Modelling World 2022, partly digested

As usual, Modelling World was an inspiration and a blast. A chance to catch-up with friends and colleagues, old and new, to share ideas and to debate about Transport Modelling. I was privileged for the second time to lead the closing debate which looks to the future, and to attend and ask as many awkward questions as I could in others.

As usual, I came away with many thoughts to mull over, and this article is a very rapid attempt to get some down before they slip away. So these are partly digested takeaways from the conference in four parts, which will hopefully link together. Let's see.

Note: More than ever these are my views and interpretations, not my employers, clients or panel members. I hope they are useful and will be delighted to debate, be corrected and adjust my views accordingly!

1. Simple Models for Local People

The Day 1 afternoon session is sometimes seen as an optional warm-up. This year I'd argue it was an essential part of the conference overall. The two sessions covered open source software and 'no-model' approaches. The latter included accessibility assessment tools, street planning and consultation aids. At one point they were referred to as needed because 'traditional models don't do active travel well'. But I think there's more going on than active travel, and some clues from Day 2 gave more hints on that:

• Paul Speirs talked about the need for 'person centric' models which present travel as real people see it. That's certainly something these are doing.
• Charlene Rohr highlighted the need for simpler, more transparent models which can be used quickly to look at many scenarios. I think she may have had 'strategic' modelling in sight, but again these local 'no model' tools I think are good examples.
• John Bates hit the nail on the head in the final session, by pointing out that 'traditional' modelling has focused on inter-urban travel, because that's where the biggest investment decisions were seen. The new tools are addressing the urban travel experience, and that involves different ways of thinking.

That last point led to a 'penny drops' moment for me, which I was able to test 'live' with the Head of TASM, Liz Jacobs:

• Traditional models and TAG are designed for schemes which need assessments of BCR, because they are costly and have uncertain benefits.
• As Liz noted, local active travel and streetscape improvements are lower cost and normally have very good BCRs.
• Therefore the question is NOT 'can this investment deliver benefit', the real question is 'can we design a scheme that WORKS and is accepted by the community'.

We have sadly seen that active travel schemes can be perceived badly, leading to negative headlines and opposition which can discredit the entire endeavour. That's why approaches are needed which are very visual, present the scheme as it will be used, and can be used to rapidly go through options and see impacts.

But as Charlene and Paul pointed out, these are attributes that large scale strategic models should also seek to adopt. So I think we should be paying VERY close attention to what these trailblazing tools are doing and see what tricks we can learn.

2. Uncomfortable Questions: Forewarned is Forearmed

In the opening plenary, Paul Speirs referenced a framework for thinking about change, based on the TED talk by Simon Sinek on the 'Golden Circle'. (Simon Sinek - The Golden Circle - TedTalks 2009 - YouTube). The concept is that we need to lead arguments for change by addressing beliefs about WHY change is needed first, in a way which targets beliefs and feelings. We then talk about HOW the change will be delivered, and then WHAT will change. Sinek's favoured examples are: Apple products, which people buy into as a design principle NOT an end product; and Martin Luther King who used the phrase 'I believe' and 'I have a dream', whereas modern politicians often have a ideas about WHAT to do but not a persuasive argument for WHY.

(This focus on WHY incidentally came up in something I recently wrote here: https://www.dhirubhai.net/pulse/asking-why-career-tim-gent. I was focusing on the need to ask WHY as part of any piece of consultancy or solution design, and how to work with people who know WHAT they normally do, and those who know HOW to change, both often overlooking WHY.)

As already mentioned, Charlene Rohr made a good case that the future is too uncertain for complex models which project a false precision about single scenarios. I am a big fan of the idea of simpler scenario models (and presented on this myself last year!). But again listening to both talks an issue occurred and I asked an awkward question which the panel described as an 'uncomfortable truth'. I'm not sure I put it across well, so will try again using the WHY HOW WHAT framework.

WHY: We all believe the future is uncertain, modelling scenarios is needed, and current models are too complex and slow! Agreed.

HOW: So we're going to build simple scenario models, which will show the key problems, and tell us quickly if a solution will 'solve' the problem. Great get to it!

WHAT: So, we've now quickly shown that you need to do MORE to solve this problem. What!? You'll need to improve this model...

It's human nature to ask for more evidence when we don't get a convenient answer, and no sleight is intended here. All I'm really saying is that simper models will get us to some inconvenient answers quicker (and some convenient answers too, but that's less of a concern!).

I would stress again that I firmly believe simpler models are needed, not least because they can bring us the difficult questions (examples available on request!). But forewarned is forearmed, we should not assume the results of simpler modelling will be easily accepted. The solution to that will probably NOT be more modelling, but more talking, communication, explanation, and exploration of what the question was in the first place.

3. The Volume is WHY

In the debate on 'The Future of Modelling' we asked the audience to rate 5 aspects of modelling in terms of importance and ease of addressing:

1. Simplicity: Creating simpler models (as discussed above).
2. Process: Making sure the correct question is asked, and the process of commissioning and using the model addresses this.
3. Data: Obtaining more data e.g. travel activity, real-time and big data.
4. Activities: Representing more detailed/complex travel choices (e.g. activity chains).
5. Volume : The ability to provide greater volume of models and modelling.

The selected order (for both importance and ease of addressing) was as above. We will write up the session and discuss that in due course, but I wanted here to puzzle over that last place for 'Volume', which doesn't seem right to me.

In the modellers survey (which will also need a write-up!), 65% of respondents said lack of resource increased pressure on modellers, and a similar proportion said that it leads to options and opportunities being unexplored. Certainly this chimes with my own experience, where the demand for modelling (complex and simple) outstrips by a fair margin the number of people available to di it. Looking at Fred Ewing's frequent and plaintive cries for modellers on LinkedIn tells us this is widespread, and was a phenomenon long before Covid or the impact of Brexit.

Are we perhaps reaching a stage where enough models exist, and the pressure will ease off? This seems unlikely given many large and small authorities are currently without multi-modal models, highway models need regularly refreshing, and we have yet to incorporate Covid into most (any?) models.

Will simpler models massively lower the workload? It may change the nature of the work, but I doubt this, because i) simplicity still requires a lot of work, with multiple scenarios and a huge amount of high quality communication (see above), and ii) simple models will often need a complex model for final scheme design stages.

Perhaps there is a fear that saying 'we need more models' feels like jumping to the WHAT question without addressing the WHY. Or perhaps it feels overly self-interested to focus on the massively high demand for modelling?

I would argue that the demand for modelling is huge, and is unlikely to fall any time soon, and that is telling us something critical. There are big questions which we need strategic, local, simple and complex models to address. That is WHY we need to produce simpler models where we can, that is WHY we need to get the process right and address the right question, and WHY we need lots of other things.

The Volume is WHY because it represents the demand for modelling, which is the first thing we need to understand.

4. Oh, and the Census

I also had some heated debates about the future of the Census JTW data, given the widespread (and probably correct) anticipation that the 2021 Census Journey to Work data will be .... less helpful than we'd like. Many of us filling in the Census will have found ourselves stumped about how to answer the Place of Work and Method of Travel questions, either because we did not know if our behaviour in April 2022 would persist, or because if we split our time 40:60 between work and home, there is no 'typical' answer.

It's worth mentioning of course that this may be a biased view from professionals able to work remotely in comfy home offices. I've heard it suggested that we only represent 30% of workers (see Tom Van Vuren's recently shared link on bias amongst planners https://www.dhirubhai.net/feed/update/urn:li:activity:6942075833612181504?updateEntityUrn=urn%3Ali%3Afs_feedUpdate%3A%28V2%2Curn%3Ali%3Aactivity%3A6942075833612181504%29).

So one could argue that 70% of the JTW results will still be there, and that this will tell us something REALLY useful about who can work from home, and who HAS TO travel for work.

I also made the suggestion that the 2011 Census would still be of use to indicate likely spatial variations in mode choice and travel patterns, and that 2021 Census should obviously be compared to that. Always look at the change over time first, and try to understand it. Indeed, this is my first 'go to' whenever I see modelled or observed travel behaviour.

Fans of big data challenged me on this because '2011 is too long ago', 'commuting is only 20% of trips' and 'mobile data can provide us with reliable and up to date information'. Well, I'll concede if I want to look solely at inter-urban travel patterns for car vs rail then I could do worse than purchase mobile data. But I would still be comparing it against Census JTW, and maybe a gravity model, to sense-check.

As for 2011 being a long-time ago and commuting being a minor journey purpose, I think this misses the point that most travel behaviour varies more spatially than it does over time or by purpose. If I want to know which areas have high car dependence in 2022, then 2011 Census is not a bad place to start (and it's free!). I can use trends in traffic counts and car ownership (plus population and jobs) to think about what might have happened since.

Long distance and inter-urban car trips are always going to be really important (they generate most of the vehicle kilometres after all), but urban travel is now a major focus, and I don't see that any data source gives us insights into local of variations in the way the Census does. This is a debate I expect to continue, and look forward to it as the 2021 Census data emerges.

ENDS