Who's Watching You in 2020?
Who's following you? Who's framing you? Who's coming for your business?
Hi,
Welcome to Roll Call, a newsletter about the state of policy and technology in public safety, and the people shaping it. If you’d like to sign up, you can do so here. Or just read on…
Today’s note is a series of predictions for 2020. I know, how original. I haven’t seen anyone else do it for public safety though!
One thing working against me on this: I have no track record. Correction: I have no track record yet. I’ve never made predictions that aren’t tied to my role, so this will mark the first time where my predictions are judged on their own merit, outside the pre-set objectives of any one company.
What’s working for me is that I am going to be super transparent and data-driven as I make my predictions. I come from a statistics background and bad punditry makes me angry, so there will be ground rules. Only three rules. First, predictions have to be falsifiable. You will know at the end of the year whether they came true or not; there will be no fluffy takes where I can “well, ackshually” my way into being right no matter what happens. Second, there will be a likelihood assessed to each prediction. Putting a number on each will pay off in a few years when we can assess whether the 20% chance I assigned was close to a true 20%. Third is that, when possible, I will put my money where my mouth is. I’ll be working through a series of company profiles in 2020, and paid subscribers will be able to see what’s in my portfolio. Obviously it won’t be financial advice, but it’s important to put some level of skin in the game to show conviction.
So here we go:
TL;DR
Neighbor surveillance tech has a moment, survives backlash, and enters a period of uneasy peace.
Majority of major cities take a stance on facial recognition and license plate readers.
One major city commits fully to FirstNet as the provider of their entire public safety tech stack.
A company emerges that centers its value proposition on responder mental health.
This is the last big year of company consolidation in the public safety space.
Deepfakes move from political to petty.
Plus there’s a rapid-fire list of potential company exits and acquisitions.
Neighbor Surveillance Tech
The three big companies to pay attention to are Nextdoor, the Citizen app, and Amazon Ring + Neighbors. Each empower community residents to see and share events that are happening around them. They digitally extend the human instinct to protect themselves by sharing suspicious activity, listening to the police scanner, and recording who comes to the door. Completely natural. However, each has come under scrutiny for being examples of how tech companies can scale their influence without oversight, privacy questions, and for reflecting the worst of humanity: paranoia, biased stereotyping of minorities, and hardening the lines between community constituents.
So they’re bad? No. It’s complicated. This group is unique in a public safety context because they are (mostly) consumer tech companies. Even the most regulation-happy municipalities can’t remove private residents’ ability to set up cameras on their property, or to see where 911 calls are coming from in their neighborhood. And, to go a step further, I’d argue that it’d be negligent for public safety agencies to not invest time and energy on the digital platforms that residents use. There are some great reasons for agencies to follow the neighbors they protect onto any platform that strengthens community ties.
I believe there is danger that a company exerts too much power in public safety when they meet two criteria:
A company controls and stores a lot of crime and safety data without public sector oversight (where “a lot” is defined as multiple cities over multiple weeks).
A company’s consumer product is being distributed by public safety agencies to community residents.
I say it is dangerous, not that it is bad. Meeting these conditions does not mean the company is a malicious actor. Meeting both, however, is a prerequisite for enabling a scale of negative consequences not possible at other companies. Citizen meets the first criteria. Amazon meets both. Amazon Ring’s growing partnership with (currently 50+) law enforcement agencies is concerning because Amazon manages footage without retention policies, understands when and where agencies have requested footage from neighbors, and can change its terms of service at any time. Add on top an incentive structure for departments to try to have Ring installed on every porch and you have the makings of a certain dystopia. For now, the tradeoff is in favor of the consumer.
I predict a policy battle between Amazon and a progressive major city, or maybe the ACLU. Someone is going to slam Ring on the lack of data retention limits and privacy considerations (like who at the company can view user videos, etc.). Amazon will fight it quietly, but then publicly position themselves as a video privacy leader by imposing public-safety-esque retention policies and CJIS-esque security protocols on their Ring division.
I also predict that Tyler Tech will acquire the Citizen app. The number will be close to $75 million, which is roughly twice what Citizen raised. They paid $150 million for Socrata last year, and have an acquisition-heavy growth model. It’ll fit well with Socrata and CrimeReports.
Nextdoor will non-exclusively partner with some public safety companies, but their record so far of building and supervising community rules for reporting suspicious behavior is a indicator of where they want to land: as a micro-local platform for public safety agencies to engage with neighbors in a non-emergency context. Lots of growth for them in other verticals - they’ll want their crime and safety division to stay quiet.
Specific Predictions:
66% chance Amazon Ring comes under intense public scrutiny for distribution model, retention, and viewing permissions. If so, 80% chance they take steps to become “consumer privacy leaders” and impose public-safety-like restrictions on their use of crime/safety related footage. They will not impose restrictions on their distribution through public safety agencies, but some cities will limit agencies from working with Amazon at the distribution level.
80% chance Citizen is acquired. They’re closer to a feature than to a platform, which tees them up for Tyler Tech. If they aren’t bought it’ll be because they believe they’re worth more than Tyler can pay.
Further Reading:
Are Neighborhood Watch Apps Making Us Safer? by Hanna Kozlowska
My piece, The New Gotham, which references Stealing Amazon Packages in the Age of Nextdoor by Lauren Smiley
Spotlight: Neighborhood Crime Apps from Sarah Lustbader at The Appeal
Murder! Muggings! Mayhem! A Profile of Citizen CEO by Steven Bertoni
All the Crime, All the Time by John Herrman
Facial Recognition and ALPR Policy
In 2020, the policy matters more than the tech when it comes to facial recognition and ALPR (automated license plate readers). Four cities banned public sector use of facial recognition last year: San Francisco, Somerville (MA), Oakland, and Berkeley. I predict that every major city will have a facial recognition policy by the end of 2020, whether it prohibits, limits, or wholesale allows its use. The first major test in 2020 will likely be Portland (OR), where the Commissioner is proposing a ban on public and private sector use.
Axon’s move to publish the recommendations of its Ethics Board and shelve facial recognition for now will be confirmed as the right one. The distinction Axon raises between ‘detection’ and ‘recognition’ is important. Hopefully, municipalities that restrict AI in this area don’t overcompensate and restrict facial detection as well.
ALPR will be similarly treated. States will likely lead the way on ALPR legislation, as state police agencies (which patrol the interstate system) have a lot of static video infrastructure already installed. A Fairfax County (VA) judge last year banned passive use of license plate readers, meaning that agencies can only collect and store the data of vehicles under active investigation. There is no denying the value in ALPR for law enforcement agencies, and there will be significant variance in the policies decided across the country.
Specific Predictions:
90% chance that all major cities have facial recognition policies by end of 2020, whether their own or subject to state policy.
Same for ALPR, 90% chance that they have their own or it’s superseded by the state
Utility will find a chilly reception for its ALPR product unless it builds excellent permissioning systems around ‘active’ (BOLO-originated, etc.) and ‘passive’ use. I predict it will be set up appropriately and they lead the mobile ALPR race at the end of 2020 (80% chance).
One Big FirstNet Commit
FirstNet published their current app catalog in mid-December. It’s got a ton of stuff! Some small to mid-sized agency is going to look at the list of tools available and say: “gosh, if we just get the right LTE-enabled devices, we can do almost everything through FirstNet.” I predict an agency of around 100 sworn officers will move all communications (PTT, messaging), smartphone-enabled services (bodyworn video, barcode scanning), and situational awareness tools (data like NCIC, response coordination, mapping, and even dispatch) and trust a FirstNet ready device as the hub.
I don’t think many agencies are going to FirstNet for apps like Callyo, but if they can get a slick mobile version of a CAD like RapidDeploy Nimbus, it might be worth seeing what the savings look like to move as much as possible to a mobile-centric tech stack. We’ll dive deeper into this soon: I have a piece coming out in January that will break down the 4 strategies for outfitting an agency with tech.
Specific Predictions:
70% chance one or more agency, between 80 and 120 sworn, goes all-in on FirstNet.
Mental Health Co
An enterprising company will capitalize on the mental health zeitgeist in public safety. You can build something really valuable across personnel management (scheduling, training) and mental health (wellness, some early intervention re: use of force, connecting responders to mental health professionals, reporting requirements, billing/healthcare).
If this is interesting to you, reach out at mitchell@rollcall.media. I’ve done loads of research on HB2502 in Arizona, which increases the responsibility agencies have to take care of responders involved in traumatic incidents. Agencies are roughing it with Excel right now; it would be amazing to build the bridge between local PTSD counselors and public safety agencies to get responders the support they need. Seriously, reach out!
100% chance I respond.
Last Year of Consolidation
This is a two-year prediction. This year will be the last round of major consolidation in public safety. After this, it will be sink or swim. I think most company executives believe that the market environment will be different post-election and that it’s safer to make big moves now while you can (mostly) predict market sentiment. Public safety organization growth is enmeshed with financial market growth - if markets pull back, then public safety spending will follow. Agencies will take fewer risks on tools and platforms.
I predict that, in public safety, there will be more companies that fail than have strong exits in 2021 (75% chance).
Deepfakes
Deepfakes are photo-realistic audiovisual content, or the act of producing such content with the aid of deep learning. Because of some high-profile examples of celebrities and political figures, everyone expects deepfakes to be a political tool - and they will. But the scrutiny is extremely high, so journalists will spend a lot of effort on unearthing the high profile deepfakes that contribute to political instability.
What most people don't expect is that deepfakes go petty. Using deepfakes adversarially at a hyperlocal level is a threat that law enforcement agencies are not ready for. It took a long time for BWC video to reach wide acceptance by officers and the courts; it has taken only a couple years for video and audio manipulation tools to grow sophisticated enough to erase that trust in digital evidence. Fake is now indistinguishable from real. Simple enough to run on a Mac, ‘cheapfakes’ will be utilized at the local level. You or I could edit the audio of a traffic stop to make an officer appear verbally abusive, or generate a false video of your boss admitting to embezzling money. Or holding a missing kid in his basement. Or anything else; it doesn't matter.
The veracity of video evidence could become questionable, if this plays out. Bodyworn video evidence is borderline impregnable because the pathway from camera to cloud storage is well-walled. Chain of custody is tight. Citizen evidence portals change the game, however, by installing an unguarded portal (ha) into digital evidence management systems. How will Axon Citizen or Motorola TipSubmit ensure that the evidence they accept is real?
Specific Predictions:
There’s a 40% chance that we see deepfakes during the 2020 presidential race that initially gain traction on social media, but are quickly exposed as fakes.
There’s a 20% chance that there is a high-profile fraud where an enterprising individual scams a company into paying out more than $1 million by using an audio deepfake, similar to this story.
There’s a 5% chance that Axon, Motorola, or another digital evidence management company is put in the uncomfortable position of having to prove the originality of a piece of evidence that was submitted to them anonymously by a citizen.
Further Reading:
Lightning Round: Exits and Acquisitions
15% chance Vista Equity takes CentralSquare public. Would be absolutely massive.
30% chance Axon buys FileOnQ. A physical evidence management system enables Axon to be the point of truth over the entire evidentiary lifespan.
25% chance Amazon buys Vivint, which would expand Amazon’s sales/installation force for Ring and really launch them into the lead on home security.
60% chance someone buys Benchmark Analytics. Almost no one would surprise me.
Thanks for reading. If you enjoy this newsletter, please share it, forward it to your friends, or just sign up here.
-MA