Two ping pong players are engaged in a duel, surrounded by a crowd of cheering supporters. We’re on the 8th floor of a building designed to look like a set from Star Wars. Is this really a government office?

This is the HQ of a crack team of Singapore Government data scientists and developers. It’s shaped like a giant robot, was designed by George Lucas, and has corridors lifted from the Death Star. Officials here are not tasked with running a galactic empire, though, but thinking disruptive thoughts about public service delivery.

The first has already emerged: why are taxis the only direct form of public transport? Wouldn’t it be good if you could get a public bus from your doorstep, reserve a seat, and have it take you directly to work without stopping multiple times on the way? That service now exists, and it’s a new hope for Singapore’s millions of commuters.

The problem with public transport

“We call this The Hive” says Liu Feng Yuan, director of government analytics at the Government Digital Services. Casually dressed in a checked shirt and jeans, he’s joint boss with Mark Lim, head of the software development team. “It’s an effort to get a combination of data scientists, software developers and designers to develop and use technology to help government agencies deliver better services.”

Ping pong match aside – the youthful team are hard at work figuring how to transport millions of people to work. The problem is that trains and buses operate on fixed routes, so many feel the need to use a car or taxi. “Cars and taxis are more flexible, but also more expensive. And from a sustainability point of view, taxis and cars aren’t good because of congestion,” Liu notes.

Over the past few years, Singapore’s Land Transport Authority has met demand by upgrading capacity, ensuring that the system can cope with huge commuter surges every morning and evening. The Beeline team augmented this by looking at the directness of the routes. “People were complaining: ‘I have to transfer from a feeder bus to the MRT [train]’, so we said: ‘Using data and mobile technology, couldn’t we think about something in-between mass transit – which costs two or three dollars – and Uber, which is twenty or thirty dollars? There’s a big gap in between.’”

How it works

The flagship scheme is called Beeline, and launched as a pilot last month. It’s working with bus operators to adapt public transport to changing commuter patterns – without requiring the city to build a new train line or launch a regular bus service.

“It’s a bit more direct than a bus that stops thirty stops,” he says. “Beeline stops for a maximum of five stops, sometimes three, and then is straight on the expressway. It’s not as personal as a taxi, but if you get 15 people on a bus, that’s a big win from an environmental sustainability point of view.”
It also tackles another problem: seating. “Sometimes you don’t know if you’re going to get a seat on a public bus, so wouldn’t it be great if you could pre-book a seat using an app? Now you can.”


“Wouldn’t it be great if you could pre-book a seat using an app? Now you can.”

The team is currently crowdsourcing the routes, finding new demand from housing blocks which previously had to make multiple changes on their morning commute. There are five trial routes, with more planned, and the app encourages people to request new routes. “If enough people want it, we’ll activate it,” he says. So far 7,000 people have made suggestions, with the most popular routes ‘trending’ on the app to encourage bus companies to launch new services.

beelinebus

The story so far

The idea was first mooted last November, with development starting in February. It’s a joint project between the IDA – Singapore’s tech agency – and the Land Transport Authority.

The project was inspired by a few different international systems. First, Hong Kong’s frenetic public minibus network, which Singaporean commuters have long desired. However, those buses operate on fixed routes with multiple stops – it isn’t the data-driven, on-demand principle that backs up Beeline. They’re also seen as messy and chaotic.

The data-driven side was inspired by Uber’s proposed ride-sharing model, and also by a couple of bus concepts in the United States. The latter, however, prioritised luxury over necessity, pricing out most ordinary commuters and struggling to make a difference.

At around $5 a ride, Beeline isn’t much more expensive than a bus and train ride in the morning, and is much cheaper than a taxi. The team hopes that it will spur a new revolution in Singaporean transport, but it’s being run as a six month pilot with strict KPIs to gauge its effectiveness and popularity. “Our role as a government agency is to catalyse new transport options for Singapore,” Liu notes. “The Prime Minister said that government agencies need to be more experimental, they need to be less afraid of failing.”


 “The Prime Minister said that government agencies need to be more experimental, they need to be less afraid of failing.”

Key for the team is engagement, he says. “How many people engage with our site, how many people book… hopefully in the longer term, we want this to be something that the bus companies can take on a commercial basis.”

Bus companies have been key to the initiative and, since January, a great deal of work has been done to win them over. “The point for them is to see if people are willing to pre-pay to book; pay a bit more for a guaranteed seat; and whether they like the concept of crowdsourcing.”

The really clever bit

The bus operators were won over the by unit’s data analysis. Using crowdsourced requests and anonymised public transport data, they could demonstrate a need for buses to serve areas where people were currently making multiple changes in their morning commute.

Liu’s team built a dashboard to show anonymised data for the morning commutes of millions of Singaporean residents. Using the digital ticketing system “I can figure out when you first tapped on the network and when was the last tap, and I can measure the journey time. The heatmap shows a blue area where people took on average 15 to 20 minutes to get to their destination. The people who were red took an average of 35 minutes – and within that there’s a wide variation.”

The data can show where there is enough latent demand in an area for a direct service. “This is where my [team’s] computer science PhDs come in handy: they search the network and find viable Beeline routes.”

Built by Beeline

The criteria was to find a route with a minimum of 20 people in need, but also during a 15 minute window. “It’s no point if they’re spread out throughout the day. We also wanted to minimise the number of bus stops – five maximum – otherwise it’s not going to be worth it. As few as possible, but enough to make it commercially viable.”

The bus companies were thrilled by the data visualisation (they don’t get access to the raw datasets). “They recognise that on the taxi side there was a lot of technology disrupting that space, and so they were quite pleased that they had data to help them plan.”

New tricks

The Beeline scheme shows how government skill sets are changing. Traditionally, economics and the liberal arts were key. Now, governments across the world are finding that data science and development skills make a huge difference to citizen’s lives.

“You need people who can run algorithms to search the network and find these optimised routes; someone to use code to pull the data through the Google Maps API, we’re talking about hundreds of millions of records of data – it’s not something you can do in Excel. Then you’ve got the person who does the frontend and visualises the data; and we need to build filters so we can look by region, by time.”

The unit has been designed to attract people with these skills – who don’t typically find the idea of bureaucracy appealing. “We want to attract younger Singaporeans who are interested in technology and public service. We’re trying to look for people trying to do things for the public good.”

They have 15 data scientists in the team, with a development team of 5 people – including a user experience designer, a couple of developers and an agile project manager. In some ways, it’s run like a tech startup but there’s one key difference: they don’t focus on monetisation. “Even if you can’t monetise it but it benefits society, you should do it because that’s where we tackle problems that a startup wouldn’t take on.”


“Even if you can’t monetise it but it benefits society, you should do it”

One of Liu’s data scientists is behind us playing ping pong (he seems to be winning). They are creating a different feeling for government, while retaining the public service ethos.

Looking ahead

What’s next for Beeline? Key for the pilot is greater takeup. “As with all matching platforms, there are certain limitations without scale. Critical mass is a virtuous cycle: we’ve only got five routes, so even if you log on you may not have a route that’s suitable. The reason we want functions to suggest routes is so that you can still have your say. The more routes we have, the more relevant it is to people. The more people we have, the more bus companies say: ‘hang on, you have a huge user base’ and want to hop on.”

Getting this user base requires a communications campaign, which starts next week. There is already a Facebook page, and they are adapting the app to allow people to recommend routes to their friends – building up campaigns to start new journeys. It would be particularly effective if neighbours clubbed together to recommend a route, quickly reaching the required number for a bus to stop by their home.

The team also wants to do a Kickstarter-style concept, where people commit to the route and make the benefits even clearer to bus companies. Technically, this is more difficult, so it has been de-prioritised until later iterations.

They are also open to ideas from the private sector. “We’re really open to new collaborations, more bus companies, more tech companies, startups who want to collaborate with us,” he says. “Frankly, logistic companies and bus companies haven’t been the most tech savvy, which is why Uber managed to make a lot of headway in the US. If we can help local SMEs in Singapore use data for the way that they do things, then that’s a big win for us.”

Potentially, Beeline could become an API, meaning that other apps and services are able to run it through their own systems. “We’re not in the business of running a company. It’s about seeding the idea, it’s about innovation for public service and for public good… anybody interested in this market might be able to tap the data.”


“We’re not in the business of running a company, it’s about seeding the idea”

Private condominiums could also propose routes, especially if they have built new blocks that aren’t yet served by public transport. And offices might want to run feeder buses for their employees if their campus is far away from where many of their staff live. Government officials took part in an early trial in August, and now will be able to leverage the service to provide feedback and benefit from direct transport to work.

The ping pong game has finished – I missed who won – and many of the team are hunched over their desks, headphones on, typing furiously. It could easily be the office of a tech startup, and in many ways it is.

But Beeline won’t IPO – it will either adapt or grow as a new service to help more Singaporeans get to work on time. All it needs is feedback, Liu says. “We’re keen to know how we can improve it!”

Contact the beeline team at ga@ida.gov.sg or https://www.facebook.com/beelinesg