Addicted to on-call

This is an explanation of the systems that lead an organisation to become addicted to on-call.

Most people recognise addiction in drugs, alcohol, nicotine, sugar, etc. A system can also have addiction in it. In a system it’s often called dependence rather than addiction. But ultimately it’s the same thing.

System dependence

System dependence or addiction stems from a short-term stimulus – that wears off – which does nothing to change the underlying reality.

  • A Farmer adds fertilizer to his field, this boosts the crop for that year, but does nothing to change or improve the underlying soil quality.
  • A government gives a subsidy to an important but unprofitable industry. The industry’s underlying profitability is unchanged, and the industry is now dependent on the subsidy.

In the system dependence, and in both of these examples, there’s been some state created by an intervention. The government intervenes, the farmer intervenes. States that are created by an intervention don’t last. The subsidy is spent, the fertilizer is consumed or washed away.

An intervention might be fine

If the intervention results in a stable system, then maybe it is fine. The farmer is dependent on fertilizer, but assuming there’s no other negative side effect, then so be it. That’s the intervention required to prop up this system. The system is addicted to the intervention.

But this might be fine.

Except, if there is a negative side effect. If the intervention reduces the systems ability to support itself.

  • Modern medicine has shifted the responsibility of health away from the practices and lifestyle of individuals, and onto the medical interventions of doctors.
  • Pesticides kill the pests (bugs) and the predators (spiders). Creating a reinforcing dependence on pesticides, and reducing the system’s capacity to regulate itself through a natural prey/predator balance.

Applying to on-call

Any manual intervention in the running of a product or service can suffer from this reinforcing dependence on the intervention.

If the product is missing a feature or piece of functionality, and we take engineers away from building solutions and use them to manually intervene in the product, we’re creating a dependence.

If those interventions are toil – a manual or repetitive task with low enduring value – then we’ve created an addiction/dependence. The more missing features, the more the engineers must manually intervene, the less the engineers work on product features. Soon, just keeping the system alive requires constant manual intervention.



  • the on-call intervention does nothing to change the underlying problem (missing features),
  • and detracts from the system’s ability to support itself (by spending engineering time on intervention and not on product feature fixes).

.. these combine into a self reinforcing feedback loop of increasing dependence on the intervention.


  1. The first solution is to avoid the intervention in the first place. This strengthens the system’s ability to manage itself. This will require us to think about the long-term health of the system instead of the short-term benefit of an intervention. Allow the system to evolve to balance itself, to be healthy and sustainable.
  2. If we are already addicted, the solution is withdrawal. Like all withdrawal this will be painful. But it is required if we are to remove the addiction. We are paying back the debt we accrued for the short-term gains of intervening in the system.