The most consistently trusted security intervention in our research wasn’t a camera system, a guard patrol, or a physical barrier. It was good lighting.
Fifty-two Sydney residents evaluated 73 security interventions — from CPTED-integrated design features through to weapons screening and intensive surveillance. Each was rated on a scale from most unsafe to most safe. The interventions that generated the highest ratings and the broadest community consensus were design-led: lighting, sightlines, active frontages, well-maintained spaces, community activation. The interventions that rated lowest, and divided respondents most sharply, were enforcement-heavy: weapons screening, intensive patrols, perimeter barriers.
This is not a comfortable finding for a profession that routinely specifies the latter and treats the former as incidental.
The study
This research was conducted through the School of Built Environment at the University of Technology Sydney, with Associate Professor Pernille H. Christensen and Gabriela Quintana Vigiola. The methodology combined three approaches: descriptive statistics to rank interventions and measure consensus across the sample; non-metric multidimensional scaling (MDS) to reveal the underlying structure of how communities perceive security — the dimensions that actually organise their evaluations; and reflexive thematic analysis of open-ended responses, generating 54 codes consolidated into 11 categories and three themes.
Security professionals, architects, and planners were excluded. The study was designed to capture how communities — the people who use public spaces — perceive protection. Not how experts categorise it.
The tipping point
The finding I keep returning to is this: beyond a certain threshold, visible security stops communicating safety and starts communicating threat.
One respondent described it this way:
“When security measures are particularly visible and in your face you have to wonder why they’re necessary, there must be a potential threat which automatically makes a place feel slightly less safe.” — Respondent 18
Another put it more directly:
“I think well designed places will always feel safer than installing CCTV.” — Respondent 47
These responses reflect a pattern in the data. High-intensity security measures — those that project overt control into everyday space — didn’t just rate lower than design-led alternatives. They also generated the highest disagreement. Communities weren’t uniformly reassured or uniformly unsettled by them. They split. Some respondents found intensive measures comforting. A significant portion found them threatening.
This is the tipping point: the threshold beyond which security becomes legible as a threat signal rather than a safety signal. Communities read a heavily secured space and think “something is wrong here” — not “I’m protected here.” The security investment and the community experience move in opposite directions.
The threshold isn’t fixed. It varies with context, with community, with how measures are communicated and justified. But it exists. And it’s crossed more often than most security specifications acknowledge.
How communities actually perceive security
The MDS analysis revealed something practitioners rarely model: communities don’t categorise security the way professionals specify it. Physical, electronic, and operational are professional categories. They don’t map onto how people experience a place.
Three dimensions organise how communities perceive security interventions.
Intensity is the dominant dimension. It captures how much overt control a measure projects into everyday space. At one end: design-integrated features that distribute safety through the environment without announcing themselves. At the other: enforcement-heavy controls that claim space, regulate movement, and communicate authority. Design-led measures consistently positioned toward the safe end of this dimension. The relationship is not linear — intensity works up to a point, then inverts.
Guardianship distinguishes whether a measure feels backed by credible human presence and response. Passive environmental features — sightlines, lighting, natural surveillance — score differently from active human oversight. A well-lit, activated street and a stationed security guard both provide guardianship, but communities read them very differently. Passive guardianship tends to integrate; active guardianship can reassure or unsettle depending on context and deployment.
Signalling versus Control captures whether a measure communicates preparedness or imposes restriction. Some interventions communicate “this place is looked after.” Others communicate “this place restricts your movement.” The same technology can land on either side of this dimension depending on scale, integration, and the presence or absence of surrounding context that anchors the intent.
These three dimensions explain much of what practitioners observe but rarely formalise: why CCTV reassures in some contexts and unsettles in others; why visible security guards can increase or decrease perceived safety depending on deployment; why over-designed perimeter treatments make civic spaces feel less welcoming than a well-planted, activated street.
What’s wrong with the dominant specification logic
Most security specification for public space works backward from threat. You establish the scenario, select the countermeasure, confirm it meets the relevant standard, and document the decision. This produces defensible outputs. It doesn’t reliably produce places communities experience as safe.
The research identifies two systematic errors in this logic.
The first is the assumption that more visible protection signals more safety. The data doesn’t support this as a general claim. Above a certain intensity threshold, visible security reads as evidence of danger. The investment in protection produces its opposite in community experience.
The second is the omission of community experience from the specification process. Tolerability thresholds cannot be determined by expert assessment. They are set by the people who use spaces — and they vary. A specification that meets threat-based criteria but exceeds community tolerance hasn’t succeeded. It has produced a control that will be contested, circumvented, or gradually abandoned.
I’ve been in briefings where high-intensity security was specified without anyone in the room having asked what community members would make of it. The threat logic was rigorous. The community impact analysis didn’t exist. In several cases the result was controls that unsettled the users they were meant to protect, created friction at design review, and in some instances were redesigned at significant cost. The security brief was technically fulfilled. The place didn’t work.
What protective placemaking means in practice
The concept the research advances is protective placemaking: the argument that protection and place quality are not competing values requiring trade-offs, but mutually constitutive conditions that must be designed together.
Design-led measures — lighting, sightlines, activation, maintenance — aren’t a supplement to security. They are its most trusted expression in everyday public space. Treating them as secondary to “real” security infrastructure misunderstands the evidence on what communities find safe.
High-intensity countermeasures are appropriate in specific circumstances: where the threat assessment warrants them, where the risk is credible and clearly communicated, and where they are proportionate and integrated with the surrounding environment rather than imposed on it. The threshold of community acceptance narrows sharply as intensity increases. Beyond that threshold, a measure may be technically correct and community-counterproductive at the same time.
Three practical conclusions follow.
First, design-led measures should be specified as primary security infrastructure, not as aesthetic additions to a security plan. The research supports their prioritisation. Treating them as optional upgrades to a barrier-and-surveillance package inverts the evidence.
Second, every specification decision involving high-intensity controls warrants an explicit assessment of where it sits relative to community tolerability. Not as a compliance exercise — as a genuine design question: will this measure, in this context, read as a safety signal or a threat signal? The answer depends on scale, integration, communication, and what surrounds it.
Third, community participation in security design is not procedural. Safety audits, iterative consultation, and co-design processes are practical tools for calibrating against tolerability thresholds. They cannot be replaced by expert assessment, however rigorous, because the threshold is set by communities, not by standards.
The question the data raises
The question this research puts to anyone specifying security in public space is not “does this countermeasure address the threat?” That is necessary but insufficient.
The full question is: will the community that uses this space experience this as protection — or as its absence?
Getting that right is not a political accommodation to community concern. It is a technical requirement for security that works. Controls that communities reject or read as threatening don’t protect people. They can actively worsen the experience of safety they were specified to provide.
Protective placemaking is the discipline of holding both questions simultaneously — threat-based and community-based — and designing toward answers that serve both. The research now gives practitioners the evidence base and the framework to do it.
This research was conducted in partnership with the School of Built Environment at the University of Technology Sydney. Principal researcher: Codee Ludbey. Co-investigators: Associate Professor Pernille H. Christensen, Gabriela Quintana Vigiola..