Can Humans Be a Barrier? The 2-Test Rule Every HSE Team in Malaysia Needs to Know
Most HSE professionals say yes — humans can be barriers. In barrier management Malaysia teams are discovering the harder truth: most human “barriers” fail at least one of the two tests that determine whether a control will actually hold when it matters most.
Walk through any HIRARC document in Malaysia and you’ll find them: “supervisor monitoring,” “operator to check,” “inspector on site,” “worker to follow SOP.” These controls are listed under the Risk Control column. They look like barriers. They’re called barriers. And most of the time, they will not hold.
This is one of the most consequential gaps in Malaysian HSE practice — and it’s hiding in plain sight in documents that pass audits every year. The problem isn’t that humans can’t be barriers. The problem is that most organisations have never tested whether their human controls actually qualify as barriers under the criteria that barrier management demands.
Here’s what that test looks like — and what it reveals.
The Foundation: What Is a Barrier?
In barrier management, a barrier is not just any control measure. A barrier has a specific, technical definition. It is a control that:
- Performs a specific positive action — not a general presence, role, or responsibility, but an identifiable thing that happens, triggered by an identifiable condition
- Functions independently — it can prevent the event or mitigate the consequence on its own, without relying on another barrier to work first
These two criteria come from the same framework applied in barrier management practice across industries — from the energy sector to aviation to high-hazard manufacturing. They are not optional standards. They are the definition of what a barrier is. Controls that don’t meet both criteria are not barriers — they are administrative measures, and they carry a fundamentally different level of reliability.
Test 1 — Specificity: Is It a Specific Positive Action?
The specificity test asks: can you name exactly what this control does, under what conditions, and what a successful execution looks like?
A role is not a specific action. A presence is not a specific action. A responsibility listed in a job description is not a specific action.
Specificity Test — Examples
Fails: “Supervisor on site” — This names a role, not an action. A supervisor standing on site does not constitute a barrier until they do something specific.
Fails: “Worker follows SOP” — This names a document, not an action. The barrier is the action described in the SOP, not the SOP itself.
Passes: “Supervisor halts the operation when pressure gauge reads above rated capacity” — This names a specific action (halt operation), triggered by a specific condition (gauge above rated capacity), performed by a specific person. This is testable. This is trainable. This is auditable.
The test reveals a simple truth: if you can’t describe the barrier without referring to a role or a document, you haven’t defined a barrier yet. You’ve defined an expectation.
Test 2 — Independence: Does It Work on Its Own?
The independence test asks: can this barrier prevent the loss-of-control event by itself — without requiring another barrier to function first?
Independence is about what happens when the system around the barrier fails. Many human controls in HIARARCs depend on other controls being in place to trigger them. When those other controls fail, the human control fails with them. That is not independence — that is dependency masquerading as a barrier.
Independence Test — Examples
Fails: “Worker isolates energy because the permit requires it” — If the permit process fails (not issued, not followed), this control fails simultaneously. It is dependent on the permit system. The permit is the barrier; the worker’s action is the permit working, not an independent barrier.
Fails: “Supervisor stops the job because training told them to” — If the training barrier failed (inadequate training, training not retained, incorrect training), this control fails. It is dependent on training, not independent of it.
Passes: “Pressure relief valve releases at 110% of rated operating pressure” — This is a hard engineering barrier. It is independent. It functions regardless of what supervisors do, what workers decide, or whether any other barrier is working.
This is why engineering barriers and hard physical barriers are inherently more reliable than human barriers. They do not require a person to decide to act correctly, under pressure, at the right moment. They act automatically. Human barriers, by contrast, require not only that the person knows what to do, but that they are present, alert, motivated, and operating in conditions that support the right decision — conditions that are never guaranteed.
Applying the 2-Test Rule: A Practical Example
Here is how the test applies to common controls listed in Malaysian HIARARCs in barrier management practice:
| The Claim | Specific Action? | Independent? | Verdict |
|---|---|---|---|
| “Supervisor on site” | ✗ Presence, not action | ✗ No | NOT A BARRIER |
| “Worker follows SOP” | ✗ Document, not action | ✗ Depends on SOP system | NOT A BARRIER |
| “Inspector checks equipment” | ✗ Role, not specific action | ✗ Depends on schedule | NOT A BARRIER |
| “Supervisor halts operation when deviation detected and defined threshold exceeded” | ✓ Specific action + condition | ✓ If trained and authority given | VALID BARRIER — with conditions |
The last row is important. A human barrier can be valid — but only when the action is specific, the triggering condition is defined, the person is trained on exactly that action, and they have the authority and organisational support to act without obstruction. That last condition is rarer in Malaysia than most safety managers would like to admit.
What This Means for Your HIRARC
Run the two-test rule on your current HIRARC. For each control listed under the Risk Control column, ask: Is this a specific positive action? Can it work independently?
If the control fails either test, it isn’t a barrier. That doesn’t mean you remove it — administrative controls and supervision still reduce risk. But you cannot treat them as equivalent to a hard engineering barrier or a well-designed procedural barrier. They sit at a lower level of reliability and your risk assessment should reflect that.
For major hazard scenarios — situations where a single failure can kill someone — relying on human controls that haven’t passed both tests is a structural risk gap. This is where DOSH Malaysia’s major hazard installation framework pushes organisations toward formal barrier analysis, precisely because informal human controls have a documented track record of failing under the conditions that major incidents produce.
The practitioner question: How many of the controls in your current HIRARC have you tested for specificity and independence? If the answer is none, that’s the gap between your documented risk system and your actual risk system.
So — Can Humans Be a Barrier?
Yes. But with conditions. A human action qualifies as a barrier only when: the action is specific and definable, the triggering condition is clear and observable, the person is trained on that specific action, they have independent authority to act, and the action works without relying on another barrier to function first.
When all five of those conditions are met, a human action is a valid barrier. When any one of them is missing, it is a hope — and hopes are not barriers. The difference between the two shows up in investigations.
Build Your Team’s Barrier Management Capability
Cikgu Barrier delivers specialist Bowtie Analysis training that covers barrier classification, specificity and independence testing, escalation factors, and barrier KPIs — for HSE teams managing major hazard risk in Malaysia.
