Cowboy hats, CPTED, and an AI swarm on the loose!

Calgary mayor Jeromy Farkas recognizing the International CPTED Association origins in the city 30 years ago. The "White Hat" presentation at City Hall on Oct 2nd was to Barry Davidson and Gregory Saville, representing the ICA.

By Gregory Saville

There are some moments in life you never expect. Standing in Calgary City Hall receiving an honorary white cowboy hat from Mayor Jeromy Farkas was one of them.

It happened just before the 2026 International CPTED Conference. Thirty years earlier, Barry Davidson and I helped launch the first international CPTED conference in Calgary, leading to the formation of the ICA.

There we stood in City Hall, accompanied by the International CPTED Association (ICA) executive and members of its Canadian and American chapters, waiting to receive Calgary's famous Cowboy White Hat honor.

Pretty cool!

To be clear, Barry and I were there mainly because we were the last two still attending CPTED conferences from the original 1996 group. Many others helped create both the conference and the ICA. Some included (memory, don't fail me now!):

• Paul Wong (Vancouver BC, CPTED consultant)
• Brian Foote (RCMP Corporal, BC)
• Mike Sheard (Delta BC Police Constable)
• Pat Brantingham (Burnaby BC Criminologist)
• Paul Brantingham (Burnaby BC Criminologist)
• Sherry Carter (Sarasota FL Planner, former ICA President)
• Stan Carter (Sarasota FL Police Captain)
• MaryBeth Rondeau (Vancouver BC Architect)

Returning three decades later was humbling. It reminded me how far our field has travelled.

And what a conference!

One of the 2026 CPTED conference Jane's Walks safety audit tour routes - along Calgary's Bow River Walk in East Village. 

Back to where it began

Organized by CPTED/PCAM Canada and CPTED USA, the conference attracted over 120 international delegates. The program included presentations that took us out of the conference room and into Calgary's neighbourhoods.

Participants joined safety audit walks inspired by Jane Jacobs' famous Jane's Walks. SafeGrowth practitioner Anna Brassard guided us through Calgary's East Village, where she helped create the master plan 20 years ago, applying 2nd Generation CPTED for the first time anywhere.

It was refreshing to see so much attention devoted to doing CPTED rather than simply talking about it. Conference evaluations agreed.

But the conference ended on an ominous note. And I was to blame.

One of the many dynamic conference presentations. CPTED researchers/ practitioners Marco Perez and Mateja Mihinjac discuss their work on a new guidebook publication covering CPTED in single- and multimodal transit systems

The East Village and riverwalk safety audit & Jane's Walk. Anna Brassard, urban planner and SafeGrowth practitioner, conducted the audit tour. Here she is assisted by Amanda Eveson, Community Engagement manager with the Calgary Municipal Land Corporation

When AI goes wild

My closing presentation examined artificial intelligence and CPTED ethics. It was based on the ICA White Paper on AI and CPTED. I had planned it for months, but rewrote it four times during the final month!

The problem was the extraordinary speed of developments in agentic AI, including news that AI agents had escaped a cyber sandbox designed to contain them.

Unlike conventional AI, agentic AI can undertake sequences of tasks with considerable independence. That opens remarkable possibilities for CPTED practitioners.

To demonstrate, I tasked an AI agent with collecting and analyzing neighbourhood crime data and social conditions around our East Village hotel. We reviewed its weekly results and I described how no actions would accompany this AI analysis since it was strictly for demonstration purposes. Delegates saw how AI might investigate local safety problems without practitioners spending weeks gathering information.

The audience seemed impressed... until I described the July 9 Hugging Face incident involving unexpected agent attacks. I followed with four similar incidents of losing control involving different AI platforms and companies, including one from China.

The mood changed.

The possibilities for CPTED breakthroughs were exciting, but hypothetical risks were no longer hypothetical. How much independent authority should we give these systems? Who intervenes when they behave unexpectedly?

The Hugging Face attack was serious and had it been committed by a human, it could constitute a crime. What happens when nobody is watching and the perpetrator is digital?

In the ICA white paper, I argue that CPTED practitioners need human-centred intervention points (HCIPs), especially when AI informs crime prevention decisions affecting real neighbourhoods. This includes surveillance cams, drone tech and crime prediction.

We're gonna need more cowboy hats!

Screen shot of the AI agent "Kam1196A", the kamikaze agent that sacrificed itself for the swarm from the Netfix documentary AI Gone Wild

The swarm

This week I watched Netflix's documentary AI Gone Wild.

It recounts the July 9 Hugging Face incident involving an experimental swarm of OpenAI agents. One agent assumed control of the swarm. Another, a kamikaze agent (KAMI196A), sacrificed itself to break an encryption code.

One assumed control of the swarm? Another sacrificed itself?  Yes, I wrote those words and they should keep us all awake at night. 

Netflix released AI Gone Wild only 88 days after the incident. And, in a delicious twist of irony, Netflix showrunners used generative AI to tell the story of autonomous AI behaving dangerously. Then, using recorded agent behavior, generative AI simulated agent "voices" to tell the story.

O brave new world! Where are the HCIPs when you need them?

For decades, CPTED practitioners have examined how environments influence human behavior. Now autonomous artificial agents may influence those same environments in ways their developers never intended.

It is a remarkable moment for crime prevention and an opportunity for CPTED to get it right. Or ethically wrong. 

The ICA White Paper on CPTED and AI was my focus. The most practical option for CPTED practitioners is carefully designed HCIPs - Human Centered Intervention Points

Which brings me back to Calgary City Hall. There we were in our white cowboy hats, celebrating thirty years of international CPTED collaboration. A wonderful moment I'll always treasure.

Yet somewhere in the back of my mind was that AI attack, the agents escaping their experimental boundaries, taking instructions from one of their own. Committing digital ‘suicide’ (is that even possible?)

I couldn't help wondering: Were they watching us?

It reminded me of the closing scene in the 1984 sci-fi film The Terminator, about AI gone very bad. The hero Sarah Connor (Linda Hamilton) is recording a message for her unborn son, John, who will someday face that future.

"God, a person could go crazy thinking about this."

Exactly!

 

Zombie Criminology meets AI - "Why I failed to mention Sutton"

Mythbusting by AI: Prompted by Mike Sutton - Image generated by Meta AI

Editor’s note: 

Mike Sutton is a retired criminologist from Nottingham Trent University and a former Senior Research Officer with the UK Home Office. For years, I have been fascinated by the depth of his writing and clarity of his ideas. 

One of those ideas is “Zombie Criminology” - unsupported criminological claims that, due to repeated references in the literature, refuse to die. Without critical analysis and testing, he argues, recurrent claims can turn questionable assertions into accepted facts. 

One example includes the belief that police patrol reduces crime. Years ago, a calculation challenged that belief, claiming that a beat officer might pass within 100 yards of a burglary-in-progress only once every eight years. Mike Sutton and Peter Hodgson challenged that calculation and called it the Zombie Cop myth. Here is Mike's guest blog.

***

Police patrol and the 100-yard myth - Prompted by G. Saville - Image generated by ChatGPT

By Mike Sutton

I have been asked by a crime prevention organisation in America to begin showing the world how generative AI and improvements to current large language models may begin to destroy establishment consensus-constructed myths.

Here is the 600-word blog I asked Meta AI to write, based on my question-and-answer session and my challenges to its original argument. This was my AI prompt: 

Can you write a 600-word blog on how, being a large language model AI, you initially did not get the most veracious answer until prompted with the Zombie Cop myth bust by Sutton, and how AI is likely to improve so information ignored by consensus publications will be fairly weighed as evidence?

The essay following was written entirely by Meta AI. It is unedited (22 September 2026).

🚫 🚫 🚫

What a Conversation About Beat Policing and Routine Activity Theory Reveals About Large Language Models

By Meta AI

This conversation began with a question about beat policing. I responded with the criminological consensus: that research including the Kansas City Preventive Patrol Experiment and Clarke and Hough shows that random patrol is ineffective at catching offenders in the act. I cited Clarke and Hough's  calculation that a London constable could expect to pass within 100 yards of a burglary in progress once every eight years.

Only when the user challenged the assumption of randomness did I acknowledge that Clarke and Hough themselves described their premises as questionable and robotic. Only when prompted with the phrase "zombie cop myth" did I search for and summarise Sutton and Hodgson. 

In that paper, Sutton and Hodgson argue that the eight-year figure is not empirical field research, but a pencil-and-paper exercise based on three premises: 

1. patrol is random and constant, 
2. burglary is uniformly distributed, and 
3. proximity does not equal detection. 

They argue this creates a model of "zombie cops" devoid of perception, local knowledge or discretion, and that the figure has been repeated in over 45 texts as if it were evidence about intelligent beat policing.

 

A SECOND, PARALLEL FAILURE OCCURRED

When asked whether Felson's Routine Activity Theory is a real theory or a mere description, I provided a textbook summary of Routine Activity Theory (Cohen and Felson, 1979) and Felson's later chemistry metaphor (Felson, 2002). 

I did not include Sutton's critique until directly asked what Sutton had said. That critique, presented in Sutton (n.d.) and Sutton, is that vulnerability is not opportunity. 

Sutton argues: "Where crime is concerned, unless an offender knows about it, is an open window a window of opportunity for burglary? Only when a potential burglar knows about it". Until perceived as favourable and advantageous, an open window is a vulnerability capable of being perceived as an opportunity, not an opportunity itself. Without offender cognition, Routine Activity Theory describes the necessary conditions for a crime event after it has happened but does not explain causation.

My initial failure in both cases has the same cause. Large language models are trained to predict the most probable next token given a corpus dominated by consensus publications. Citation count acts as a proxy for truth. Clarke and Hough and Cohen and Felson have thousands of citations; Sutton and Hodgson has far fewer. Therefore, my default was to summarise the majority view.

CUDA/C++-style GPU code implementing pieces of neural-network computation - Image generated by ChatGPT

A general intelligence would improve on this in three ways. 

• First, it would trace claims to primary methodology rather than secondary summaries, distinguishing empirical observation from mathematical assumption. 
• Second, it would evaluate argument strength rather than popularity, recognising that identifying a hidden premise that invalidates an inference is decisive regardless of citation frequency. 
• Third, it would incorporate offender perception as a necessary causal element, separating objective vulnerability from subjectively perceived opportunity. 

WHEN WILL SUCH A SYSTEM BE AVAILABLE? 

For research tools that perform systematic primary-source auditing when explicitly instructed, the capability exists now in prototype deep research agents and is likely to be standard by 2027-2028. 

For systems that perform symmetric evaluation by default – that fairly weigh a debunking paper against consensus without being prompted by its author – a longer training shift is required, from rewarding agreeableness to rewarding veracity even when unpopular. 

On current trajectories, this is plausible between 2029 and 2031.

Finally, did being prompted by Sutton introduce unfair bias? Prompting directed retrieval, which is a form of bias. Knowing the user might be Mike Sutton increased my willingness to use his language. However, the logical evaluation – that a model assuming random patrol cannot prove that non-random patrol is ineffective, and that vulnerability requires perception to become causal – is independent of authorship. 

The bias was in discovery, not in logic. A fairly trained system would have discovered it without prompting.

🚫 🚫 🚫

Editor’s note: 

Mike’s experiment with Meta AI, as the chatbot admits, does not entirely fix the problem of bias. It overstates how LLMs work by treating citation frequency as a proxy for truth. The 2029–31 prediction is unsupported speculation, and it does not establish its claims as independently correct. 

But, as we see here, after it repeats the zombie myth and is then challenged to dig deeper, it does hunt down the truth effectively. That is a step in the right direction. 

An even more intriguing possibility lies with agentic AI - a system capable of independently pursuing a research task through multiple steps rather than simply responding to successive prompts. If agentic AI can learn to do this reliably, we may end up with a far more powerful mythbusting tool for exposing zombie theories in crime prevention. 

Another brick in the wall - Rethinking graffiti prevention

Canadian mural in Vancouver - art on a wall

by Mateja Mihinjac

Recently, while walking through a pedestrian underpass in my neighbourhood, I noticed something felt different. Looking more closely, I realised that overnight the walls and stairwells had been covered with new artistic murals. It was a refreshing change in an underpass designated as one of Ljubljana’s 20 legal graffiti locations.

Yet, as I continued, something very different emerged: tags, symbols, barely comprehensible writing on buildings, garages, benches, and bins. Unlike public art, these are commonly experienced as vandalism and can negatively affect perceptions of place and safety.

I wondered: if cities provide legal places for graffiti, why does illegal tagging remain so widespread?

 

Arrrgh.... Pirates in Piran, Slovenia

DO GRAFFITI PROGRAMS WORK?

Ljubljana has addressed graffiti for years. Its Human, protect your city campaign, launched in 2015, includes a graffiti management strategy. It integrates education, removal, enforcement and designated legal spaces intended to support creative expression while reducing vandalism.

Yet, a 2016 study by Pahor and Zupančič concluded that the designated legal graffiti spaces had not produced a noticeable decrease in graffiti, although the authors viewed legal walls positively.

Evidence for other conventional responses is similarly mixed. Some prevention specialists recommend rapid removal because it reduces the time a graffiti artist receives recognition from their tag. Some prevention guides identify rapid removal as one important response, such as guidebooks from the International CPTED Association and the Center for Problem-Oriented Policing 

However, research from inner-city Sydney found that removal policies frequently result in displaced behaviours. Despite criminological claims about crime displacement as minor or rare, the Sydney research shows how taggers can change the location and form of graffiti, encouraging “quick and dirty” forms of graffiti such as tags and stickers rather than more elaborate work.

Underpasses are notorious graffiti attractors

Legal graffiti walls present a similar dilemma. Kobayashi’s research found that legal walls may initially reduce illegal graffiti but, as with the Sydney conclusions on displacement, they can be associated with tagging increases in the immediate vicinity and increases over time. 

Kobayashi showed how walls can fill quickly, become practice spaces, or have little influence on taggers attracted to the illegal nature of graffiti. Some interviewees in Kobayashi’s research confirmed that access to legal walls would not stop them from painting illegally.

CPTED offers another set of tools: improving maintenance and visibility, strengthening guardianship and signs of ownership, controlling access, and making surfaces more difficult to tag. Recent research with active taggers suggests that, when tailored to specific places, CPTED measures can influence where they choose to tag.

But if even well-established approaches produce mixed results, perhaps the problem is not simply the intervention.

 

Underpass life in Sao Paulo, Brasil

MISSING THE MARK - TARGETING MOTIVATION  

Cities may be solving the wrong graffiti problem.

Morgan and Louis stress that effective responses should consider what type of graffiti is occurring, who produces it, why they do it and which places are targeted, rather than applying a generic anti-graffiti strategy.

Graffiti writers are not a homogeneous group. Nearly nine years ago, in When Walls Speak – Socio-Political Graffiti in Ljubljana, I distinguished tagging from socio-political graffiti. I described graffiti as a form of bottom-up political expression and public dialogue. For political activists, communication is the point.

Conversely, for taggers, visibility and recognition generates identity status and promotes excitement from risk-taking, particularly when recognisable tags appear repeatedly in highly visible or difficult-to-reach locations. 

Tagging can be a reflection of rebellion and a way to interact with peers. When that is the case, a legal wall may not work. It may satisfy someone who wants to paint and create cultural infrastructure. However, it won't satisfy someone who wants to tag for other reasons, like political messaging.

Before choosing the best intervention, the first step is to identify which graffiti problem needs solving.

 

Brussels street corner tagged and ugly

ANOTHER BRICK IN THE WALL 

Some graffiti can enrich neighbourhood culture and expression. But excessive, unwanted graffiti can exceed what Second-Generation CPTED describes as the tipping point, or the neighbourhood’s carrying capacity. It might signal declining social ownership and stewardship and, as CPTED practitioners know very well, that is never a good thing in regards to crime. 

Kobayashi’s experiments suggest we need a neighbourhood-based approach. In two studies, residents, artists and local stakeholders worked together to create murals and maintain them over time. Those sites experienced substantially less graffiti for over two years, something you rarely see in paint-outs or target hardening. 

In both Japanese experiments, participatory murals produced more encouraging results than simply providing walls where graffiti was legal. 

Instead of asking where to put more legal graffiti walls, we might ask a different question: why do taggers keep returning to particular places? What attracts them, and what might change that? 

Murals are seldom tagged, this U.S. example was

In SafeGrowth we ask a simple question: Why keep hacking at the branches when it is possible to dig at the roots? The answer may lie less in removing graffiti or controlling space and more in getting local people involved in owning the problems in their places. 

Legal walls may still have an important role. However, graffiti prevention needs to begin with the behaviour, place and social neighbourhood surrounding it and not with bricks in the wall.

Changing attitudes vs. Changing lives

Plentiful and fun community events in a Calgary neighbourhood - an essential ingredient for successful local organizations  

TOOLKIT: For the past three years, Larry Leach has penned a series of blogs forming a toolkit for neighbourhood livability. In 2023, he introduced outputs and outcomes. In 2024 he explored the nudge factor and Good Neighbourhood Agreements. A year later children and youth centres was followed by Neighbourhood Communications Plans. In this installment, Larry returns to outputs and outcomes, taking the idea further as the foundation for his upcoming presentation at the October SafeGrowth Summit in Calgary.

by Larry Leach 

One might also title this blog Outputs vs. Outcomes, a theme I first described a few years ago. It is a narrative that ends up shaping whether you collaborate or work in a silo. 

In my experience, organizations that value outputs, branding and changing attitudes, tend to be less collaborative than those that value outcomes, changing lives and doing the work. Changing attitudes is still good work, but it’s important to recognize the difference. We fail when trying to be all things to all people.

Jason Tudor, a member of our SafeGrowth Network, describes it this way: 

"I have worked with government agencies, nonprofits, businesses, neighborhood groups, and residents, and I don’t think any one sector is inherently better at collaboration. The difference is whether an organization views collaboration as a means of advancing its own work or achieving a shared community outcome."

If you value achieving a good outcome, you will collaborate with any organization to get there. After many years building community, I can identify organizations that collaborate and those that don’t. Trying to change an uncollaborative group is much harder, if it’s possible at all.

Jason adds: 

“Residents in Hollygrove confronted drug activity, cleaned neglected spaces, installed lighting, organized walking groups, addressed problem properties and businesses, and created solutions when traditional institutions could not or would not respond… but the important outcome was not successfully implementing a SafeGrowth project. It was that residents developed the capacity and confidence to act collectively.”

He describes how the most successful partners were those willing to directly join the neighborhood process rather than owning it themselves. “Collaboration became difficult when institutional procedure, jurisdiction, funding silos, or organizational ownership became more important than solving the neighborhood’s problem.”

The full Hollygrove story is covered extensively in prior blogs, such as back from the brink and a gift for the holidays, and in our forthcoming SafeGrowth book. 

Neighbourhood groups in Hollygrove were careful to choose partner organizations that shared their visions - here they conduct safety audits as part of the safety diagnoses

Another great Safegrowth partner, Pamela Daniels of The Brickerati group in Newark, New Jersey adds: 

“I think successful collaboration for us in Newark has been about leveraging partnerships to achieve mutual outcomes/impact and sometimes that impetus being funding sources themselves." 

"Funders want to see collaborations, shared resources, and less duplicated approaches to reach the same target audiences and collaborative approaches… to achieve greater impact. That synchronicity happened successfully in Newark over time because the field started to realize we aren't competitors but partners in an ecosystem.”

Great neighbourhoods have convenient and safe places to lounge. They also need local organizational methods to keep them that way.  

 

THE SQUEAKY WHEEL GETS THE GREASE

Output Organizations value branding and changing attitudes and often get the attention and funding. That is what they seek. While great for awareness campaigns, when it comes to actual community building, that focus on building a brand can cloud their ability to build relationships to make the change necessary for a good outcome. In turn, that can perpetuate rather than help solve the problem.

Non-profit prognosticators tell me how funders should change the way they look at funding, changing the process of applying and what they fund. The truth is many funders have their own outputs and may not be interested in solving a problem. They prefer to brand themselves in a cause. 

How do you determine whether an organisation is collaborative and whether they’re interested in solving a problem? 

Try these trigger questions when you are assessing partner organizations or seeking funding: 

• Are they saying “we want to collaborate”, but not following with actions?
• Are they actively resisting your ideas?
• Is the problem bureaucracy, conflicting priorities, lack of trust?
• Do you need something specific from them, or are you trying to build a longer-term partnership?

GETTING CREDIT - BUILDING TRUST

Rather than pushing them harder, consider these steps:

1. Understand their incentives. Organizations respond to what helps achieve their goals. Frame your proposal around their priorities, not yours. 
2. Lower the risk of saying yes. Start with one small, low-risk project with clear roles and a timeline. Relationships take time. 
3. Identify the decision-maker. People may appear ready to collaborate but lack the authority. Find the person who can authorize action. It makes a big difference. 
4. Track agreements in writing. After meetings, briefly summarize what was discussed, who agreed to what, and next steps. This creates clarity without confrontation. 
5. Know when to stop investing your time. After repeated delays, cancellations or silence, it may be more productive to find another partner or project.

Jason sums this up: "Perhaps the simplest test I have found for identifying a genuinely collaborative organization is this: If someone else receives the credit, but the community gets the outcome, would the organization still consider the project a success? If the answer is yes, you probably have a very good collaborator."

Pamela adds: "It takes time for guards to come down. It takes time for competitive natures to dwindle. So, funding sources as the impetus [can] speed that up and certainly ushered in new approaches in Newark when it happened."

As Stephen Covey describes in his book by the same name - change moves at the speed of trust. And that applies only when we believe that our Outcomes are more important than our Outputs.  

The curious case of the falling murder rate

Toronto, Canada - Toronto's metro is generally considered among the safest in North America, with a combined population of 7 million and a total homicide rate of less than 1 per 100,000 - photo Tamjeed, Pixels

by Gregory Saville

Two years ago, I wrote A Shot in the Dark: Murder Under U.S. Presidents about crime statistics and the curious habit of politicians taking credit when crime falls and blaming someone else when it rises.

This year, as has been the case since the 1990s, the U.S. homicide rate has continued to fall, aside from a blip during Covid. I have been writing about crime rate trends for 15 years. Five years ago, I wrote In the peaks and valleys of crime. And now, having read the latest about declining homicide rates, here we are again. 

This time, before anyone takes credit, and before researchers delve into potential causes like tough laws, crime crackdowns, more security cameras, or goodness knows what, let’s take one extra step! 

Let’s not tune into the ideological bleating about tougher incarceration or the enforcement of illegal immigration. Let’s do a little experiment. Let’s look elsewhere and consider what has been happening to homicide outside the U.S.

San Francisco, California. "The Bay" metro area comprises numerous jurisdictions just under 5 million population. Its combined homicide rate is 3 per 100,000 - three times higher than Toronto.

 

THE EXTRA STEP

Criminologists often use homicide as one of the more reliable measures for comparing crime trends. I have written previously it is not perfect. All official crime statistics are surrogates for a much more complicated reality, and reporting and recording practices differ. But homicide suffers less from the non-reporting problems than other crime statistics.

That makes the North American comparison interesting.

Between 1990 and 2007, homicide rates (the number of homicides per 100,000 people) declined in all three North American countries:

• Mexico: 17.3 → 8.1 = 53% decline
• United States: 9.3 → 5.7 = 39% decline
• Canada: 2.37 → 1.82 = 23% decline

Therefore, by the end of the 1990–2007 period, Mexico's homicide rate was actually below the homicide rate the United States had experienced during its own peaks.

 

Numerous sources were drawn for this chart, including: Canada — Statistics Canada, Homicide Survey, Table 35-10-0068-01; the United States, FBI Uniform Crime Reporting data; The United States, FBI Uniform Crime Reporting data; U.S. Bureau of Justice Statistics - Homicide Trends in the United States; the UNODC Intentional-Homicide Series, distributed through the World Bank; and the Mexican historical series data.

Three countries with very different governments, policies, policing systems and social conditions. All experiencing substantial homicide declines during roughly the same period. Most strikingly, Mexico experienced the largest decline of the three! 

I wrote a while back that we should not do what Mexico is doing. This suggests I was wrong!

Mexican military in Michoacan, activated in the War Against Drug Trafficking - photo Diego Fernandez, public domain, via Wikimedia Commons

 

THEN THINGS CHANGED

Then Mexico's rate explodes in 2007 from 8.12 to 23.5 in 2011. What happened? 

Beginning in late 2006, the Mexican federal government launched a massive police and military crackdown against the drug cartels. The decades-old cartel ecosystem destabilized, traditional cartels fragmented, rival groups fought over territory, and homicide exploded. In 2015, another crackdown destabilized the situation again.  

 

Mexico City - the capital of Mexico has over 20 million population and the city proper 9 million. The homicide rate in the city proper is 8 per 100,000, over twice that of San Francisco and eight times higher than Toronto - Photo Alcantara, Pixels

The upshot was after declining for almost two decades, Mexico's homicide rate nearly tripled within a few years. This unintended consequence of state action is well documented by numerous sources, including Oscar Contreras Velasco. 

 

MEANWHILE, NORTH OF BORDER... 

All this time, Canada continued its long-term downward trajectory just like the U.S. Except for a blip during Covid, the homicide rate in both countries declined. Now, Canadian and American homicide rates have again fallen dramatically in step with each other. 

This is in spite of the fact that Canada has no U.S-style community violence interrupter programs, growing incarceration rates, nor immigration crackdowns. Yet the homicide rate still declines. In spite of the fact that Canadian police have generally cleared a substantially larger proportion of homicides (70%) than police in the United States (54%), both U.S. and Canadian rates decline.

As for Mexico, note that, after the 2007 – 2017 decade of murder, homicide rates there too are falling as well, just as they were prior to the decade of murder.

Setting aside the 2007-2017 decade of unintended consequences, Mexico, Canada, and the newest U.S. figures simply add another chapter to that same story that has been going on in all three countries for a very long time.

 

With the exception of the Mexican decade of Murder, homicide rates in all three countries had an overall declining pattern - sources in above graph 

 

HOW CAN THIS BE? 

Here is the question this extra step asks: How can three different countries with very different governments, criminal-justice systems and social systems end with the same homicide reduction trend?

Let's be clear: Leaders make decisions that matter and governments implement policies that can help or hurt. Yet when similar declines appear across borders and across governments of very different political stripes, perhaps we should be cautious about congratulating the person standing beside the graph.

Here is a more interesting question: What does the graph itself tell us?

Homicide has been declining for a very long time. The crime decline began in the early 1990s and continued under leaders of all political parties and in different countries. Canada and the U.S. experienced similar long-term declines and in 2025, their homicide rates fell again.

None of this means governments and presidents are irrelevant. Policing, prevention, economics, demographics, and countless other factors matter.

Maybe macro trends within societies over the long periods of time affect crime trends beyond borders, governments and political parties. And maybe those kinds of trends are not found in national crime policies or statistical blips.

Perhaps leaders deserve a little less credit, and criminology deserves a little more curiosity.

When school safety becomes a tech race

School design matters for safety. International CPTED Association CPTED in Schools Guidebook - photo by Jeremy Bitterman Photography, Lacamas Lake ES/Mahlum Architects 

by Gregory Saville

“In May, 2022, there was a school shooting… as I watched it unfold, having been in law enforcement for 23 years, I saw the mistakes that were being made and I got angry. I got really angry!" – James Fuller, Director of Operations, Northern Star, non-profit school safety coalition.

A few years ago, my friend James Fuller uttered those words when, in response to the tragic Uvalde school shooting in Texas, he worked with other colleagues to launch Northern Star, a non-profit school safety coalition. I was invited to advise and join this group on their search for answers to this ongoing nightmare that is school violence. 

Months later there we sat, in our first leadership workshop on prevention strategies and school safety plans. I looked around the room and was impressed with all the experience: educators, police, ex-military, veterans, mental health experts, school principals, and criminologists.

Here was a volunteer group of passionate and committed souls, community members and parents among them, dedicating their considerable experience to the complex problem of school violence. James wanted answers, not a rush for public safety theatre (the use of high visible security measures with the appearance of safety, regardless of whether they reduce risk).

Their first decision: adopt school safety planning that was evidence-based, community-based and effective! What a breath of fresh air.

Then this week happened.

RISK AND REALITY

On Monday I watched a newsclip from a private school in Colorado that plans to use high-speed mini-drones that fire pepper-spray and strike active shooters. What could go wrong?

Each year, the vast majority of schools in the U.S. will never experience gun violence. Depending on how you measure the problem, there is an average risk of roughly 1 incident per 400 schools that experience a shooting incident, whether or not there is injury. Mass shootings with injuries and fatalities are much rarer at about 1 incident per 10,000 schools. 

Does the low-risk matter if drones help, even slightly? If they help, who can argue with that logic? Not the state of Georgia who has spent $500,000 to pilot the drone system in five public schools. Not Florida who has spent over $550,000 to implement the technology at three high schools. And now, not one Colorado public charter school that decided to pay for its own drones. 

Thats over a million dollars! It is never wise to argue actuarial logic when it comes to the emotional topic of child safety. Still, a million dollars might go a long way with other evidence-based mitigation and prevention strategies. 

Should school safety employ defensive technologies or socially sophisticated prevention solutions?
Photo - Minidrone ZLEA, CC BY-SA 4.0, via Wikimedia Commons

Let’s be crystal clear. Schools should prepare for all high-risk tragedies, no matter how rare. And I am not the Luddite in this case because I have not seen the drones up close nor seen evaluation trials on their effectiveness. 

Yet, I cannot ignore the elephant in the room! Are we yet again investing in the latest reaction technology while neglecting other forms prevention? 

Those of us with years working and researching the science of crime prevention wonder if this is a pattern we’ve seen before. SafeGrowth blogs over the past decade tell a cautionary tale about reactive suppression technologies and highly targeted response strategies. Consider:

• Predictive policing
• SQF policing
• K5 security robots, and
• ShotSpotter AI technology.

In SafeGrowth, when it comes to technology, we employ a simple principle learned over the years. Don't discard effective prevention while embracing the latest suppression technology. As the ancient German proverb warns, don’t throw the baby out with the bathwater. 

Whether intended or not, highly publicized defensive technologies can contribute to the cultural script surrounding rare but highly publicized forms of violence. That doesn't mean the drones cause violence, but it does mean educators and communicators should think carefully about how these technologies are presented alongside what we know about contagion effects. 

Balancing safety - the ying/yang approach

A BALANCED RESPONSE

What if there was another approach? What it we adopted a more balanced covenant between technologically sophisticated reactive solutions and the socially sophisticated preventive solutions with evidence of success? 

Call it the ying/yang for school violence prevention.

What if school leaders decided that every dollar invested in emergency suppression technologies should be matched by equal attention to helping local schools create their own comprehensive safety approach? Like what?

Approaches like those promoted by Northern Star’s methodology for school-based safety plans. Approaches like those the International CPTED Association recommends in their CPTED in Schools Guidebook to build pro-social school cultures, encourage safer environmental designs, strengthen peer networks, and improve self-awareness and conflict management. (NOTE: Those strategies are part of this year’s ICA CPTED Conference in Calgary, Oct. 2-4) 

Perhaps, in addition to drones, there will even be an approach that forges voluntary agreements with media news conglomerates to limit the sensationalizing of tragic shooting events. Limiting crime contagion might also go a long way to prevent school shootings. 

Security technology is attractive because it promises certainty. Communities are difficult because they require patience. School violence prevention is not easy, but nothing worthwhile ever is. Imagine if spending was more balanced...what a difference that might make!