response

V ESSAY T H E E N D O F T R U S T U 0044

On April 22, 1990, I was one of the tens of the inds o people who packed into Toronto's Nathan Phill i quare in front of City Hall, to commemorate Earth Da he firsi one in twenty years. I was eighteen years old, as fai back as I could remember, I'd had an ambient a eness that there was an environmental crisis in the ofi from the PCBs m the water to the hole in the ozone layer. But as the day re on; as the speakers took to the stage, as the crowd buzzed and surged, I aiized that this was serious. It didn't matter how big and hopeless the cause as; the

ragile ecology of the only known planet capable of sustaining hum ; life ir he universe was at risk of permanent and irreversible collapse

when I became a climate bore. For years after, I tried to con; ceth people around me that we were hurting toward a crisis that would a: rsel

dleas Id eVTe '°Ved: * seas' killi"g winds. ; ,s c ^ migrati°n tTiS&5 '°0med in future, andby dm.

uZ Z Z T l° be Undeniabfe'il be too late.

voke intemarionaleffortrt fizzled Kyoto ^o en<^h p°httcal pressure .pro nonbindine Copenhagen, Paris. All unambous

But a funny thineh!.'' °f C°Urse'blown Past by al™st every signatory

assumed that I was exacrcx • °S people 1 had the Climate Talk wit!

accept the danger-they justd^ ̂ danger' Now' most People I talk witl

denial"-the moment at which the ™^ WG crossed the Point of "pea* change is that big of a deal ^ ̂ QT °f people who just don't think climate

science communication (movies m^T^3516 decline- Some combination ol consequences (floods hurrirsn ^ " Convenient Truth) and undeniable weather events) convinced a critirSi' r°UghtS' blizzards> and other extreme

1 he consequences are onlv °f people that the problem is urgent, c earer, scientific literacy becomes m m°re Unden*able as the science becomes s arpen tlieir explanations it's harrT^ Widespread'and the communicators

• s hard to imagine how the number of people

O Cory Do row • Peak Denial 0045

who ack! ige the reality of climate change could ever significantly decline.

From nc the number of people whose direct experience makes them

climate- c :e believers will only grow. This p: its a new tactical challenge for activists. It's no longer so import­

ant to cos e people that climate change is real-now we have to convince

them tha can and should do something about it. Rather than arguing about pre . is, now we're arguing about solutions, and specifically whether

solutions exist. Before it was a battle between truth and denial; now it's

a battle b en hope and nihilism. It's n climate change. Many of us went through an individual journey

similar to one with cigarettes: rationalizing a dangerous and antisocial habit by g on the doubt expensively sown by tobacco companies. After

all, toba; ;e greenhouse gases, does not manifest its consequences right away-th e-and-effect relationship is obscured by the decades intervening

between irst puff and the first tumor. And just as with climate change, by the tin: -. symptoms are unignorable, it's tempting to conclude it s too

late to qi light as well enjoy the time we have left."

Which brings me to privacy. Breaches of our privacy share many of the same regrettable characteristics

as climate change and cigarettes: your first unwise disclosure and your first

privacy breach are likely separated by a lot of time and space. It s hard to explain how any one disclosure will harm you, but it's a near-certainty that after a long run of unwise disclosures one of them will come back to bite > ou, and maybe very hard indeed. Disclosures are cumulative: the seemingly inno cent location-tagged photo you post today is merged with another database tomorrow, or next year, and links you to a place where a doctor was \\ritin& a methadone prescription or a controversial political figure was present.

Getting people to overshare is big business, and the companies that ben efit from public doubt about the risks of aggregating personal information

have funded expensive PR campaigns to convince people that the ris

overstated and the harms inevitable and bearable. But the fact is that collecting, storing, and analyzing as much pers

V ESSAY T H E E N D O F T R U S T 1 0046

information about the public as possible is dangerous business. Th ential for harm to the subjects of this commercial surveillance is real, .grows with each passing day as the database silos get bigger.

People who've had their personal information collected face a list of risks: blackmail, stalking, government or police overreach, workpi retali­ ation, and identity theft are at the top of that list, but it continues \ :id on (for example, some breach victims had their names forged on letr o the FCC opposing net neutrality).

What's more, the harms of privacy breaches don't fade over t they worsen. Maybe your account on a breached addiction-recovery site ied to an alias that isn't connected to your name today, but a decade later a erent site where you used the same alias gets breached, and some clevt erson uses both breaches to re-identify that formerly anonymous alias of rs.

Privacy is a team sport. The data from my breached accounts n com­ promise you, not me. For example, your messages to me might re J that you contemplated divorcing your spouse years before, a fact that is revealed when my data is breached, years after you'd both reconciled.

Add to that the effect on social progress. In living memory, it was a crime to be gay, to smoke marijuana, to be married to someone of another r ace. Those laws gave way after massive societal attitude shifts, and those shifts were the result of marginalized people being able to choose the time and manner to reveal their true selves to the people around them. One quiet conversation at a time, people who lived a double life were able to forge alliances with one another and with their friends and family. Take away people's right to choose how they reveal themselves and you take away their ability to recruit supporters to their cause. Today, someone you love has a secret they have never revealed to you. If we take away their ability to choose how they reveal themselves to the world, you may never learn

at secret. Your loved one may go to their grave sorrowing without your ever knowing why.

Which is all to say that every day that goes by auto-recruits more people to he burgeoning army that understands there is a real privacy problem. These

O / Doctorow • Peak Denial 0047

pe are being recruited thanks to their visceral experiences of privacy br ses, and they're angry and hurt.

t means our job is changing. We spent decades unsuccessfully sound- in- alarm about the coming privacy apocalypse to an indifferent horde, an ist like the environmentalists and the tobacco activists, we failed. The cu nt, dismal state of privacy and technology means that millions have co to harm and more harm is on the way. The only thing worse than all th iffering would be for it all to go to waste.

:e people who finally understand that there's a problem don't necessarily ur stand that there is a solution: we must undertake an overhaul of tech­ no v, law, social norms, and business that makes the future fit for human ha tion. The processes set in motion by the aggregation and retention of da ave unavoidable consequences, and we're not going to stop using email or ellite navigation or social media, so it will be tempting to conclude that the ca - is hopeless, and that the only thingto do is surrender to the inevitable.

h is is privacy nihilism, and it is the new front for privacy activists. Technology has v wen us unprecedented, ubiquitous surveillance, but it has also given us unprecedented strong encryption that makes evading surveillance more possible than at any time in history. We have witnessed the rise of digital monopolists who abuse us without fear of being punished by a competitive market, but we've also lived through the rise of digital tools that make it possible for ordinary people to organize themselves and demand strong, vigorously enforced antitrust rules. Our lawmakers foolishly demand bans on working cryptography, but they're also increasingly embarrassed and even politically destroyed by privacy breac and they're joining the chorus demanding better privacy rules.

The bad news is that convincing people to pay attention to harms that are a long way off is so hard that it may very well be impossible. But the good news is that convincing people that the disaster they re living t roug c

averted, through their good will and forceful deeds, is muc , muc a species, we may be tragically shortsighted, but we make up for ,t by being

so ferociously indomitable when disaster looms. It's looming now. Time to start fighting. •

A COMPENDIUM OF LAW ENFORCEMENT SURVEILLANCE TOOLS

0039

By Edward F. Loomis

• Facial Recognition Systems

| 0 GPS Trackers

G License Plate Readers

0 Drones

) Body Cameras

0 Cell Tower Simulators

0 Parallel Construction

With public attention focused on the federal government's warrantless surveillance, the state and local use of surveillance methods has largely been ignored by many citizens. Follow­ ing 9/11 and the subsequent creation of the Department of Homeland Security (DHS), many military-grade surveillance technologies have migrated through DHS grants into the hands of state and local police departments for the supposed purpose of thwarting terrorist acts. These high-tech tools are in everyday use by officers across the country, capturing enormous amounts of data on unaware citizens who come within their range. Woven throughout this issue are a few of those tools.

Mug shots of suspected criminals have proven useful in solving crimes since their use by the Pinkerton National Detective Agency in 1870. With the develop­ ment of digital comput­ ers, experiments began that automated identity recognition through distinguishing facial fea­ tures. In the years since 9/11, automated facial recognition systems greatly matured to meet the needs of U.S. troops to monitor local popula­ tions in Afghanistan and Iraq. The technology was soon adapted by the law enforcement community to moni­ tor popular events for security purposes. With the rapid development of digital cameras, cell phone cameras, cloud computing, social net­ works, and public post­ ing of tagged images, a database of millions of personal facial images can easily be created by applying available surveillance methods.

Facial recognition systems can examine the distinct features of

A Compendium of Law Enforcement Surveillance Tools

• Facial Recognition Systems O GPS Trackers O License Plate Readers O Drones O Body Cameras O Cell Tower Simulators n Parallel Construction

0040

each individual's face in photos, video, or a real­ time camera feed and match it against those in a facial database. Some systems score candidate matches and rank them in order of relative confirmation probability. However, accuracy of these auto­ mated systems varies widely depending on the race and gender of the subject. In tests con­ ducted by a researcher at the MIT Media Lab, identifications of white males were demon­ strated to be accurate 99 percent of the time, whereas the failure rate on dark-skinned females approached 35 percent.

As arrested subjects are photographed by local, state, and federal police, their mug shots are stored in that juris­ diction's database for use in future automated criminal searches. Where cooperative sharing agreements exist, such as in the San Diego County region, where a consortium of eighty-two local, state, and federal law

enforcement agencies shares access to the Automated Regional Justice Information System (ARJIS), photo­ graphs may be queried by member jurisdictions to check traffic-stop subjects for outstanding warrants or to support criminal investigations.

Beginning in 2007, ARJIS received Depart­ ment of Justice grant funding to pursue devel­ opment of a mobile, regionally shared facial recognition prototype named the Tactical Identification System (TACIDS). Since 2013, the San Diego County Sheriff's Department has been using hand­ held tablets that connect to the TACIDS facial database. Without asking an arrest subject his or her name, an officer can capture the suspect's image with a tablet or smartphone and upload the photo to the TACIDS database. The system returns pos­ itive matches from its over 1.3 million booking photos of prior arrestees for a virtual lineup.

One can imagine the privacy implications of a massive virtual lineup database supporting an automated facial recog­ nition system. According to data compiled in 2016 by the Georgetown Law Center on Privacy and Technology, at least twenty-six states "allow law enforcement to run or request searches against their databases of driver's license and ID photos." Additionally, sixteen states permit FBI use of facial recognition technology to "compare the faces of suspected criminals with their driver's license and ID photos."

Most troubling is the fact that no state was found to have laws that comprehen­ sively regulate the use of facial recognition systems. The study described that the Pinellas County Sher­ iff's Office "runs 8,000 monthly searches on the faces of seven million Florida drivers—without requiring that officers have even a reasonable suspicion before running

0 Edward F. Loomis 0041

V DATA

1/2 One in two American adults (over 117 million) is in a face recognition network run by law enforcement.

30% MIT testing of three commercially-released facial recognition systems found a 30% lower success rate for "darker females" than "lighter males."

a search." The study found that of fifty-two agencies reporting the use of facial recogni­ tion systems, only the Ohio Bureau of Crim­ inal Investigation had policies prohibiting its officers from using them to track persons engaging in protected free-speech activities. Of those fifty-two agen­ cies, only four had pub­ licly available policies on the use of facial recog­ nition systems, and only nine claimed to log and audit officer searches for improper use.

For those who have personally been mis­ taken for someone else by a stranger, the fact that facial recognition software might errone­ ously select us from a growing national lineup database can be very unsettling. When South Wales police activated a facial recognition system against 170,000 attendees at a soccer match between Real Madrid and Juventus in Cardiff in June 2017, the cameras identified 2,470 people as criminals. Of

Source: Center on Privacy & Technology at Georgetown Law; MIT Media Lab.

the potential criminals identified, 2,297 were wrongly labeled by the facial recognition software—a 92 percent false positive rate.

In FBI testimony pre­ sented before the House Committee on Over­ sight and Government Reform in March 2017 on the FBI's use of facial recognition technology, an admission was made that the technology is used only as an inves­ tigative lead, not as a means of positive identi­ fication. The reason for that became obvious when the facial recog­ nition system's accu­ racy was later cited as having been tested and verified only to return "the correct candidate a minimum of 85 percent of the time within the top 50 candidates." •