the app that rides along

Problem statement

App-based food and parcel delivery on motorcycles exploded after 2020 across Latin America and Southeast Asia, and the riders are dying: ECLAC reports that motorcyclist fatalities in Chile rose from 112 in 2013 to 231 in 2023, with "the rise in delivery drivers using delivery applications, such as Uber Eats and Rappi" identified as one of the main factors (an attribution ECLAC takes from a La Tercera press investigation, not from a causal study); Colombia registered 800,000 new motorcycles in 2022 and motorcyclists are 60 percent of its road deaths; in Mexico City "despite the introduction of a special license aimed at improving motorcycle safety ... accidents caused by motorcycles remain high" (103 of 224 road deaths in the first half of 2022 were motorcyclists). Malaysia counted 112 delivery-rider fatalities between 2018 and May 2022. But the standard diagnosis — reckless couriers — is not what careful observation shows. In an observational study of 4,950 riders at six signalised intersections in Shah Alam, Malaysia (52.3 percent of them identifiable as food-delivery riders), delivery riders were less likely to ride without a helmet or run red lights than other motorcyclists, and no different on stop-line violations — but 2.66 times more likely to use a phone while stopped and 2.49 times more likely to use a phone while moving. The unsolved problem is that the delivery platform's interface and incentive structure — accepting orders, navigating, confirming pick-ups, all on a handset, against per-delivery pay and time-sensitive bonuses — is itself the risk exposure, and interventions so far (licences, helmet enforcement, rider education) target the rider rather than the system that puts the phone in his hand.

Why this matters

Motorcycle fleets are growing rapidly in low- and middle-income countries and, per ECLAC, are used largely commercially there and overwhelmingly by men (78 percent of users), so platform delivery is turning a mass-employment sector into a mass-casualty one; Colombia already has the worst fatality rate (15.8 per 100,000) among the 35 countries in the ITF's IRTAD database. Because the riders are contractors, the crash cost falls on them and public health systems rather than on the platform whose design and pricing shape the exposure. If the observational evidence generalises — that delivery riders are otherwise more compliant than average but distracted by the work tool — then the safety gain lies in redesigning the tool and the incentives, which is cheaper and more scalable than enforcement against hundreds of thousands of individual riders (Malaysia alone had an estimated 70,000 in 2021).

What’s been tried and why it hasn’t worked

Regulators have reached for the instruments they have for individual road users: Mexico City's special motorcycle licence, Uruguay's helmet and speed decrees, Chile's zero-tolerance alcohol policy, awareness campaigns — none of which touches phone interaction while riding for work. A 2025 systematic review of last-mile delivery workers (Journal of Transport & Health) attributes delivery-rider stress and risk to "time pressure, uncertain income, and digital surveillance," and other studies describe riders breaking traffic rules to protect income and avoid customer complaints, yet the observational data suggest that in Malaysia the excess risk shows up specifically as phone handling, not as red-light running or helmet non-use — evidence that the platform workflow (order acceptance windows, live navigation, chat with customers) is being executed at speed. Malaysia's road-safety institute MIROS reported 70 percent of delivery riders "riding dangerously, especially during peak hours," and those peak hours are when delivery demand concentrates. Platforms have not been required to make their rider apps operable hands-free or to pause order offers while the vehicle is moving; the intervention space that would change the exposure — interface design, incentive design, algorithmic dispatch — has been treated as private product design outside road-safety regulation.

What would unlock progress

Reframe the delivery app as safety-critical in-vehicle equipment, subject to the distraction-design principles applied to car infotainment: motion-locked order acceptance (offers hold until the bike stops), voice-first navigation and confirmations, batching of decisions at pick-up and drop-off, and incentive structures that do not reward seconds saved between stops. Adjacent precedents: automotive driver-distraction guidelines (task lock-out above a speed threshold), aviation's sterile-cockpit rule, and truck telematics that score and coach rather than fine. The evidence gap to close is causal — whether phone handling by delivery riders is driven by app design, by incentives, or both — which is testable with instrumented rides and A/B interface changes on a willing platform.

Entry points for student teams

An HCI/engineering team could prototype a motion-aware delivery-rider interface (order offers and chat locked while moving, voice confirmation, glanceable turn cues) and test it against a standard flow in a riding simulator or closed course, measuring glances and phone touches per kilometre — a prototype platforms and regulators can react to. A behavioural-economics team could model how per-delivery pay, acceptance-time windows and peak bonuses translate into time pressure and design incentive variants for a field trial. A public-health/data team could replicate the intersection observation protocol in a Latin American city (Santiago, Bogotá, Mexico City) to test whether the Malaysian pattern — more compliant, more distracted — holds there. Relevant skills: human–computer interaction, human factors, behavioural economics, road-safety epidemiology.