During a late-August heatwave, the city’s tree dashboard placed a red pin beside a shopping street where dozens of residents had reported wilting plane trees through a mobile app. A maintenance crew was dispatched the following morning. Across town, the small trees along the route to Riverside Clinic had yellowed leaves and bare soil around their trunks, yet the dashboard showed no urgent signal. People waiting for buses there had noticed the heat, but few had used the app. Some relied on prepaid data, some used shared phones, some had never heard of the reporting channel, and some had stopped expecting a response after earlier requests disappeared. The system did not record those conditions. It read the missing reports as a quieter problem.
This is a digital-rights issue because a public service has allowed the ability to produce data to shape the ability to receive care. The dashboard is not malicious simply because it processes reports. It may help crews respond quickly, avoid wasted journeys and identify recurring faults. But a report is not a neutral unit of need. It is evidence filtered through connectivity, language, confidence in institutions, available time and the design of the reporting tool. If those filters differ between neighbourhoods, a priority score based on reports can make a pre-existing inequality look like an efficient allocation rule.
The hidden decision is the definition of a signal#
Data systems often make their strongest judgment before any prediction is calculated: they decide what counts as a signal. A leaking irrigation pipe may be recorded by a sensor. A damaged branch may appear in a work order. A tree without shade may only enter the system when someone photographs it, finds the correct category and submits the image. Treating all three inputs as equally available creates a misleading comparison. The service is no longer measuring only the condition of trees; it is measuring where the city’s chosen channels are easiest to use.
Data systems often make their strongest judgment before any prediction is calculated: they decide what counts as a signal.
The dashboard should therefore distinguish observed conditions from reported conditions. It could label areas according to whether recent field inspections, reliable sensors or only public reports support the displayed status. A low-report area should not automatically be classed as low need. It should be classed as low information. That distinction changes the next action. Instead of moving the area down the queue, the system can schedule a verification visit or draw on a separate heat-risk plan.
This is not a demand for perfect information. Municipal services regularly make decisions with incomplete evidence. The relevant question is whether the uncertainty falls on the organisation that holds the resources or on residents who have fewer means to make themselves legible. If the latter must repeatedly create evidence before a street receives attention, the platform has shifted part of the public service onto them. A technically smooth workflow can still be unfair when it makes a resident’s digital trace the ticket for basic maintenance.
Test outcomes across places, not only model accuracy#
An audit should begin with the outcome on the ground. For each area, the city can compare actual tree stress from scheduled inspections with app reports, sensor coverage, watering visits, canopy shade and local heat exposure. It should then ask whether similar conditions produce similar responses. If two streets have comparable soil dryness but one receives watering because it generates far more reports, the discrepancy is not explained by horticulture. It is an allocation choice made through the data pipeline.
The audit needs a geographical scale fine enough to reveal gaps inside apparently well-served districts. A district average can conceal a bus stop with no shade, a housing estate separated by a major road or a clinic route used by people with limited mobility. Those details matter because public trees affect everyday access to safety and comfort. The audit does not need to track individual residents in order to see them. It can use aggregated service records and environmental conditions, while keeping personal reports separate and protected.
Results should lead to a defined correction, not simply a recommendation. The city might set a minimum inspection frequency for blocks with high heat exposure and sparse reporting, or reserve emergency watering capacity for areas where coverage is poor. It might require the contractor to document completed checks in those locations, rather than rewarding only the volume of jobs generated through the app. A correction rule makes the response predictable and reviewable. Without it, staff are left to improvise around an opaque score, and residents must hope that somebody notices the same gap they have noticed.
Community knowledge can challenge the map#
People who use a place understand things that the dashboard cannot infer. A school caretaker may know that children now wait in direct sun at collection time. A community nurse may see that an older patient avoids a formerly shaded path. A local shopkeeper may recognise that a tree has become hazardous after repeated storms. Their observations can be valuable precisely because they explain consequences, not only physical symptoms.
However, community reporting must remain an additional route, not an unpaid substitute for public inspection. It would be wrong to tell residents that a street can be irrigated only after they organise a monitoring group, learn species identification or send weekly photos. Neighbourhoods with more time, stronger associations and more reliable connections would again supply richer information and obtain more attention. A public authority can invite participation without making participation a condition of being served.
That requires careful design. Reports should be possible without creating an account, available through telephone and in-person channels, and acknowledged in language that explains what will happen next. The city should make clear that one report is not proof of a diagnosis, but that it can trigger a check. It should also publish aggregated feedback: where inspections were added, which irrigation patterns changed and where a concern could not yet be verified. This response loop respects contributors and helps reveal whether the service treats some channels as decorative while relying exclusively on others.
Redistribution begins where the record is thin#
The usual response to missing data is to ask for more of it. A city may promote the app, add QR codes to parks or run digital-skills sessions. Such options can improve access, provided they are voluntary and supported. They cannot be the whole answer. A person who shares a device, works multiple jobs or distrusts a process that has ignored them is not failing a civic duty by leaving no report. Their lack of trace may be exactly the reason the service must take greater care before deciding nothing is wrong.
Redistribution in this context means directing staff time, inspection routes, sensors and water toward places where the standard evidence channel is weak. It also means reviewing procurement incentives. A vendor paid according to the number of digitally generated tickets may have little reason to look for invisible problems. A contract that values equitable coverage, verified field conditions and resolved gaps gives the service a different purpose. Technology can support that purpose, but it cannot define it by itself.
Responsibility should be visible as well. The department that owns the priority rule must be able to describe its inputs and limitations. The parks team must have authority to override a score after an inspection. The contractor must retain auditable records of work carried out. Elected officials should receive regular reports on whether watering and shade protection are distributed across areas with different levels of connectivity. Saying that “the system ranked it lower” is not an explanation when people are suffering the consequences of that ranking.
The trees near Riverside Clinic do not need residents to become data workers before their dry soil becomes credible. They need a service that recognises the difference between no evidence and evidence of no problem. A fair municipal dashboard can still use reports, sensors and predictions. It must pair them with uncertainty, territorial audits and a commitment to move resources towards places that are least visible through its preferred channel. Digital rights are protected not by refusing data, but by refusing to make a person’s capacity to generate it the price of being counted.
