Blog

Transport2035 - Or How I Learned to Stop Worrying and Love Assumptions

This post from Jesse Prendergast, a Transport Planner and Analyst at MRCagney, was originally published on Linkedin.

Predicting the future isn’t easy. Modelling complex systems isn’t quick. Yet tackling the climate emergency requires decisive action based on an uncertain future. How can we resolve this paradox together?

In the summer of 2019, Paul Winton from the 1PointFive project stood in our O’Connell St offices with some gloomy news. If Auckland carried out all the transport projects and policies currently in the pipeline, it still wouldn’t be enough to meet the targets we needed to help keep global warming below 1.5 degrees. In fact, even with the planned expansion of our rapid transit network, it would barely make a dent. What’s more, Auckland (as the largest city in New Zealand, and with the most capacity for high quality public transport and cycling) will most likely need to do more than meet the target if New Zealand as a whole is to achieve its goals.

In a room full of sustainable transport planners, this was somewhat of a shock. If this was true, then people needed to know. Over the next two months, Beth Schuck set to work building an alternative model to Paul and came to the same conclusion: we simply weren’t doing enough. And so, Transport2030 was born – an interactive tool where users could select transport projects and see how it affects emissions in Auckland. Since then we have developed similar tools for Christchurch, Waikato, and Victoria (Australia). 

Shortly before the infamous Auckland Lockdown of 2021, Waka Kotahi NZ Transport Agency commissioned a nation-wide version of Transport2030 to look at emissions in the year 2035. In a stroke of genius, we called the tool Transport2035 (T2035 for short).

The Purpose

At its core, T2035 is designed to help bring the transport industry and decision makers together, with a common evidence base, to discuss the future of transport in New Zealand.

Transport planners love information. Data, statistics, and models – I know I could look at them all day (and many people do). Across New Zealand there are many transport models that serve a vital role in the detailed planning of our future. However, when taking these models to decision makers, there are a few key issues:

1. They are complicated – big models take a long time and a lot of knowledge to set up, so they cannot be run as frequently as we might like.
2. Results are complex – there are many great insights generated, often in large spreadsheets.
3. Summaries take time – to properly take the outputs of these models and turn them in to easily digested information takes expertise and time.
4.  Not everyone has one – these models take time and resources to build. Time and resources which not all councils have access to.

To get us all on the same page quickly T2035 needs to be simple, easy to interpret, fast, and available to the whole country.

The Challenges

T2035 is the start of a suite of works that Waka Kotahi has planned in the Emissions Modelling space, designed to help guide the country to a low carbon future. As such, it needs to be a good entry point for new councillors and the public, while also scratching the itch for transport planners who want to jump into the details of future transport projects and their impacts on transport emissions.

This means T2035 needs to meet some key criteria, a few of which I’ve outlined here.

• Intuitiveness – Transport2035 is primarily meant as an engagement tool, and nothing turns people away like a complicated set of numbers, graphs, and long bodies of text. T2035 needs to be able to be picked up out-of-the-box and easily navigated.
• Flexibility – No region of Aotearoa is the same, and T2035 needs to be used by all of them to test a wide range of options. Prescriptive options or limited functionality won’t cut it.
• Current information – To make sure that T2035 is as useful as possible, it needs to make use of the best, most up-to-date information we have on transport in Aotearoa.
• Robust information –The data used by T2035 needs to be consistent across the whole country, to ensure that the tool reacts predictably for the user.
• Clarity - T2035 also needs to produce results that are uncluttered and easy to understand for a user of any background. Most importantly, it needs to show the size and scope of all the changes required to meet our emissions targets.
• Complexity – T2035 needs to allow experienced transport planners to communicate complex ideas in a simple way. We need to bring both parties to the table to achieve results.
• Speed of delivery – To meet the original deadlines for this project, a prototype of T2035 needed to be created in just six weeks, with a release ready version incorporating user feedback ready in 12 weeks. We needed to build it in a way we could easily tweak, while keeping the core calculator robust. (I’m delighted to say we achieved this. But hey, we were in lockdown - what else did we have to do?)

Some of these criteria are directly at odds with each other:

• Clarity vs Complexity. How do we build a tool that captures the integration of so many factors (e.g. an uptick in working from home, electrification of our vehicle fleet, or the impact of denser urban form on travel choices) while keeping a non-technical audience engaged and confident?
• Current information vs Robust information. How do we make sure T2035 presents a current picture, while accounting for the fact that not every region has the same data available?

These tensions exist in so many data science applications, and there is no one right way to solve them. The answer is: it depends. What is the key purpose or your tool? For T2035, we had that defined:

T2035 needs to be simple, easy to interpret, fast, and available to the whole country.

The Solution – Transport2035

Using the purpose of the tool as a guide, we made decisions on the user inputs, background data, design, business logic, and outputs to arrive at the final product. Some key decisions included:

• A focus on simple user interface (sliders and reduced text) to keep the dashboard uncluttered and accessible.
• Allowing users to play will a full range of scenarios (even if they are a little, or a lot, unreasonable), this is a conversation starter. Want to put everyone on a bike, or a bus? Be my guest!
• Keep the data consistent, even if it sacrifices accuracy. The purpose of T2035 is not to give you a detailed plan, but to quickly point you in the right direction. Let’s get everyone in the same ballpark and let’s get there fast.

Try Transport2035 for yourself and see if you think that the design and the tool meet the goals I’ve described!

The Aftermath: What Have I Learned?

The development of Transport2035 has been valuable, not only to the creation of what I hope to be a useful tool, but in teaching me how I can build the next one better.

• Agile development is important – no amount of sitting and thinking will get the thing done, and you will NEVER get a complete list of requirements ahead of time.
• Build it, break it, and improve it. Better yet, let other people break it – they can do that faster and better than you.
• People have knowledge you don’t – give them the opportunity to provide feedback by releasing or sharing your work early.
• Sometimes less specific is better, especially when doing high-level strategic analysis.
• Assumptions are okay if they are clear. All modelling contains assumptions. Not all of those assumptions are good. However, if you make them clear to your audience, then they can make their own decisions on the usefulness (or not!) of your work. In the words of George E. P. Box: "All models are wrong, but some are useful."

Transport2035 was a lot of fun to develop, and I’ve learned a bunch about emissions modelling, emissions reduction, and data communication. I’ve learned that Aotearoa has a long was to go to meet our targets, and to do this we’ll need to pull every lever we have. Most importantly, I’ve learned we won’t get there unless we’re on the same page, and we’re on that page soon.