Transport modelling: art or science?

Best of friends: Rob Bain (left) and Tom Van Vuren (right), both wearing “Transport modelling is not science” badges. Bain kept his on, but Van Vuren soon took his off!

Best of friends: Rob Bain (left) and Tom Van Vuren (right), both wearing “Transport modelling is not science” badges. Bain kept his on, but Van Vuren soon took his off!

Art or science? That was the question that provided the basis for a firey – but ultimately good-natured – discussion at last week’s Modelling World event in London.

Speaker Rob Bain, a consultant who specialises in advising investors whether or not to invest in transport projects, made no bones about where he stood on the topic, wearing a badge proclaiming ‘Transport modelling is not science’.

To illustrate the point, the forthright Scot presented data showing how badly wrong demand forecasts for Australian toll roads projects had been.

“It would be very attractive to regard transport modelling as a science because there’s a whole package of implications that roll out from there,” said Bain. “It holds out the possibility of theories of travel behaviour, models of travel, as comprehensive, perhaps as powerful, and with the predictive capability of those that exist with physical phenomena like boiling water. We know water boils at sea level at 100°C, it does this all the time.

“The difference is molecules of water have no alternative,” said Bain. “I simply do not believe that the same exists in transport. We do not have laws of behaviour as such and it’s because of this issue – people. We’re infinitely more complex and have an almost limitless ability to adapt and innovate in response to environmental changes.”

Bain said that, because of this, error margins in transport forecasts would always be large. “This huge range of error is entirely expected,” he said, pointing at his Australian toll road data. “The errors should come as a surprise to no one.”

But he said modellers were too reluctant to talk about the uncertainty inherent in their forecasts. “The fact of the matter is you have senior members of your profession out there actually selling levels of precision about forecasting that simply do not exist to please their clients. Too often we see the modelling world become the salesman of projects and this does you as professionals an enormous disservice.”

Bain said the problems with forecasts weren’t entirely technical. “You have to look at how models are used and who they’re prepared for. I was a modeller myself for a number of years and I know that pressures are placed on modellers to come up with figures to fit a financial model rather than necessarily an accurate portrayal of what they believe future states of the world might look like.”

Panellist Tom Van Vuren, divisional director of Mott MacDonald, provided a modeller’s response to Bain’s critique. “One of the key elements of what we do as modellers is to take a scientific approach to estimating models, to build repeatable experiments and to build models that actually are based on an amount of theory and an amount of science. That definitely is science and I think we should not lose that.”

Van Vuren distinguished between the models “which are actually not that bad” and the input parameters such as forecasts of GDP or land use development patterns. “If we did a little bit more soul searching about the reliability of our inputs to our forecasts then I think we could go a lot further,” he said.

The exchanges sparked numerous questions and points from the audience. John Burgato of Atkins thought that input parameters would always be a problem for modelling. “Beyond the short-term timescale there is no way you can say even what the errors are coming into the model,” he said.

Bain then presented findings of an online survey he had conducted with the modelling community. The majority of respondents had said that modelling was an art rather than a science, but opted for “both” when given the chance.

Not everyone was impressed by the survey. Neil Chadwick of Steer Davies Gleave said the science-art debate was “completely irrelevant”. “We shouldn’t detain ourselves with this nonsense,” he huffed.