Value of intuition

[Martin Cassini]

On BH, Stephen Bayley said something along the lines of, "Measurement is a modern obsession. They say if it can't be measured, it doesn't exist. Piffle! How do you measure culture?" My experience of efficient interaction between road-users when lights are out of action is dismissed as irrelevant by traffic engineers. If we lived by measurement alone, we wouldn't achieve the change that intuition can inspire. By all means apply science for evidence, but appreciate the punch of the hunch.

[Martin Cassini]

Response from Keith (first of a few):
K: Accusations that traffic engineers are brain-washed and concrete-minded is one reason we choose not to confront armchair experts. If those outside the profession read a traffic engineering textbook they might see there is a science behind it which requires something beyond anecdotal evidence of specific observations at particular sites.

In the case of locations where lights fail, in London alone in the last 3 years there have been over 150 casualties. Interestingly, most of these occurred at night. When we compare the length of time that lights are out to when they are working, there are a disproportionately high number of accidents at these sites as a result of uncontrolled behaviour. If it wasn't for these accidents many of these locations, the junction would have experienced a lower than average accident rate. It is therefore blindingly obvious from this EVIDENCE that the arrangement is not safe, despite the anecdotes from the thousands of people that manage to get through the junction without harm. I do not expect the casual road user to carry out this research, but they might respect the professionals who do it for them, and who then spend time and effort attempting to mitigate the problem, while at the same time making the junction, crossing or network as green, sustainable and resilient as possible, while attempting to discourage use of the private car, and provide a simple, aesthetic, negotiable (particularly for visually impaired and children), value for money solution.

Research I read recently showed that around 50% (if not more) of drivers regularly break the speed limit, yet in shared space environments all drivers will need to reduce their speed to near walking pace to avoid pedestrians feeling menaced - between 3-5mph or 5-8kph. Could anyone guarantee that this would happen 24-7? You have to wonder why drivers would choose to drive at walking pace (why not walk!), unless of course these environments were few and far between, at quieter locations with a greater focus on the pedestrian and urban realm.

This is fine, indeed UK standards do not prevent uncontrolled junctions from being introduced at any site - as a consequence, many are.

As the Director of traffic and highways at a London borough has stated, his decisions on network control methods will be with him for life - not just the duration of his tenure. If someone dies on his roads (and statistics show that one or two will every year), to avoid a jail sentence he might be called upon at Inquest to demonstrate that he was not negligent. I'm sure there aren't many armchair experts that could genuinely make that call without the knowledge that their judgement could stand up to such scrutiny. It is simply not sufficient to argue that he thought it would be OK.

The traffic engineer is interested in a number of outcomes, not just safety but network capacity and journey time, to name a few more. It's no good looking at a single site; whole journeys with the compound impact of delay must be considered. 'Society' will only change to some miraculous and 'never-before thought of' approach to junction control if it can be demonstrated to local and national governments that 'shared space' or filter-in-turn is, in every case, at all times and in all possible ways, significantly more 'efficient' than the well-developed system of priority rules and signal control. Thus far, this has not been achieved (indeed the development of this system from a base of no controls suggests quite the opposite) and so it is likely that schemes will only be introduced at sites that are somehow considered to be low risk.

With modelling, we are not attempting to reproduce the precise processes that lead to the actions. This would be desirable, but would require the computing power of a HAL or Deep Thought. Instead, we have to make do with the closest representations we can muster. We could consider average conditions, but this might not be enough to cover enough eventualities to be safe or reliable, so we look at sensitivity to average or even 85%ile behaviour.

The comments of those questioning the strength of calibrated and then validated modelling techniques demonstrate very clearly the problem with anecdotal evidence and armchair expertise. Without it, we are left with only demonstrating that a particular action/ response was good for that particular moment, by that particular person (or possibly those around them) and so does not provide any basis against which to design. The unscientific approach leads to a very expensive and inefficient alternative of suck it and see at every new location.

You might think that factors such as human behavioural response make it impossible to achieve accurate models, but don't be fooled by how remarkably predictable we all are. Even extraordinary behaviour is predictable. It's not difficult to build gap acceptance models for vehicular and pedestrian behaviour based on simple rules - these have been around for over 50 years. We just need to refine these, to enable very accurate forecasts of junction capacity based on composition and volume of different modes. As a traffic engineer looking to find optimum network-wide solutions that are safe and as efficient as possible, my work is far more convincing if it can be demonstrated to be based on evidence, rather than hearsay.

We can now very simply, yet very accurately, model behaviour at standard regulated junctions in 3D and real time. I have used this technique to assess how well a shared space junction performs with identical levels of traffic and pedestrian demand. It seems to be showing that we might expect uncontrolled junctions to have less capacity and more delay. If this is the case, and they are no safer - why bother? We need data and evidence if we are going to take this any further than chat rooms.

[Martin Cassini]

Nicely put. I will ask Keith to weigh in with his comments. Of course that means joining the network ... He might want to remain unattached. But it would be good to see him here, and I don't see why joining would damage his role as objectifier. We already enjoy endless heated, constructive debate. He seems to me to be evolving from a traditional traffic system practitioner into a fully fit, unprejudiced thinker on the subject.

[phunksta]

That smacks of a huge misunderstanding of the scientific method.
You can't test what you can't measure - that doesn't mean it doesn't exist. In fact the ratio of things we can measure to those that exist is 1:Infinity by definition.
You can't measure how wakeful a driver is, but you can measure their speed. The first is a killer, the second is a poor substitute for a road saftey 'factor'

[Martin Cassini]

There's a particular traffic engineer who has become my partner on the JET (Junction Efficiency Trial), Keith Firth. He points out, quite rightly, that as far as hard scientific evidence goes, observation that is not rigorously conducted, and traffic that isn't counted, amounts to anecdotal evidence, and is of little practical use. Hence the trial, to measure traffic and analyse road-user interaction, journey time, etc, that can provide definitive results and robust evidence. In fact Keith, who has a background in traffic signal installation and control, is getting into these ideas. We're acting as quite good counterweights to each other and so far it's proving a constructive partnership.

[kevaquarian]

"My experience of efficient interaction between road-users when lights are out of action is dismissed as irrelevant by traffic engineers"

Why??? On what basis? If something can be observed then surely it deserves considering? I really don't get this. The only conclusion I can draw at this stage then is that the ones you speak of are brainwashed concrete-minded victims of the system - unable to break out into new ways of thinking?