A walking tour of surveillance infrastructure in Seattle

Note: this guide is a work in progress and may change at any time! We’ve done our best to cite our sources, but this page has not been professionally fact-checked.

This workshop was first run as part of two pilot workshops with the Tech Equity Coalition, in partnership with the ACLU of Washington, in October 2019. A zine based on this work was included at the CtrlZ.AI zine fair and the HOT MESS digital exhibition in 2020.

Introduction

In this tour of downtown Seattle, we’ll practice spotting some of the layers of the “smart” city that are hidden in plain sight, collecting and storing data about our lives, as well as the kinds of thinking that justify their existence. Each surveillance technology in our field guide includes the following categories to help you “spot” surveillance technology in the wild: Address, Appearance, What it does, How the tech works, Social importance, Discussion and finally, References.

Tour route

This is the route we will be taking on the walking tour. Click on each stop to pop up its location, and feel free to explore it on Google Maps, e.g. with Street View. The route spans 1.3 miles. Below, we outline each of the surveillance tools/sites listed above.

Surveillance cameras

Address: Practically everywhere, but the above example is at 523 Union St.

Surveillance camera hotspots (red = more likely)

Appearance: Poles, ledges, overhangs, rooftops. They are often spotted watching parking lots, doors, banks, intersections, and government buildings. Indoors, they are typically spotted on roofs and near cash registers.

Different types of surveillance cameras.

What it does: The camera has a memory. It can record video or other data and add it to a store of records over all time. The camera can be controlled remotely: it can swivel, zoom, or change height.

How the tech works: Camera recordings can be analyzed for patterns and shared with other entities, both private (your neighbors) and public (the local police).

It might be connected to a network (via Internet or radio frequency), which lets it send video to anywhere, receive instructions from anywhere, and lets other people, who might be anywhere, watch the video stream.

Discussion

What are other ways to question the need to have cameras, or surveillance, at all? What sort of society would we build around this way of life? What are your individual or communal experiences of “light shining more brightly on some than others”? What if each camera were replaced by a person? How would that change how you feel?

References

Street-level surveillance overview (EFF) Video surveillance system overview (ACLU) What’s wrong with public video surveillance? (ACLU)

Amazon Go

Address: 2131 7th Ave

Appearance: Looks like it could be any other convenience store… but it’s not! Inside, you must scan an app to enter, and there are no cashiers. Notice the gates where you must scan a QR code from your Amazon account to enter.

What it does: Amazon Go tracks your movement using overhead cameras to determine your browsing habits. Overhead cameras in the ceiling of the store, tracking consumers’ movements in the store.

How the tech works: Amazon can use your purchases to know more about you using patterns. For example, if you buy Hanukkah decorations, they might know you’re Jewish. Or certain foods might be correlated with certain health issues. They can combine your in-store purchases with your online Amazon purchases for even more predicting power.

What is Amazon doing with their knowledge about you? There’s no oversight or transparency. Your data could be sold to third parties without your consent.

Discussion

References

Facial recognition used at a convenience store in Seattle (another story) How much is your data worth? “No cash needed at this cafe. Students pay the tab with their personal data.”

Automated license plate reader

Address: 699 Spring Street Fig. 1. One automated license plate reader (we believe) perched over a highway onramp in downtown Seattle, here, the traffic entering the ramp onto the I-5 Express highway that cuts vertically through downtown Seattle.

Appearance: An automated license plate reader (ALPR) is a little camera that is either mounted to a pole (stationary) in high-traffic locations or the top of a police car (mobile) (Fig. 2.). Fig. 2. An ALPR can be mounted (usually high above major roads) or mobile (in/on police cars).

What it does: An ALPR photographs the license plate of every car that passes by and records the time and place of the encounter, as well as the plate number (Fig. 3.), and sends the information to a central storing place (called a database). Based on the information from an ALPR (e.g. “plate number ABC1234 detected at the intersection of Pike and Pine at 1:20 PM”), and the type of ALPR, a particular city agency may take an action. Fig. 3. An up-close view of a stationary ALPR.

There are three main kinds of ALPRs in Seattle. Stationary ones (type #1), owned by the Dept. of Transportation, are used for traffic purposes, to estimate travel time. Mobile ones, owned by the Seattle Police Dept., are used for parking enforcement (type #2) or law enforcement (type #3), to ping a police officer directly when a “wanted” license plate is spotted. These three kinds of ALPRs have different data retention periods; police ALPR data can be stored for up to 90 days, whereas other ALPR data is (supposedly) deleted immediately. In Seattle, the Seattle Department of Transportation has at least 99 stationary ALPRs deployed, and the Seattle Police Department has 19 vehicles with mounted ALPRs.

How to spot: ALPRs are usually mounted up high near high-traffic areas, like highways, downtown areas, interstates, and bridges. Maps of stationary ones are difficult to find because cities don’t want drivers knowing where they might be issued speeding tickets.

Regulations on ALPR use—both the technology and the data collected—are mostly nonexistent nationally, as well as in Seattle. That means that the agency that owns the system can choose whether and how they want to retain data, or track vehicle movements. Check out the map in Fig. 4: though SDOT says it does not track individual drivers’ movements, data from an ALPR system could easily be combined to do so. Fig. 4. Our guess at ALPR locations in Seattle. (This is not a definitive map of ALPR locations.) As a car moves, its location could be recorded by multiple ALPRs. (

Image: WSDOT Traffic GeoPortal)

Because of the lack of regulations, nationally, data-sharing is rampant with license plate data. According to EFF, many law enforcement agencies share plate data directly with each other, even across borders. ALPR data also makes it into private databases such as Thomson Reuters’s CLEAR, access to which can be bought by agencies and private corporations. In Seattle, SDOT and SPD say that they do not share data directly from ALPR systems, but it’s unclear what agencies might be able to access data with a request (per the two Seattle Surveillance Ordinance reports on ALPRs).

When it comes to ALPR data, beware of scope creep (Fig. 5): due to pervasive collection and data-sharing, your license plate could leave its original context and purpose and be used in ways you never consented to, such as private investigation or targeted advertising. Fig. 5. A diagram of scope creep: how your data, for example an ALPR sighting of your license plate, could leave its original context and purpose without your knowledge or consent.

How the tech works: ALPR is one of the older surveillance technologies; it was first invented and tested in the UK in 1984 to detect stolen cars. It uses a technique called optical character recognition (OCR), from a field called computer vision, to automatically make a guess at the letters and numbers in a picture of a license plate. This guess is probabilistic; i.e. it could be wrong. Database technologies allow all the information collected by ALPRs to be collected, and questions asked of it.

Interventions

In 2015, California and Minnesota passed strict laws placing limits on ALPR data-sharing. Minnesota also bars law enforcement from photographing a vehicle’s occupants. (Source: STOP)

Discussion

Of the three types of ALPRs, which ones do you think should be used in Seattle? Is the convenience of travel time estimates (e.g. WSDOT’s chart) or more efficient law enforcement worth the privacy leaks? What are less-invasive ways that we could achieve the same goals? How might ALPR use, and data collection, impact you collectively, as an innocent person who is not directly targeted by the state?

Further questions

What agencies have access to these systems? Which cities are using this or considering? What are the WA state rules regarding law enforcement using private sources of ALPR data?

Which tech companies are providing these systems and how much info do they keep (and what is it used for)?

(We are leaving answers to these questions out of our introductory writeup, but encourage you to find out the answers for your city. Thank you to Tech Fairness Coalition members for asking these questions!)

References

2018 Surveillance Impact Report — License Plate Readers — Seattle Department of Transportation 2018 Surveillance Impact Report — Automated License Plate Recognition (ALPR) (Patrol) — Seattle Police Department 2018 Surveillance Impact Report — Parking Enforcement Systems (Including ALPR) — Seattle Police Department EFF — Automated License Plate Readers Automated License Plate Readers & Law Enforcement — 2019 They Are Watching (ACLU) — Automatic License Plate Reader “How ICE Picks Its Targets in the Surveillance Age” (The New York Times, 2019) “How Britain Exported Next-Generation Surveillance” — on the invention of ALPRs (James Bridle, Medium, 2013) Data Driven: What We Learned (EFF study on ALPR data)

Acyclica

Address: Corners of Spring & 5th and Spring & 4th

Appearance: Flat black circles on top of traffic signal control boxes, which are large, gray or painted metal boxes, typically found at street corners.

What it does: The Acyclica device casts a fake Wi-Fi network and tracks phones that try to join the network in passing cars. Since each phone has a unique identifier (called a MAC address – like your Social Security Number, but for a device), different Acyclica installations can track your personal location as you pass them in the city.

How the tech works:  You know how your phone or laptop auto-connects to Wi-Fi networks?  To do this, your device is shouting to the world a ton of your personal information in something called a probe packet. A probe packet contains the MAC address as well as the list of all the past Wi-fi networks that your device has tried to join before, which can reveal a lot about you! (See Fig. 1.) Acyclica listens for these probe packets, and keeps track of the different places it has heard your MAC address to create a location history.