On Monday morning the municipal cleaning dashboard turned green. Its map suggested improvement in nearly every neighbourhood: resolved reports were up, a weekly line rose sharply, and large circles marked areas with the most activity. Across town, Pilar stood beside overflowing bins on a street that did not appear on that display. She does not use the reporting app. Her phone is old, the sign-up process felt difficult, and when she once called, she was told that sending a photograph was the quickest route. The waste was there. What was missing was the route that would turn it into municipal data.
A public visualisation can be valuable. It can make crews' work easier to follow, support questions about delays, and reveal patterns that a case list conceals. Trouble begins when its colours are treated as the entire territory. A chart does not speak for itself: it summarises observations gathered through particular channels, under particular rules, during a particular period. Unless those conditions are visible, an apparent improvement may become an unfair conclusion about streets where fewer people have the access, time, or trust needed to make a report.
An empty area may be missing evidence#
Several steps sit between rubbish on a pavement and a point on a map. Someone has to notice the problem, be able to report it, have their report recorded, and supply a location the system can accept. The final screen usually shows only the last step. A street with no markers may be clean, but it may also be uninspected, unfamiliar with the app, represented by phone calls without precise geolocation, or excluded because a report did not fit the form.
Several steps sit between rubbish on a pavement and a point on a map. Someone has to notice the problem, be able to report it, have their report recorded, and supply a location the system can accept.
That ambiguity matters when resources are allocated. A policy that sends more cleaning capacity to places with more reports may reward neighbourhoods whose residents are more able to use the digital channel. Other streets can then be described as quiet when they are simply quieter in the database. No hostile intent is needed for that outcome. It follows from treating recorded demand as though it were all existing need.
The dashboard should state plainly what it represents: “This map shows reports received through the app and their handling. It does not measure all litter or all reports made through other routes.” This is not an admission of failure. It prevents residents, journalists, and decision-makers from assigning the display a certainty it does not have.
Coverage belongs in the result#
Coverage describes where a way of seeing reaches. For street cleaning, it is not enough to count reports by road. The city needs to know what share came from the app, telephone service, scheduled inspections, collection crews, or neighbourhood organisations. It needs to know how many reports had reliable locations, how many were placed manually, and how many could not enter the map at all. Timing matters too: a street inspected yesterday carries different evidence from one not visited for six months.
This can be communicated without publishing personal information or overwhelming readers. A panel beside the map can show sources by proportion. Different shading can identify areas whose figures depend almost entirely on digital reports. Inspection routes can be published by broad grid or small area, while unlocated reports can be counted separately. Very small numbers should be combined to protect privacy. The aim is not to identify a caller; it is to show which parts of the city are being heard unevenly.
That distinction changes the public conversation. “Few observed incidents” is not the same statement as “little information available.” Once both are visible, a green zone cannot be used to justify fewer visits without an account of how the evidence was gathered. Residents also gain a concrete basis for asking the service to check places with weak coverage.
Bias starts before the colour is drawn#
Digital exclusion is the obvious issue, but it is not the only one. An app may require an email address some people do not use, work poorly with a screen reader, be partly untranslated, or consume too much data. Time creates another filter. A person working shifts, caring for someone, or sharing a phone may see the overflowing bin yet be unable to stop and complete a report. A building with a concierge may generate many reports through one organised resident; a street where previous complaints went unanswered may stop reporting altogether.
The service itself can create selection. A crew following the same route repeatedly will notice more problems there. A form designed around bins may describe waste around a market badly. A metric that counts only closed cases may make processing look faster even when a report has been bounced between teams. Before calling a neighbourhood disengaged, the city should examine how its design makes participation easy for some people and hard for others.
The response is not to discard app data. It is to place them alongside other evidence: sampled inspections on streets with few reports, routine observations by crews, telephone contacts, in-person reporting, and conversations with local groups. Every source has limits. Their combination makes it less likely that fluency with a smartphone alone decides which problem receives attention.
Limits make interpretation more useful#
The visualisation should say what comparisons cannot establish. More reports may mean more waste, but they may also follow a campaign that made the app better known. Fewer reports may signal cleaner streets, an outage, or frustration after slow replies. A higher number per kilometre does not automatically make a district worse; it may have markets, construction work, events, or many more passing visitors. Without this context, colour rankings are an unstable story about place.
Instead of a triumphant label such as “cleanest neighbourhood,” the interface can prompt better questions. Which reporting source changed? Which areas were not inspected? How many cases remain open? Does the actual response time match the promised one? These questions do not make the map obscure. They help people distinguish an observation from a final judgement.
The public also needs a way to challenge what the map misses. If Pilar cannot find her street, she should be able to report it without creating an account, attaching a picture, or knowing the correct category. A telephone line, a short form, in-person service, and partnerships with community groups are complementary routes, not inferior queues. Each should provide a tracking reference and enter a record that can be audited alongside app reports.
A small test that changes the service#
For eight weeks, the city could test this approach in three contrasting areas. It would retain the existing dashboard, add a coverage indicator, schedule sample walks through low-report streets, and provide an in-person reporting point. Each week, a named decision-maker would compare the channels, response times, and areas that become visible only through inspection. The question is not whether the app is good or bad. It is what the city stops seeing when the app becomes its only listener.
Useful outcomes would be specific: show sources and update dates on the map, extend inspections where evidence is thin, simplify the telephone route, or repair a category that loses reports. The trial should also check that a report made in person is handled with the same dignity and speed as one sent from a phone.
A municipal map earns trust when it displays both its findings and its edge. When the dashboard next announces improvement, Pilar should be able to see how much of the city supports that claim, where evidence remains weak, and how her observation can count without first becoming an expert app user.
