The silence of the Shades

How rational non-reporting turns local failures into regional crises

When the communal tap on Cockbill Street runs dry, nobody sends word to the Palace.

These are not the same notes as the one about informal community resilience making need invisible over long cycles. That is a real loop and it runs here too. But there is a faster, more specific mechanism at work, and it does different damage.

When the Cockbill Street tap fails and residents carry water from Small Gods rather than report the fault, the Small Gods pumps begin absorbing demand they were not built for. The Small Gods pumps are not large. By day three, their pressure is noticeably lower. The residents of Small Gods begin supplementing from the standpipe two streets further on. That standpipe serves a dozen houses. It now serves three times as many. By day five, both Cockbill Street and Small Gods are short of water. The Dollymount Ward pump, connected to the same distribution main, begins showing reduced pressure too. By the time the original failure reaches someone with the authority and inclination to do something about it, what should have been one repair order and two days of localised disruption is a multi-district water emergency.

This is the detection delay amplifier. The delay is not the context for the damage. The delay is the damage. Every day the failure goes unreported, the failure grows. Not because the broken tap is getting worse, but because the systems adjacent to it are absorbing a load they were not designed for, and each of those systems is now closer to its own failure threshold.

The repair that would have cost one order and two days now costs three orders and a week, and requires coordination across three districts rather than one. The cost of the original failure is not the broken tap. The cost is the compounded consequence of the time that passed before anyone fixed it.

The system dynamics

The detection delay amplifier runs in two phases.

In the first phase, a failure occurs but is not reported. The failure is managed locally through workarounds: carrying water from elsewhere, sharing resources, reducing consumption, absorbing the loss. The workarounds are functional. The system appears, from outside, to be operating. The failure is invisible to anyone who is not directly experiencing it and who is not being told about it.

In the second phase, the workarounds impose load on adjacent systems. The adjacent systems were not designed for this additional load and had no margin built in for it. Under the additional load, they begin to degrade. Their degradation may generate its own workarounds, which impose load on further adjacent systems. The failure is propagating outward, still invisibly, through the dependency structure of the network.

When the failure eventually becomes visible, through a system that has no further workaround available, or through an individual who happens to be in the right place to notice, the revealed failure is not the original one. It is the original failure plus everything it has caused in the time it went unaddressed. The cost of the response scales with the time elapsed, not with the size of the original fault.

Why non-reporting is rational

The mechanism is often discussed as a communication problem: people did not know how to report the failure, or did not know who to contact, or did not realise the failure was relevant to report. These are real barriers in some situations. They are not the barrier in the Shades.

The Shades does not report because it has learned, through accumulated experience, that reporting produces nothing. This is not a conclusion reached through a single disappointment. It is a policy position, maintained rationally, based on evidence collected over many years of sending word and receiving no response, or receiving a response weeks later, or receiving a response that addressed a different problem from the one reported, or receiving a polite acknowledgement that logged the report and produced no action.

The decision not to report is not apathy. It is a form of resource conservation. Complaining requires time and energy. In a community that does not have either to spare, the expected return on investing them in a complaint to the Palace is near zero. The rational decision is to direct that energy elsewhere: to the workarounds that have previously proved more reliable than the official channel.

This is the feature of the mechanism that most complicates any technical fix. If people do not report because they cannot find the complaint form, the solution is a better form. If people do not report because reporting has never worked, the solution is not a form. It is a demonstrated change in what reporting produces. And that demonstration must happen first, before the reports can be expected to change, which creates a sequencing problem: the improved response must precede the reporting, but the reporting is what would make the improved response visible and sustainable.

Sicily and the invisible decay

Parts of southern Italy have received piped water on a rationed schedule for decades. In much of inland Sicily, in parts of Basilicata and Calabria, the distribution network cannot sustain continuous supply. Water is delivered on a rotation: every other day, or every third day, with hours rather than days of continuous pressure. Households have adapted. Roof tanks accumulate supply during the on-hours. Private cisterns hold reserve. Arrangements with trucked delivery services cover gaps when the rotation fails. The adaptation is widespread, largely informal, and functionally adequate for daily life most of the time.

The consequence of this adequacy is that the underlying distribution network has been deteriorating for thirty years with insufficient political signal to justify the investment that would fix it. The adaptation that makes the situation liveable also makes the failure of the formal network invisible at regional and national level. Officials assessing water service provision in these areas see low reported outage rates, because outages are absorbed by household tanks before they become reported incidents. They see low complaint volumes, because residents have learned to manage without reliable piped supply and do not expect a complaint to change that. The formal data reads as adequate. The formal data is not measuring the right thing.

When the 2023 Sicilian water crisis produced emergency responses across multiple provinces, investigators found distribution networks that had been visibly deteriorating for years: pipes losing significant volume to leakage, pumping stations running on equipment well past its service life, distribution pressure inadequate to reach upper floors in the absence of household storage. None of this had produced the political signal that would have justified intervention, because the adaptation had been absorbing it continuously. The regional authorities responsible for the network had, in a technical sense, no reason to treat the situation as an emergency. The formal indicators had been quiet. The delay had accumulated for three decades. When it resolved into visibility, it resolved as a crisis that required emergency expenditure many times what orderly maintenance would have cost.

Jakarta: Adaptation against an accelerating floor

The northern coastal districts of Jakarta have been sinking for decades. The mechanism is primarily groundwater extraction: deep wells drilled to supply households and commercial buildings where piped water is unavailable or unreliable pull water from aquifers faster than they recharge, causing the land above to subside. Parts of North Jakarta are now several metres below sea level and continuing to sink at measurable rates each year.

The residents of flood-prone northern districts have adapted to flooding through informal infrastructure accumulated over generations: raised floor levels, seasonal adjustment of stored goods, local knowledge of which roads flood first and which routes remain passable. These adaptations are effective. Floods that would be emergencies in an unadapted neighbourhood are managed disruptions in an adapted one. Life continues.

The adaptation performs the same function as the Cockbill Street workaround at a longer cycle time. The formal infrastructure response to Jakarta’s subsidence, comprehensive in some planning documents and negligible in others, has had less urgency than the underlying risk warrants, partly because the informal adaptation has been absorbing the consequences in ways that reduce their immediate political visibility. The risk continues to accumulate. The adaptation becomes harder as the floor level drops relative to sea level. The workarounds require more effort and more investment from the households maintaining them each year. The original failure, in this case decades of inadequate groundwater management and delayed piped water provision, continues to compound beneath a surface of managed adaptation.

The load transfer mechanism here is geological and slow. But the structure is identical: the adaptation that allows daily life to continue is the same mechanism that allows the underlying failure to continue unaddressed.

Bangladesh: When the workaround is the hazard

The detection delay amplifier runs its most consequential variant when the workaround adopted in response to an invisible failure is itself the source of harm.

Through the 1970s and 1980s, international development programmes in Bangladesh promoted the drilling of shallow tube wells as a replacement for surface water sources that had been causing cholera and typhoid epidemics. The tube wells were understood to be safe. They removed the pathogen risk of surface water. Adoption was widespread, with hundreds of thousands of wells drilled across the country by the early 1990s.

The tube wells drew from aquifers naturally contaminated with arsenic. The contamination was not known at the time of drilling. Arsenic poisoning through drinking water is chronic, not acute: the symptoms develop over years of exposure and include skin lesions, peripheral neuropathy, and elevated risk of several cancers. There is no immediate signal that the water is harmful. Households that had replaced genuinely dangerous surface water with tube well water were making a rational decision based on available information. The workaround was better than the original problem.

By the time the arsenic contamination was identified as a systematic problem in the mid-1990s, tens of millions of people had been drinking contaminated water for years. The World Health Organization described it as the largest mass poisoning in history. The original problem, unsafe surface water, had generated an adaptation, tube wells, that resolved the reported failure while creating an unreported one. The delay between the adoption of the workaround and the identification of its harm was not a failure of communication. It was a failure of detection: the harm was real, it was accumulating, and it was invisible to everyone, including the people experiencing it, until it had run for long enough to produce symptoms at a population scale that could not be missed.

This is the detection delay amplifier without any element of rational distrust of institutions. The people using the wells were not declining to report a problem they knew about. They were unaware that there was a problem to report. The failure to detect was not a social or political phenomenon. It was an epistemic one: the measurement systems in place were measuring the wrong things, and the harm was taking place in the gap between what was measured and what mattered.

What the delay actually costs

The cost of the detection delay amplifier is systematically underestimated because it is measured against the wrong baseline. When a multidistrict water emergency is eventually resolved, the cost reported is the cost of the emergency. What is rarely calculated is the cost relative to what the response would have been if the failure had been detected and reported on day one.

That calculation is almost never done, because it requires knowledge of what was happening on day one, which was precisely what the non-reporting prevented. The compounded failure looks, at the point of response, like a large failure that required an expensive response. It does not look like a small failure that became a large one because nobody reported it. The two are indistinguishable at resolution time without detailed retrospective investigation, which is rarely conducted because the emergency has passed.

This means that the cost of the delay is regularly absorbed into the apparent cost of the original failure, making the original failure look worse than it was and making the delay invisible as a cost category entirely. The budget argument for investing in systems that generate earlier detection, in trusted reporting channels, in demonstrated responsive action that would change the rational calculus of non-reporting, is never made, because the cost it would reduce does not appear in the accounts as a separate line.

The tap and the network

When Vimes eventually discovers that Cockbill Street has been without water for five days and the Small Gods pressure is dropping, the question he has to answer is not just how to fix the Cockbill Street tap. It is how to restore three distribution zones that have been running above their design load for five days, how to address the Small Gods pump that has developed its own fault under the strain, and how to do all of this in less time than the multi-district failure is going to take to compound further.

He fixes it, because he is Vimes. But the question that the repair closes before it can be asked is this: if reporting the Cockbill Street tap on day one would have cost one order and two days, how much of the five-day bill, the three-district bill, the emergency-mobilisation bill, was actually purchased by the Palace’s longstanding practice of not responding when Cockbill Street sends word?

The tap broke. The silence made it expensive.