MSOM Grand Challenges · SDG 11

What Cities Told Us

How to make cities inclusive, safe, resilient, and sustainable?

Saif Benjaafar (Michigan), Long He (Notre Dame), Alexandre Jacquillat (MIT), Bo Lin (NUS), Sheng Liu (Toronto), Wei Qi (Tsinghua), Zuo-Jun Max Shen (HKU), Phebe Vayanos (USC), Jingwei Zhang (Cornell)

Aerial view of a dense city center with a rail station beside a river — Central Melbourne from above. Photo: Dietmar Rabich, CC BY-SA 4.0

The framework

Stories from our interviews, sorted by the nine grand challenges in our paper.

Application Domains

Access

Supply

Coordinate

Housing ServicesSDG 11.1

TransportationSDG 11.2

Urban ResilienceSDG 11.5

Urban SpaceSDG 11.7

Housing Services · CoordinateC4Implementable housing allocation1 story ›Transportation · AccessC5Sustainable transportation for underserved communities4 stories ›Transportation · SupplyC6Design and operations of autonomous systems3 stories ›Urban Resilience · SupplyC7Disaster response resource allocation5 stories ›Urban Space · AccessC8Equitable access to urban space5 stories ›Urban Space · SupplyC9Designing safe and inclusive urban space with data3 stories ›

C1System AssessmentDescriptive and predictive analytics8 stories ›C2System DesignPrescriptive analytics3 stories ›C3Practitioner EngagementHuman-AI interface3 stories ›

Methodological Foundation

Application Domains

Access

Supply

Coord.

HousingSDG 11.1

TransportSDG 11.2

ResilienceSDG 11.5

SpaceSDG 11.7

C41 story C54 stories C63 stories C75 stories C85 stories C93 stories

C1Assessment C2Design C3Engagement

Methodological Foundation

Application domains

C4Implementable housing allocationHousing Services · Coordinate1 ›C5Sustainable transportation for underserved communitiesTransportation · Access4 ›C6Design and operations of autonomous systemsTransportation · Supply3 ›C7Disaster response resource allocationUrban Resilience · Supply5 ›C8Equitable access to urban spaceUrban Space · Access5 ›C9Designing safe and inclusive urban space with dataUrban Space · Supply3 ›

Methodological foundation

C1System AssessmentMethodological foundation8 ›C2System DesignMethodological foundation3 ›C3Practitioner EngagementMethodological foundation3 ›

Questions for discussion ›Things we heard in the interviews ›

C1 · Methodological foundation

How do we know whether a policy actually worked?

Outcomes show up slowly and partially, and effects spill over to nearby streets.

In the paper: How to rigorously quantify the impacts of urban infrastructure and policy interventions?

A cyclist on a separated bike path beside a road — Bike path in Utrecht. Photo: Madalijna, CC BY-SA 4.0

From our interviews

8 stories

Transport

Car trip data was a poor guide for bus routes

A ride-hailing platform that also ran buses first tried using its own car-trip data to predict where bus riders would be. It didn’t work, because people who take cars and people who take buses are different groups.

Tradeoff: the data you have vs. the data you need

Also in C2 ›

What happened next

Car trip data was a poor guide for bus routes

Instead, the company used a model that guessed each user’s home and workplace from anonymized trip times (morning trips start at home, evening trips at work) and checked the guesses against addresses some users had saved. It was good enough, though not perfect. Frequent car users were left out, since they would not switch. Local teams screened the suggested routes by hand, and stop locations always needed manual judgment.

Transport

Only the complaints got fixed

A city fixed bike lanes based on reports to 311, its non-emergency service line. The person we interviewed said these reports are biased: people report only if they know the service exists, and it mostly draws people who already know how to navigate the system. Problems no one reported did not get fixed, leaving large equity gaps.

Tradeoff: waiting for complaints vs. regular inspections

What happened next

Only the complaints got fixed

About two years before our interview, the city began regular inspections in addition to taking complaints. Staff now ride every bike lane once a month and record problems themselves. When planning, the city also tracks how much it invests in neighborhoods it has officially marked as underserved.

Public space

A survey that counts answers, not who answers

As part of a city-funded needs assessment of a shopping street, a local community group has to survey property owners, merchants, and shoppers. The city sets a target number of responses. It does not check whether those who answered represent the neighborhood, for example whether the lowest-income residents are included.

Tradeoff: a response target vs. a representative sample

Also in C3 ›

What happened next

A survey that counts answers, not who answers

Not solved yet. The city official we interviewed said this might be something the city could do better.

Transport

Bike lanes judged mainly by ridership

After new downtown bike lanes opened, ridership jumped and the city promoted those numbers. The city then leaned heavily on ridership as its sign of a successful bike lane. Suburban lanes will never reach downtown numbers, so suburban residents ask why they should get lanes at all.

Tradeoff: ridership counts vs. other benefits

What happened next

Bike lanes judged mainly by ridership

Not solved yet. The person we interviewed said lanes have benefits beyond ridership, and that it is not clear what success should look like. When choosing where to build, the city already looks at what a lane connects, such as access to jobs, food, and parks.

Transport

How long until we know a street got safer?

The person we interviewed thought it takes at least five years of data to tell whether collisions go down after a new bike lane opens. One year is not enough.

Tradeoff: early results vs. enough data on safety

What happened next

How long until we know a street got safer?

Not solved yet. There is no standard for when to evaluate a new lane: after six months, two years, or five? After one year, the city can measure how many people ride and what the public thinks of the lane. The person we interviewed would like research on when and how to run before-and-after studies.

Transport

Benefits promised, only recently checked

To apply for national funding, a transit project has to show expected benefits under equity and climate criteria. For one transit extension, reports covered how many riders might be women or low-income and what jobs it might create. Checking whether those benefits appear after a project opens has only come up in the last five years or so.

Tradeoff: projected benefits vs. measured benefits

What happened next

Benefits promised, only recently checked

Checking is starting. Government has begun discussing performance measures that projects would keep reporting on after they open. Some benefits, such as emissions avoided, are hard to estimate. The person we interviewed said the models are not perfect, but you try to report on what you can.

Public space

Safety, measured indirectly

A city agency that funds programs on shopping streets judges success mainly by foot traffic. It also looks at spending data from card companies and at storefront vacancy rates. It does not measure safety directly. The official we interviewed, speaking personally, said safety is highly subjective: what feels safe to one person may not to another.

Tradeoff: what can be counted vs. how safe people feel

Also in C9 ›

What happened next

Safety, measured indirectly

The agency looks at safety indirectly: whether a street is clean, busy, and well lit, on the idea that people may then feel safe. It also watches for a rise in quality-of-life complaints, which the official sees as linked to safety. Its grants pay for extra cleaning, lighting, and art or events that draw people in, mainly to support small businesses.

Transport

The app’s recommendations bury the shared ride

In a ride-hailing app, a model picks the few ride options you see, and it is tuned to get you to book right away. Anything slower than a car arriving in two or three minutes makes people less likely to book. A shared ride is cheaper, but it shows a 12-minute wait and a 5-minute walk, so the model has a hard time giving it a spot.

Tradeoff: bookings today vs. value over time

Also in C2 ›

What happened next

The app’s recommendations bury the shared ride

No clean fix. The team of the person we interviewed pushed for the company’s models to also count long-term value, such as repeat use, and argued for taking a short-term hit for a long-term bet. It was a constant internal battle, and hard to win because long-term value had no good measure.

For discussion

How can experiments and causal inference credibly measure what a city intervention achieved?

How do we turn predictions into decisions, with models that can still be solved and explained?

Detailed models predict better but are harder to solve and harder to explain.

In the paper: How to integrate predictive and prescriptive analytics for improving urban systems?

Aerial view of a highway interchange next to a downtown — Highway interchange, Dallas. Photo: Michael Barera, CC BY-SA 4.0

From our interviews

3 stories

Transport

Car trip data was a poor guide for bus routes

A ride-hailing platform that also ran buses first tried using its own car-trip data to predict where bus riders would be. It didn’t work, because people who take cars and people who take buses are different groups.

Tradeoff: the data you have vs. the data you need

Also in C1 ›

What happened next

Car trip data was a poor guide for bus routes

Instead, the company used a model that guessed each user’s home and workplace from anonymized trip times (morning trips start at home, evening trips at work) and checked the guesses against addresses some users had saved. It was good enough, though not perfect. Frequent car users were left out, since they would not switch. Local teams screened the suggested routes by hand, and stop locations always needed manual judgment.

Transport

The app’s recommendations bury the shared ride

In a ride-hailing app, a model picks the few ride options you see, and it is tuned to get you to book right away. Anything slower than a car arriving in two or three minutes makes people less likely to book. A shared ride is cheaper, but it shows a 12-minute wait and a 5-minute walk, so the model has a hard time giving it a spot.

Tradeoff: bookings today vs. value over time

Also in C1 ›

What happened next

The app’s recommendations bury the shared ride

No clean fix. The team of the person we interviewed pushed for the company’s models to also count long-term value, such as repeat use, and argued for taking a short-term hit for a long-term bet. It was a constant internal battle, and hard to win because long-term value had no good measure.

Transport

The paving schedule decides the bike lane

A city picks which streets to rebuild based on how worn the pavement is, not on cycling outcomes. A raised bike lane requires moving the storm drains, which only happens in a full rebuild. If a street is only due for a new surface, the bike lane ends up as painted lines.

Tradeoff: the paving schedule vs. the preferred bike lane design

What happened next

The paving schedule decides the bike lane

Not solved yet. The person we interviewed said what is missing is integrated planning: road budgets that also take cycling outcomes into account. Then a street due only for a new surface could be upgraded to a full rebuild.

For discussion

How should prediction and optimization be combined to design urban systems that work and can be explained?

How do we work with practitioners from the start, not just at the end?

Researchers and practitioners know different things, and stakeholders often want different things.

In the paper: How can stakeholders be engaged throughout the full research lifecycle?

People, including a wheelchair user, around a table at a workshop — Stakeholder workshop on an access and inclusion plan, Brisbane. Photo: John Robert McPherson, CC0

From our interviews

3 stories

Public space

A survey that counts answers, not who answers

As part of a city-funded needs assessment of a shopping street, a local community group has to survey property owners, merchants, and shoppers. The city sets a target number of responses. It does not check whether those who answered represent the neighborhood, for example whether the lowest-income residents are included.

Tradeoff: a response target vs. a representative sample

Also in C1 ›

What happened next

A survey that counts answers, not who answers

Not solved yet. The city official we interviewed said this might be something the city could do better.

Transport

Allowed by the government, ticketed by the police

The government allowed private app-booked buses to operate, and some cities welcomed them as a way to add buses and reduce congestion. But the buses had no approved stops, so buses waiting at the curb for riders got tickets from traffic police for blocking traffic.

Tradeoff: one agency’s goal vs. another agency’s rules

Also in C8 ›

What happened next

Allowed by the government, ticketed by the police

The interview did not describe a fix. What was missing, in the view of the person we interviewed, was official approval: approved stops, and traffic police treating it as an approved service whose bus may wait about five minutes. Support at one level did not mean everything else was cleared.

Public space

One space, many different users

Older adults, children, women, young people, and people with disabilities have different needs, and planners must weigh them when these groups share the same community spaces.

Tradeoff: one space vs. many groups

Also in C9 ›

What happened next

One space, many different users

Neighborhood planners run resident discussions on how green space is used and invite different groups, including children, to help design. The planner we interviewed said sharing a space across different times and uses can ease conflicts, and facilities are adjusted after a follow-up evaluation. Even small details, like where seating is placed, affect how people experience the space.

For discussion

How can optimization tools help practitioners and stakeholders decide together?

C4 · Housing Services · SDG 11.1 · Coordinate

How do we build housing allocation tools that caseworkers will use?

Caseworkers are the social workers who manage each client’s case. Assessments are self-reported and incomplete, and both people and available units keep changing.

In the paper: How to translate data-driven housing allocation policies into implementable decision-support systems?

A long red and cream housing block with arched gateways — Karl-Marx-Hof social housing, Vienna. Photo: Thomas Ledl, CC BY-SA 4.0

From our interviews

1 story

Housing

Ranked by need, not by what will work

Housing for people experiencing homelessness is assigned through a mix of eligibility checks, self-reported vulnerability assessments, waiting lists, and caseworker judgment. Many of these prioritization tools were designed for other purposes and are not directly linked to how people fare after they are housed. And people waiting for housing move around, so they can be hard to reach when a unit opens.

Tradeoff: ranking by need vs. matching what will work

What happened next

Ranked by need, not by what will work

Caseworkers, the social workers who manage each case, know things the data rarely captures, such as clients’ housing needs and preferences. Cities are now working with researchers on tools that use predicted outcomes, like staying housed, when matching people to housing. Not solved yet: how to bring client preferences and caseworker knowledge into those tools.

For discussion

How can data-driven allocation tools fit real housing systems and the people who run them?

How do we bring good transit to fast-growing cities and underserved communities?

Data is unreliable, budgets are small, and needs like personal safety rarely show up in the data.

In the paper: How can the reach of sustainable transportation be expanded in developing countries and among vulnerable groups?

People waiting at a roadside bus stop under a cloudy sky — Bus stop in Barbados. Photo: Berit from Redhill/Surrey, CC BY 2.0

From our interviews

4 stories

Transport

A cheap way to test bus routes ran into safety concerns

To test new bus routes without paying for buses up front, a platform published routes and timetables with no bus assigned. If people booked, it sent a shared car at the last minute and called it an upgrade. Women cancelled these rides at much higher rates. The person we interviewed explained that a bus with 15 other people is a public space, while sharing a car with a driver and a passenger you don’t know feels like a risk.

Tradeoff: cheap, flexible supply vs. feeling safe

What happened next

A cheap way to test bus routes ran into safety concerns

The interview did not describe a fix for this. The person we interviewed still calls it one of the team’s smartest ideas: the tests gave a lot of information without spending much money.

Transport

A premium bus stuck in ordinary traffic

A platform ran app-booked buses with air conditioning and guaranteed seats in several very large cities. Riders who used them loved them, but the buses sat in the same traffic as everyone else, and riders still had to walk 5 to 10 minutes, in poor walking conditions, to unofficial stops with no shelter or sign.

Tradeoff: the service you sell vs. streets you don’t control

What happened next

A premium bus stuck in ordinary traffic

The team got the app, routes, and schedules working well, then hit a wall. What the service needed was faster bus travel, such as bus lanes, plus better stops and better walking conditions, all of which require government support. The lesson the person we interviewed drew from their career: the problems are political, not technical.

Transport

Why a shared ride is never half price

You might expect two riders sharing a car to each pay half. But sharing saves the platform only one cost: the driver’s time. Insurance on the trip is a large cost that can’t be split between riders, and detours to pick up the second rider add cost. So even a perfect match is far from half price. Riders were also more uneasy sharing a car with one other person than a bus with fifteen.

Tradeoff: small savings vs. inconvenience

What happened next

Why a shared ride is never half price

No fix was found. Bigger discounts would need lower costs. Asking riders to wait about 10 minutes so the app could group more trips clashed with the habit of “press a button, get a car.”

Transport

The airport shuttle paid off, the city bus didn’t

One platform ran an app-based bus service across several very large cities in lower-income countries, and an airport shuttle in a city in a high-income country. The shuttle was much more profitable: a car ride to the airport cost several times the shuttle fare, and traffic, though chaotic, was much more predictable. In the bus cities the need was clearly greater, but riders paying a premium still sat in unpredictable traffic and could switch to the metro or a two-wheeler.

Tradeoff: where the need is vs. where the money is

Also in C8 ›

What happened next

The airport shuttle paid off, the city bus didn’t

The airport shuttle was simply the better business. The person we interviewed sees the bigger problem as political rather than technical: the bus service needed faster bus travel, better stops, and better walking conditions, which require government support. A private operator alone hits a wall.

For discussion

How should transit be designed for underserved communities when data and budgets are limited?

How do we design robotaxis, delivery robots, and drones that improve access and safety?

Autonomous fleets still need chargers, curb space, and people behind them.

In the paper: How to design and operate autonomous transportation systems to improve access, equity, and safety in urban services?

A small delivery robot on a sidewalk next to parked cars — Grocery delivery robot, Finland. Photo: Ypsilon from Finland, CC0

From our interviews

3 stories

Transport

Cheaper rides, more traffic

The person we interviewed worries that if self-driving taxis make rides very cheap, and nothing regulates them, people will take many more rides. The person compared it to text messages: people sent far more once each one stopped costing 50 cents. The cars may be electric, but the result could be “carbon-free congestion.”

Tradeoff: lower prices vs. more congestion

What happened next

Cheaper rides, more traffic

Not solved yet. The view of the person we interviewed: cities need ways to manage self-driving vehicles and push them toward shared rides and public-transit uses. Congestion pricing, a charge for driving into busy areas, is a good start in some cities.

Transport

No driver, no one in charge

Fixed-route buses may be the easiest place to start with self-driving vehicles, since routes and times are set. But bus drivers also keep order on board, as an unspoken part of the job. Without a driver, who steps in when a conflict breaks out on board?

Tradeoff: easier automation vs. no one in charge

What happened next

No driver, no one in charge

Not solved yet. The person we interviewed counts this among the real challenges for shared driverless vehicles.

Resilience

Footbridges too low for fire trucks

Footbridges over city roads can be too low for large fire trucks to pass under. For fires in buildings over 100 meters tall, this can stop the needed equipment from getting there.

Tradeoff: everyday street design vs. emergency access

Also in C7 ›

What happened next

Footbridges too low for fire trucks

Not solved yet. The options the person we interviewed mentioned were extreme, like taking the equipment or a bridge apart. The person said management research should focus on better equipment solutions, such as drone rescue systems.

For discussion

How should autonomous fleets be deployed and governed so they improve access and safety?

How should a city split its disaster budget between big infrastructure and local fixes?

Big infrastructure is expensive and slow to build. Local fixes are cheap but small.

In the paper: How to allocate limited budgets between major infrastructure and localized mitigation tools for disaster risks?

A row of large steel flood gates across a river — Thames Barrier, London. Photo: mattbuck, CC BY-SA 4.0

From our interviews

5 stories

Resilience

Footbridges too low for fire trucks

Footbridges over city roads can be too low for large fire trucks to pass under. For fires in buildings over 100 meters tall, this can stop the needed equipment from getting there.

Tradeoff: everyday street design vs. emergency access

Also in C6 ›

What happened next

Footbridges too low for fire trucks

Not solved yet. The options the person we interviewed mentioned were extreme, like taking the equipment or a bridge apart. The person said management research should focus on better equipment solutions, such as drone rescue systems.

Resilience

A hotel withstood the storm

Upgrading a city’s old drains is a long-term job. During a heavy rainstorm, a hotel withstood the storm by installing its own flood barriers.

Tradeoff: slow citywide upgrades vs. quick local fixes

What happened next

A hotel withstood the storm

The government adopted the idea and spread it quickly. The person we interviewed stressed training and placing responsibility on building managers and security staff. For old districts, the person said the focus should be protecting the people most at risk, such as older adults and children, with community-level measures like flood barriers.

Resilience

Where to keep emergency supplies

We asked how emergency supplies should be placed when no one can predict where a rainstorm will hit.

Tradeoff: central stockpiles vs. supplies close to people

What happened next

Where to keep emergency supplies

The person we interviewed said supplies should be split by type. Heavy equipment like pump trucks and excavators is stocked and dispatched by the government. Sandbags and food are handed to neighborhoods, building managers, and companies’ local warehouses, such as an online retailer’s, so they are stored nearby. These local holders can hand supplies out on their own.

Resilience

No one will pay to rebuild old drains

Extreme rainstorm disasters often have several causes at once, so upgrading drains alone is not enough. Rebuilding old drains disrupts traffic for a long time and brings no direct return, so no one wants to invest. A deep drainage tunnel built in one city was judged too costly for other cities to copy.

Tradeoff: one big tunnel vs. many small fixes

What happened next

No one will pay to rebuild old drains

The person we interviewed suggested treating old districts as weak points and protecting them through emergency evacuation and flood barriers. New districts can still plan green space as a buffer. In this person’s view, there is no shortcut: fixing old drains is a long-term job.

Resilience

Preventing fires instead of fighting them

In some dense neighborhoods, a fire can set off a chain reaction, and the narrow roads there can’t be fixed with the government’s current tools. The person we interviewed said traditional firefighting is close to its limit.

Tradeoff: fighting fires vs. preventing them

What happened next

Preventing fires instead of fighting them

The city went after what the person we interviewed called the main cause of fires: low-quality e-bikes. It managed e-bike charging stations and took action against unsafe products at the source. The person’s rule of thumb: management measures can solve about 80% of problems, and specialists handle the rest.

For discussion

How should a city balance large infrastructure and local measures in its disaster budget?

How do we get service into neighborhoods that providers would rather skip?

Public goals have to line up with what private operators and local groups can afford.

In the paper: How to develop practical mechanisms and operational policies to ensure equitable access to urban services and facilities?

A row of rental e-scooters parked on a sidewalk — Rental e-scooters in Portland, Oregon. Photo: Steve Morgan, CC BY-SA 4.0

From our interviews

5 stories

Transport

Allowed by the government, ticketed by the police

The government allowed private app-booked buses to operate, and some cities welcomed them as a way to add buses and reduce congestion. But the buses had no approved stops, so buses waiting at the curb for riders got tickets from traffic police for blocking traffic.

Tradeoff: one agency’s goal vs. another agency’s rules

Also in C3 ›

What happened next

Allowed by the government, ticketed by the police

The interview did not describe a fix. What was missing, in the view of the person we interviewed, was official approval: approved stops, and traffic police treating it as an approved service whose bus may wait about five minutes. Support at one level did not mean everything else was cleared.

Transport

The airport shuttle paid off, the city bus didn’t

One platform ran an app-based bus service across several very large cities in lower-income countries, and an airport shuttle in a city in a high-income country. The shuttle was much more profitable: a car ride to the airport cost several times the shuttle fare, and traffic, though chaotic, was much more predictable. In the bus cities the need was clearly greater, but riders paying a premium still sat in unpredictable traffic and could switch to the metro or a two-wheeler.

Tradeoff: where the need is vs. where the money is

Also in C5 ›

What happened next

The airport shuttle paid off, the city bus didn’t

The airport shuttle was simply the better business. The person we interviewed sees the bigger problem as political rather than technical: the bus service needed faster bus travel, better stops, and better walking conditions, which require government support. A private operator alone hits a wall.

Transport

Scooters required where fewer people ride

Rental scooter companies operate under city contracts. They must meet the city’s vehicle requirements in lower-income “equity zones” and offer lower prices to low-income riders, usually paid for by the company. Those zones bring fewer rides and less revenue, and moving scooters there costs money.

Tradeoff: required coverage vs. efficiency

What happened next

Scooters required where fewer people ride

Some cities offer a trade: serve the equity zones, and you may run more scooters downtown. Some cities waive fees for companies that serve those zones.

Public space

Cleaner streets where property is worth more

Some shopping streets have a business improvement district, funded by fees on property owners based on property values. The money pays for extra street cleaning and running the district, such as hiring a director. Areas with higher property values raise more. For districts with small budgets, rising costs for these basics leave little money for other things.

Tradeoff: local funding vs. equal quality

What happened next

Cleaner streets where property is worth more

The city gives extra grants to districts with small budgets, and separate grants for lighting and art that draw people in. When grants cluster in one part of the city because more groups there do this work, the city looks for groups elsewhere and sometimes has them partner with a bigger organization.

Public space

Insurance before cleanup

Community groups can manage and maintain a public space that is underused or has other problems. But first they must get insurance, agree to take on liability instead of the city, and fill out another agency’s paperwork, which may need a legal review. Groups without the funding or capacity cannot do this, so some neighborhoods are left underserved.

Tradeoff: the city’s liability vs. who can take part

What happened next

Insurance before cleanup

Not solved yet. The city official we interviewed answered in two parts: more resources, meaning both money and tools that help groups learn to manage a public space, and better coordination among city agencies with different missions. The official said officials across agencies know the solution but lack the funding or capacity.

For discussion

What contracts and mechanisms can align private operators with equitable access?

How do we use data to make public spaces people feel safe in?

Safety is hard to measure, and different groups want different things from the same space.

In the paper: How to design safe and inclusive public spaces using data?

A street turned into a garden with trees and benches — Superblock street in Barcelona. Photo: Cataleirxs, CC BY-SA 4.0

From our interviews

3 stories

Public space

Safety, measured indirectly

A city agency that funds programs on shopping streets judges success mainly by foot traffic. It also looks at spending data from card companies and at storefront vacancy rates. It does not measure safety directly. The official we interviewed, speaking personally, said safety is highly subjective: what feels safe to one person may not to another.

Tradeoff: what can be counted vs. how safe people feel

Also in C1 ›

What happened next

Safety, measured indirectly

The agency looks at safety indirectly: whether a street is clean, busy, and well lit, on the idea that people may then feel safe. It also watches for a rise in quality-of-life complaints, which the official sees as linked to safety. Its grants pay for extra cleaning, lighting, and art or events that draw people in, mainly to support small businesses.

Public space

One space, many different users

Older adults, children, women, young people, and people with disabilities have different needs, and planners must weigh them when these groups share the same community spaces.

Tradeoff: one space vs. many groups

Also in C3 ›

What happened next

One space, many different users

Neighborhood planners run resident discussions on how green space is used and invite different groups, including children, to help design. The planner we interviewed said sharing a space across different times and uses can ease conflicts, and facilities are adjusted after a follow-up evaluation. Even small details, like where seating is placed, affect how people experience the space.

Public space

Security cameras, placed department by department

Different city departments install their own security cameras, which leads to duplication. The planner we interviewed also raised a less obvious point: places with no people may need more monitoring.

Tradeoff: each department’s own cameras vs. one shared network

What happened next

Security cameras, placed department by department

The government is pushing departments to share data and pool their camera resources. The planner suggested planning cameras together, so they can also serve needs like environmental monitoring.

For discussion

How can data guide the design and daily operation of safe, inclusive public spaces?

Questions for discussion

C1 · System Assessment ›

How can experiments and causal inference credibly measure what a city intervention achieved?

C2 · System Design ›

How should prediction and optimization be combined to design urban systems that work and can be explained?

C3 · Practitioner Engagement ›

How can optimization tools help practitioners and stakeholders decide together?

C4 · Implementable housing allocation ›

How can data-driven allocation tools fit real housing systems and the people who run them?

C5 · Sustainable transportation for underserved communities ›

How should transit be designed for underserved communities when data and budgets are limited?

C6 · Design and operations of autonomous systems ›

How should autonomous fleets be deployed and governed so they improve access and safety?

C7 · Disaster response resource allocation ›

How should a city balance large infrastructure and local measures in its disaster budget?

C8 · Equitable access to urban space ›

What contracts and mechanisms can align private operators with equitable access?

C9 · Designing safe and inclusive urban space with data ›

How can data guide the design and daily operation of safe, inclusive public spaces?

From the interviews

Things we heard

Paraphrased, with a line of context for each.

Some people would take a 20-minute rickshaw ride to the stop. But once they sign up, the bus had better come on time.

On an app-based bus service in several very large cities. Riders cared much more about frequency and reliability than about route design.

C5 Sustainable transportation for underserved communities ›

The problems are not technical. The problems are political.

A lesson the person we interviewed drew from years of working on city transport from several sides, including advising governments and working at technology companies. What held their bus service back was slow buses in traffic, poor stops, and poor walking conditions, which need government support.

C5 Sustainable transportation for underserved communities ›

It might be electric, but you’d just have carbon-free congestion.

On the challenges self-driving vehicles could bring to cities. Lower prices tend to bring more trips.

C6 Design and operations of autonomous systems ›

A lot of the concern is fear of the change, not the change itself.

A city transportation manager on opposition to new bike lanes, which often softens once the lane is built and the promised benefits show up.

C3 Practitioner Engagement ›

We know the solution. We don’t have the funding or the capacity.

A city official, asked what one decision-support tool researchers could provide. The official pointed first to limited funding and capacity.

C8 Equitable access to urban space ›

There’s only so much money and only so much you can do.

A planner on balancing limited public resources between long-term development and urgent gaps in residents’ basic needs.

C2 System Design ›

If nobody makes money from it, nobody invests.

A disaster-safety expert on paying to upgrade old drains, which disrupts traffic and brings no direct return.

C7 Disaster response resource allocation ›

What Cities Told Us

The framework

How do we know whether a policy actually worked?

From our interviews

Car trip data was a poor guide for bus routes

Car trip data was a poor guide for bus routes

Only the complaints got fixed

Only the complaints got fixed

A survey that counts answers, not who answers

A survey that counts answers, not who answers

Bike lanes judged mainly by ridership

Bike lanes judged mainly by ridership

How long until we know a street got safer?

How long until we know a street got safer?

Benefits promised, only recently checked

Benefits promised, only recently checked

Safety, measured indirectly

Safety, measured indirectly

The app’s recommendations bury the shared ride

The app’s recommendations bury the shared ride

For discussion

Related reading

How do we turn predictions into decisions, with models that can still be solved and explained?

From our interviews

Car trip data was a poor guide for bus routes

Car trip data was a poor guide for bus routes

The app’s recommendations bury the shared ride

The app’s recommendations bury the shared ride

The paving schedule decides the bike lane

The paving schedule decides the bike lane

For discussion

Related reading

How do we work with practitioners from the start, not just at the end?

From our interviews

A survey that counts answers, not who answers

A survey that counts answers, not who answers

Allowed by the government, ticketed by the police

Allowed by the government, ticketed by the police

One space, many different users

One space, many different users

For discussion

How do we build housing allocation tools that caseworkers will use?

From our interviews

Ranked by need, not by what will work

Ranked by need, not by what will work

For discussion

Related reading

How do we bring good transit to fast-growing cities and underserved communities?

From our interviews

A cheap way to test bus routes ran into safety concerns

A cheap way to test bus routes ran into safety concerns

A premium bus stuck in ordinary traffic

A premium bus stuck in ordinary traffic

Why a shared ride is never half price

Why a shared ride is never half price

The airport shuttle paid off, the city bus didn’t

The airport shuttle paid off, the city bus didn’t

For discussion

Related reading

How do we design robotaxis, delivery robots, and drones that improve access and safety?

From our interviews

Cheaper rides, more traffic

Cheaper rides, more traffic

No driver, no one in charge

No driver, no one in charge

Footbridges too low for fire trucks

Footbridges too low for fire trucks

For discussion

Related reading

How should a city split its disaster budget between big infrastructure and local fixes?

From our interviews

Footbridges too low for fire trucks

Footbridges too low for fire trucks

A hotel withstood the storm

A hotel withstood the storm

Where to keep emergency supplies

Where to keep emergency supplies

No one will pay to rebuild old drains

No one will pay to rebuild old drains

Preventing fires instead of fighting them