Harm as Infrastructure: The Systems That Depend on What We Say We Want to Remove

A thought I can’t quite shake

I’ve been thinking about poker machines lately, which is not a sentence I expected to write.

Not because gambling is new, and not because Australians haven’t been arguing about pokies for decades. What’s been sitting with me is something more uncomfortable: the places that benefit from gambling losses are often the same places that hold communities together. The local club. The subsidised meal. The meeting room. The familiar faces. The place where people gather after funerals, celebrate birthdays, and run fundraisers.

That doesn’t make gambling harmless. It also doesn’t mean the story is as simple as “remove the harm and everything improves”. Because what happens when the thing causing harm is also quietly supporting the system around it?

I recently read Lauren Leek’s Substack piece The Machine Is Friendly. That’s Why It Wins., and her framing of harm as something structural, even load-bearing, sharpened something I’d been circling for a while. The idea becomes difficult to un-see once you start looking for it.

When harm is doing work

In many systems, harm isn’t simply tolerated. It performs a function. It keeps something else operating. Remove it, and the system does not necessarily become healthier. Sometimes it becomes unstable because the harm was quietly doing work nobody wanted to acknowledge.

That is a confronting thought because it asks us to consider whether some harms persist not because we have failed to solve them, but because solving them would require redesigning the systems around them. The problem is not sitting at the edges. The problem may be embedded in the centre.

We see versions of this everywhere. Social media platforms are an obvious example, where engagement is the currency and attention is the engine. Emotionally charged content tends to outperform nuance with alarming consistency. Outrage spreads faster than reflection. Certainty beats complexity.

Nobody sits in a room and says, “Let’s optimise for social division.” But systems optimise for outcomes, not intentions. If outrage drives engagement and engagement drives growth, then over time the system learns what works. The result can start to look less like malfunction and more like adaptation.

The question beneath the question

I wonder sometimes whether we spend too much time asking, How do we stop this behaviour? and not enough asking, What function is this behaviour serving?

Those are very different questions. One leads towards moderation policies, controls, and interventions at the edges. Necessary work, certainly. The other pushes us towards incentives, governance, architecture, and business models.

Trust and Safety has often been positioned as a reactive discipline. A protective layer applied after systems have already been designed. I’m increasingly convinced that is too late. If harm is embedded in the infrastructure of a system, then Trust and Safety has to become part of the design process itself, asking not only how to mitigate harms but what outcomes the system rewards, who benefits, and what fills the gap if harmful dynamics are removed.

Because gaps do get filled. Restrictions in one environment can push behaviours elsewhere. Tighten controls in one space and activity migrates to another. Address symptoms without understanding structural incentives and problems often reappear wearing different clothes.

The issue hasn’t disappeared; more often, it has simply moved.

The cost of convenience

This is the part we perhaps discuss less openly: safer systems may not always feel as frictionless. They may require slower decisions, stronger governance, or business models that prioritise sustainability over endless optimisation.

That can feel uncomfortable because we have become accustomed to convenience. We have become used to systems that work quickly, invisibly, and with minimal resistance. The harder question is what those systems quietly depend upon in order to function.

Perhaps it’s data extraction. Perhaps behavioural manipulation. Perhaps labour practices hidden several layers down a supply chain. Or perhaps communities sustained, in part, by losses that accumulate elsewhere.

The specifics change, the pattern doesn’t.

Moving from reaction to redesign

I suspect this is one reason conversations across policy, technology, governance, and community design matter so much. Not because any one group holds the answer, but because structural problems rarely fit neatly within professional boundaries.

It’s also what I find increasingly interesting about spaces like the Trust & Safety Festival Sydney. Not as showcases or networking exercises, but as opportunities to ask harder questions alongside people approaching them from entirely different angles.

Questions such as: What is this harm doing? Who benefits? What happens if we remove it? And perhaps most importantly, what needs to be built in its place?

Those questions are messier than compliance checklists. They are also, perhaps, where meaningful change starts.

Because safer systems are not created by stripping out visible harms and hoping everything else holds. They are created by redesigning incentives so harm is no longer required for the system to function.

About the Author

Kim Chandler McDonald is the Co-Founder and CEO of 3 Steps Data, driving data/digital governance solutions.

She is the Global VP of CyAN, an award-winning author, storyteller, and advocate for cybersecurity, digital sovereignty, compliance, governance, and end-user empowerment.