Why model, when you can experiment?

The junction outside the Bank of England is truly awful; a vast open space of tarmac, motor traffic thundering through in five directions, and pedestrians accommodated on tiny pavements. What should be a beautiful civic space is devoted to motor traffic flow.

From Google Streetview

From Google Streetview

To be fair to the City of London, they have recognised the problem, and are looking to make improvements. It seems they are examining the potential for closing off motor traffic from certain directions, or at certain times of day.

But here’s the method they are choosing to employ for examining the options –

At the moment we are establishing how wide the impact might be if we make big changes at the junction. This will give us the starting point of what we will need to look at in detail. We should complete this work by September 2014.

Our next task will be to build a computer traffic model to assess what is likely to happen if traffic is prevented from crossing the junction for example in certain directions or times of day. Information from pedestrian and cycling movements will also help to develop solutions. This is likely to be a big piece of work and will take some time to complete but it is very important to have credible options for alterations to the junction. We hope to have this work completed by early 2016.

They are building a computer traffic model to do so – in their own words, ‘a big piece of work’ that is going to take one and a half years to complete. Eighteen months. There is no word on how much this is going to cost.

I imagine the complexity here is due to the fact that we don’t really know how to model people cycling and walking, as described in this excellent post by smalltown2k. It’s really very difficult, and the City appear to be attempting to do so. Now obviously the ability to model these kinds of movements is going to be very important in the future, and it is valuable that we can start to assess what might happen to traffic flow if we acknowledge how people walk and cycle about, and how they might shift mode under different conditions.

But really, rather than just building a hugely complex model from scratch to find out what happens when a junction is closed to motor traffic, couldn’t the City just do it, on a trial basis? If the result is genuine chaos, then the trial can quickly be abandoned.

There are good reasons for thinking a trial of this kind – closing roads at Bank temporarily – would not result in chaos. The main one is that the area is ringed by major arterial roads, composed of London Wall to the north, Aldgate and Tower Gateway to the east, and Upper Thames Street to the south.

Screen Shot 2014-07-18 at 12.57.31

All are designed to carry large volumes of motor traffic, and all lie very close to Bank itself. These are the roads that should be carrying through traffic; the area around Bank should, realistically, only be carrying private motor traffic that is accessing the area. Certainly, the Bank junction should not be carrying through motor traffic in an east-west direction, as there are two major roads to the north and south – just a few hundred metres away – that were built for this purpose.

So – why not just try this? Try it now, rather than spending eighteen difficult months building a model from scratch. You’ll get results that correspond to the real world, and much more quickly!

This entry was posted in Uncategorized. Bookmark the permalink.

14 Responses to Why model, when you can experiment?

  1. KristianCyc says:

    “Keep It Simple, Stupid” is a motto often heard in the city’s software houses. It acknowledges that despite the magnificence of software and all it has made possible in the world today, technical solutions are often complicated, costly and can fail completely, so they should be considered last.

    Thus if you can get what you really need without it, then stick with the simple solution.

  2. KristianCyc says:

    I should also add though that Upper Thames St is due to become “crossrail for bikes” so will also be currently undergoing analysis for traffic reduction. Of course, they could simply reduce it to two lanes (from four) as part of your suggested experiment. This should be straightforward as they’ve already tried that when introduced the olympic lanes

    • Arwyn says:

      “All models are wrong…although some are useful” An adage those who build models use all the time, but something which is seldom taken on board by those who attempt to inteperate them.

  3. Paul says:

    Gilligan , just after he was appointed, said that he would like to see more experimental schemes put in on a cheap trial basis. The TfL “can’t do that” attitude seems to have won together with an unjustified faith in computer models as Arwyn points out.

  4. rdrf says:

    I think there is a long history of highway engineering being done on some sort of “trial” basis: if you don’t get the right results, then the scheme is revised or scrapped.

    However, you do have to be wary, not least in terms of what the results are that are being used as indicators of success. It would be quite possible for an increase in cycling and/or pedestrian casualties to be used an indicator failure even if , with a big increase in cycling and walking, casualty RATES (per journey, time cycled/walked r per pedestrian crossing) have declined.

  5. Andy R says:

    If TfL ‘just did it’ presumably they’d still have to do a major monitoring exercise after implementation to see where motor traffic was redirecting to (or if, in fact, it was disappearing) – no small study, and possibly approaching the expense of creating a model. Just doing it also risks unintended consequences – something that could cost the economy like London’s billions, however short-lived the experiment. At least by creating a model up front it will allow for a much wider number of scenarios (not just closing the immediate roads) to be looked at – both now and into the future.

    • Jitensha Oni says:


      But doesn’t anything anyone does risk unintended consequences? Applying the results from a poorly constrained model for example (see Arwyn’s comment and smalltown2k’s piece) . That’s not a valid reason for either activity or inactivity.

      In any case one thing we do know is that the average speed of travel in London has been pretty much constant for over a century, which kind of suggests a steady state independent of anything in traffic models, and robust to perturbations. A general model would be useful for the future, once, as smalltown2k says, cycling and walking movement databases are up to the job, but let’s not pretend it’s going to work here.

      Anyhow the key thing for me is that people mostly responded to the initial CoL consultation by saying that they wanted improvements for active travellers with a few nice plants. This has somehow morphed in the offices of the high council into “improve function and safety for all modes of transport” where “Information from pedestrian and cycling movements will also help to develop solutions.” Which means the active travellers are now subsidiary to the main aim. Once you start arbitrarily “shifting the goalposts” in this manner, then the model is not being constructed to optimise what people have said they want. What counts?

      PS Putney Bridge will be closed for months. That may have lessons for the Bank project. Is anyone collecting data?

      • Michael J says:

        There have been various roadworks/building works on the roads leading to and from Bank for ages, and I bet they didn’t bother much with models there. All they need to do is put some cones down and see what happens for a day or two. They could do it in the daytime after rush hour if they’re really worried. Most of the traffic around there is cabs, minicabs and delivery vans, and they’ll soon adjust to route changes.

    • Paul says:

      All of which begs the question of how meaningful a model can be. In chemical engineering, yes; but the behaviour of drivers, cyclists and pedestrians is somewhat more difficult to predict than that of molecules. I would hope that traffic flows are being measured anyway so untoward changes should be obvious.

      • michael says:

        I’m skeptical of the entire category of ‘models’ if what is being modeled is dependent on social behaviour rather than purely physical systems. The poor predictive power of models in economics being a rather obvious example.

        When it comes to transport – presumably there have been meta-studies evaluating the success or failure of such models in general at actually predicting what would happen? (Though even then one couldn’t be sure about counter-factuals, that is, whether the model was correct about the outcome of the options that were not, in the event, chosen). Assuming such studies have been carried out, does anyone know what they found?

  6. Tim says:

    As long as the limitations of a trial period are recognised.

    I think it’s entirely possible that the kind of infrastructure changes I’d dearly love to see could result in some habitual car-commuters discovering that their commute is taking longer, and perhaps that a more pleasant direct car-free cycle-route has become a viable option. Perhaps they might even consider selling their car, and so on. But these changes would take far longer to filter through than the period over which a trial would take place. And these changes may also be dependent on other new infrastructure elsewhere (ie more of a network). Perhaps a good model (if such a thing exists) could attempt to factor in longer-term changes in a way that a one-month trial (for example) could never hope to.

    I’m still discovering new cut-throughs and more cycle-friendly options in my area for journeys I’ve been doing for years.

  7. Nico (@NicoVel0) says:

    The attitude at TfL beggars belief. There is a country-wide trial across the North Sea called the Netherlands, and London itself ran a big traffic reduction trial 2 years ago called “The Olympics”, and life carried on as usual. A mayor with balls would have kept the Zil lanes at the end of the Olympics to see if London could cope when school started again, but instead we have Boris who thinks pollution is bollocks. The number of opportunities missed and lack of courage are criminal.
    Also call me cynical but I expect the model will be carfully chosen to give the results TfL want. Remember of their old model counts a taxi with no passenger as 1 unit, but a cyclist as 0.5?

  8. Nico (@NicoVel0) says:

    PS Tim, I’m actually annoyed that I have to try and test and experiment to find a route that is safe, not too unpleasant and not too indirect when I’m on the bike (or walking for that matter). When I drive I just drive, maybe with Google maps directions, and I get that. Everything but the car is second class in London.

  9. Pingback: Trying it out | As Easy As Riding A Bike

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.