Public transport bodies: producing lots of data, not necessarily making the most of it

Why does the private sector do this better? Image: Getty.

There’s an app called Citymapper that is probably the one thing I would consider necessary to my day-to-day life, and the justification for owning a smartphone. The things I do on the move - listening to music, checking the web, etc. - are things I don’t need to do while moving; but knowing where I am and where the thing I am moving towards are in relation to each other is an everyday essential.

Obviously, this is what maps are for, and there are maps all over London, but it’s a big city, and coming to it for the first time can be very confusing. Transport for London has a journey planner on its website, but it often feels very unintuitive. For example, here’s what happens when you ask for help with a journey that a daytripper to the capital might take - from Paddington to Tottenham’s football ground, White Hart Lane:

efore I'd moved to London and learned the language of the tube or bus maps, and was merely a tourist, I found this format incredibly confusing. I was using the journey planner because I didn’t know how to get around the city, yet I was being given a list of places I’d never heard of and told to travel between them without any indication of where they were relative to each other. For journeys with lots of steps, or which involve buses, this format can make simple journeys seem convoluted. (And while the TfL site has had a redesign recently to clean it up, it's still the same basic idea.)

Citymapper, though, does this:

It's the same information presented as a map instead of a step-by-step list. It offers text instructions as well, but the key thing is it shows the user where they are and where they’re going, and where the stops along the way are. (And, when I went to New York City earlier this year on holiday, it was better there than the official step-by-step journey planner too.)

Citymapper doesn’t have a special data source - it’s simply using the official data that these transport systems generate in a different way. TfL, like many transport authorities, has what's known as an application programming interface - or API - which tells external apps and sites how to communicate with its servers, and how to retrieve the information they need. This isn't just trains, but also bus departure times, bike hire docking station statuses, taxi information, fare breakdowns, and so on, and is why a search on Apple’s App Store or Google’s Play Store for "London transport" delivers dozens of results.

It's left me wondering this: why is it that public bodies like TfL are so good at producing data, but so bad at using it? Why is an organisation like TfL - which has such a strong brand identity, and which puts so much care and attention into its tube maps, for example - happy to sit back and let third parties control how the public experiences its information?

I emailed Azmat Yusuf, Citymapper’s founder, about this, and he got back to me:

"Wouldn't say transport authorities are bad at using data, they are focused on infrastructure running on time which requires utilising their own data. We are focused on apps and we like to do that well.

Most transport authorities are opening up data, some more reluctantly than others, since it’s a cost to them. But starting to believe and understand that they are better off letting others like us develop applications. Also there's an argument that the data belongs in the public domain anyway."

It’s an interesting idea, that the data “belongs” in the public domain. TfL is a government body, and in a democracy that should mean that we - the public - have a right to its data. Yet we’re living through a strange time, when privatisation and outsourcing make it hard to define what counts as government and what as private company; and where the government is committed to starving TfL of subsidies so that it has to fund itself entirely out of fare revenues, so that it operates less like a public utility and more like a private firm.

Perhaps we should simply be grateful that TfL goes to the lengths it does to provide such comprehensive real-time data feeds across its jurisdiction. The situation could be worse - as I discovered when I spoke to Tom Cairns, founder of Realtime Trains.

His site uses a large range of data sources to piece together a picture of exactly how a train moves around the UK relative to its timetabled route. The problem is that those data feeds are highly reliant on human input to reflect things like cancellations and delays - and if somebody doesn't keep updating those feeds as fast as new information comes in, the official sources, and by extension Realtime Trains, would return innacurate information.

Cairns and his two colleagues use the knowledge of how the network operates - from the lengths of trains to their top speed, to how fast signals change on certain track sections to how long it takes a train to pull away from the platform - to model the likely future arrival time of a train relative to how well it’s already done in its journey. “I have people who go around the UK noting down the timing of the wheels after they start and compare that to the signalling equipment, to get a more accurate picture of what is happening,” he told me. “There’s a lot of work that’s gone into improving the accuracy of the information by hand.”

The problems he faces, though, illustrate the difficulties small developers face when trying to work with a large institution with a reluctance to share its secrets.

"When I get to train companies about getting information directly they say ‘why should we give it out to you, a tiny third party on the periphery?’ In their industry, it’s slow moving. They never had competition in this field either, and they didn’t see a problem with what they’re doing. The opening of [data] took many of us by surprise, in part because we knew they didn’t see a problem.”

Cairns also says that good ideas on third party sites, like his map of track obstructions, have a habit of popping up on the official journey planners a few months down the line. "It was added to the National Rail development roadmap," he explained. "But of course it’s three years later and it still hasn’t appeared." Institutional willingness countered by institutional lethargy.

TfL is due to update its API in the autumn to make it adhere to more universal data standards, and theoretically make it easier for developers to use. I emailed TfL to ask what the motivation was for letting third-party developers loose on its data - whether, for example, it was because of budget limitations - but digital PR lead Rubin Govinden told me that they don't comment on what other people have done.

He did say this, though: 

"There could be eight different travel apps out there and they could all be offering their own service, but the one thing they have in common is they're using the most accurate and up-to-date information. They don't have to wonder ‘is this data pukka?’ - yes it is, yes it's up-to-date, they're all going to be saying the exact same thing at the same time in the same way. What we're giving developers is the ingredients they need to produce the products that our digital customers want.”

When pushed for more specifics, he repeated that he didn’t want to name any individual app or site: “If it's to work successfully, lots of people have to work together, and we're doing our bit by making our data freely and openly available in as many formats in a recognisable format that can be used by everybody.”

This echoes Yusuf's views - TfL's focused on the bit of its job that involves running the trains on time, and the data bit is the afterthought. It's perhaps one model that other public bodies - and not just train companies - could look to when figuring out just how much they want to commit to developing their own tools to make use of the data they produce.


Why is it acceptable to kill someone? On the mysterious history of Britain’s road death toll

A London speed camera, 2004. Image: Getty.

A decade ago I became fascinated by a graph. This one:

I had been tracking the underlining data for years. The figures were easy to remember. Every year it was 3,500, plus or minus a percentage point or two.

Yet when the 2008 data was released, it had fallen to 2,538. This was 1,000 less than the figure in 2003. I plotted the above graph, and as I said, I became fascinated.

Because this is a really important graph. This is a plot of the number of people killed on Britain’s roads each year.

In Great Britain, collectively, we used to kill nearly 3,500 people on our roads every year. Consistently or, dare I say it, boringly: 3,500 deaths a year, 10 a day. It was accepted, in a, “Well yes it’s bad, but what can you do about it” kind of way. There was no clamour for change. Newspapers weren’t running headlines about the deaths mounting up, as they do with knife crime.

Meanwhile a train crash would be front page news for a week. Take the train that derailed at Hatfield on 17 October 2000, a tragedy in which 4 people died. That led to huge media interest, massive upheaval on the railways, and, ultimately, as the re-nationalisation of Railtrack, whose failings had caused the crash. Yet more than twice as many people will have died on the roads that day. Nothing was written about those deaths. Nothing changed.

In 2000, four people died in train crashes, while 3,409 died on the roads.

Here are those figures again.

1997 – 3,599 people killed on our roads

1998 – 3,422

1999 – 3,423

2000 – 3,409

2001 – 3,450

2002 – 3,431

2003 – 3508

But, in 2004 the figure dropped below 3,400 for the first time, to 3,221. Then in 2005 to 3,201.

2006 – 3,172

2007 – 2,946

Below 3,000! This was change. Significant change: 500 lives a year were not being lost. If you use Britain’s roads, your life may have been one of them.

2008 – 2,538

2009 – 2,222

When the 2010 figures came out I was amazed by the headline figure: 1,857.

That’s still far too high, of course, but it was 1,701 lower than seven years earlier.

This was a major story that deserved a ton of coverage, which it failed to get. Having shown no concern for when we were killing 3,500 people, it wasn’t overly surprising that the fact we were now killing 1,700 fewer wasn’t celebrated.

At any rate, the graph had flat-lined for years, then, in half a dozen years, it halved. Why?

The lack of media coverage resulted in an absence of answers. One commentator, Christian Woolmar, observed that there was no clear answer to why this had happened. But he went on to point out that there had been a fall in the average road speed over this period.

My anticipation of the 2011 figures troubled me, because I expected them to go up. Obviously I didn’t want them to: I desperately want zero deaths on our roads. But something happened in 2010 that I was sure would lead to more fatalities and bring a halt to the falling trend.

I was right. In 2011 we killed 1,901.

Sometimes, being right is shit.

The news was better in 2012. The fatality rate was 1,754. So was the 2011 figure just a blip, due to some significant snowfalls that year? No: the trend was over.

The number of people killed on our roads has remained stuck in the 17 hundreds. 

2013 – 1,713

2014 – 1,775

2015 – 1,732

2016 – 1,792

2017 – 1,793

2018 – 1,782

We have returned to a flatline on the graph – and if anything, I’m more fascinated now than I was before. Road deaths flatlined at 3,500 for years, then fell sharply, then flatlined again at half the rate.

This can’t have happened by accident. I wished I could explain it. I wish we could repeat it. No: I wish the second flatline hadn’t happened, and the fall had continued. If the rate of fall had continued, we’d have reached zero deaths on the road by now. You’d be right to question whether this is possible – but if you can half the number in a few years, why can’t we eradicate them altogether? The railways are an example of what is possible. The last time a passenger died in a train crash on Britain’s railways was in 2007.

It was time to figure out the answers to two questions. Why did the death toll fall? And why did it stop falling?

The obvious reason for a reduction in deaths on the road is the improvement in car safety features. This could create a gradual fall in the death toll as new, safer cars replaced older ones. But I’m not sure it can explain a 40 per cent fall over a 4 year period.

There’s a way to check whether cars on the road became almost twice as safe between 2003 and 2010: you can compare the figures with the rest of the EU. Car safety features are international, and any new feature would have appeared around the same time across the continent.

So I found the EU figures for 2000 to 2017, indexed for 2000 and plotted the graph for multiple countries. It was a busy graph. For clarity the following graph only includes Britain, Germany, France, Spain and Italy along with a straight line drop for comparison.

The good news is that things are improving across Europe – but no country had quite the same trajectory as Britain. They all have a fall much closer to a straight line of the sort you’d expect a general improvement in car safety would produce.

One thing I did notice is that, from 2013, these five countries stop falling. The technology based solutions of recent years, such as automatic emergency braking, don’t appear to be saving lives as of yet.

So, yes, cars are safer – but that doesn’t seem to explain why British roads suddenly became 40 per cent safer between 2006 and 2010.

In 1999, the New Labour government announced that it was going to reduce deaths on our roads. The target was a 50 per cent reduction by 2010. As you now know, it succeeded. This was a major achievement for a government. The kind of thing you would bang on about all the time. “Deaths on our roads halved by Labour!” But the party wasn’t in government when the 2010 figures were released – and it’s hard to take credit for your achievements from the opposition benches.

That it was government policy is not a full explanation, and how this happened is a little opaque. From what I can gather there was a wide ranging approach. The fire and rescue service changed their practices: because they recognised that survival rates were directly dependent on how quickly people got to hospital, this became the priority. Disturbing a police crime scene was allowed if it saved a life. Accident black spots were located, highlighted and safety measures implemented. Throughout that period road safety campaigns focused on speed, with “Speed Kills” being the dominate message for that decade. The government also changed the laws on speed cameras.

RoSPA, the Royal Society for the Prevention of Accidents, has a lot to say about speeding and speed cameras. Its “Speed Camera Factsheet” states that, “Cameras are a very effective way of persuading drivers not to speed, and thereby reducing the number of people killed and seriously injured.” It reports that an independent review published by the Department for Transport (DfT) in 2005 said that “cameras significantly reduce speeding and collisions, and cut deaths and serious injuries at camera sites”, adding that cameras sites were delivering 100 fewer deaths per year.

Cameras first appeared in 1991, and revenue from court fines and fixed penalties went to the Exchequer. However in 2000 a trial scheme saw local councils keep the fines to pay for the cost of speed and red-light cameras. The pilot was so successful that, in 2001, legislation enabled this to happen across the country. The cost of providing and operating cameras moved from the local authority to the law breaking motorist.

The golden age of the speed camera had begun.

There was a tweak to this legislation in 2007. Fines reverted back to the Exchequer’s piggy bank. The DfT switched to funding cameras through a road safety grant. The intention was to create a greater mix of road safety measures agreed between local authorities and the police.

The number of people killed on British roads in 2007: 2,946

The number of people killed on British roads in 2010: 1,857

So perhaps the creation of the Road Safety Grant had a significant impact.

The second question: why did the death toll stop falling?

In 2010 I was unaware of Labour’s target to halve deaths on the roads. But, the change in government was enough for me to predict that the fall was over.

When the Tory/Lib Dem government negotiated its way into power in May 2010, the press declared that it was the end of the horrible nanny state – a return to personal freedom, liberty and the rule of common sense.

The way that this was to play out in real practical terms was on our roads. The evil speed camera was in the firing line. The narrative was that these cameras were just there so councils could extract cash from the poor public. Completely ignored were the facts that the fines were only handed down to dangerous, law-breaking drivers, and that councils no longer got the cash from fines.

Soon after the election the coalition government said that “Labour's 13-year war on the motorist is over” and pledged to scrap public funding for speed cameras. The Road Safety Grant to local authorities was cut from £95m to £57m. This meant that the government was now receiving an estimated £40m more raised in fines than it was spending on road safety. The cut to the grant reduced the camera maintenance budget by 27 per cent. It removed all the funding for new cameras, speed humps and other safety measures.

And the golden age ended.

Councils across the country announced their change of policy. Oxfordshire County Council switched off its speed cameras on 1 August 2010. Money was saved; lives were lost.

Eight months later, on 1 April, Oxfordshire’s cameras snapped back into life when the council reversed its decision because deaths on the county’s roads had immediately increased.

Turning off speed cameras sent out the message that we were no longer taking speeding seriously. The road safety campaigns changed their focus. The message that Speed Kills fell away and was replaced by drink- and drug-driving messages. It’s easy to miss that these campaigns move from encompassing virtually every driver to targeting a minority. A switch from confronting a socially acceptable behaviour to re-enforcing something already unacceptable. The state is no longer challenging everyone to be safe – only the small minority of bad people.

Yet speed still kills. The World Health Organisation states that an increase in average speed of 1 km[h typically results in a 3 per cent higher risk of a crash involving injury, with a 4–5 per cent increase for crashes that result in fatalities.
The majority of safety measures installed before 2010 remain in place and are saving lives. But with the funding gone councils are no longer installing new measures and the death toll is no longer falling.

So you can make a strong case that the pattern of road deaths was the result of government policy.

Which begs the question of our government: why has it accepted that it’s OK to kill, or be killed, on our roads?