Played

It isn’t clear how Facebook’s long-term interests are served by enabling Fake News to market to its users, essentially creating a back door around its own advertising policies. Facebook makes money from advertisements for Fake News, but in the long term it may come to hurt Facebook, with suspicion and lost goodwill outweighing earnings from this category of advertisements. If Facebook chooses to regulate Fake News as political speech, Zuckerberg et al. are setting themselves up to be useful idiots for websites trying to make a quick buck off sensationalist and false stories.

As for the users, they are being intentionally misled with incorrect articles about political actors, which have the potential to impact issue awareness and candidate choice. At worst, people are basing their vote on misinformation-for-profit. At best, users may be getting quick entertainment out of these links (if they recognize them as false), but for the most part it seems like the Fake News operators are getting the benefit of the arrangement. Removing paid advertisements for these sites from users’ Facebook newsfeeds is not going to negatively impact their lives. Furthermore, these individuals remain free to like the Facebook pages for Fake News sites and share their posts organically with friends.

We are merely proposing that Facebook cut off the use of its paid features to promote links to Fake News to wider audiences, in accordance with its existing advertising policies. Advertisements for Fake News should be regulated like ads for “controversial content” and dietary supplements. This would cut off one stream of revenue for these Fake News websites, forcing them to gain traffic from Facebook entirely through organic reach. Failure to ban this type of advertising would suggest that Facebook values its own revenue over the need to curtail bad actors who are using its platform to intentionally spread misinformation harmful to our democratic society.