Autumn is traditionally a season where suburbs trains have problems. There are a few reasons for that, including leaves on the rails (I found an official flyer explaining the process of which I’ll talk someday).
I’ve come to notice that very often, once there’s an incident on a line (someone using emergency signal thus causing a train to stop), other trains are unscheduled. Indeed, I think this is worsening the situation (escalation of incidents) resulting in possibly more incidents, up to a point where the traffic being stopped for too long, people who are blocked in a train open the doors and walk on the rails, thereby causing the whole traffic to Paris (yes, I’m in France) to come to a total halt for obvious security reasons.
Until this extreme situation (that happens once or twice a year), there are intermittent problems that the attached Systems Dynamics diagram tries to clarify. For my knowledgeable readers, it’s two intertwined archetypes: a “fix that fail” with a “shifting the burden“. Here’s why.
- R1: First, there’s a train incident that cause trains to be late. With late trains, there is an increase of people waiting on the platforms to board the next train. Of course, the more there are people waiting, the more there is a risk of incident in the next trains, thereby increasing the number of train incidents.
- B3: When trains are late, that increases the perceived complexity of traffic because the whole schedules have to be changed. So of course, an immediate and symptomatic answer is to reduce the number of trains in circulation thereby reducing the perceived complexity of traffic. This response is somewhat natural, but as we see next, it is a “fix that fail’ archetype.
- R2: by reducing the number of trains in circulation, there is an opposite increase of the number of people waiting on the platforms for next trains, thereby increasing the risk of train incidents: we’re back into R1. The intended fix failed, thereby worsening the situation!
So we see that this first loops (R1-R2-R3) form a fix that fail archetype. Now I want to show how that situation is perpetuating itself through a “shifting the burden” archetype. Let’s continue the investigation:
- B4: when the perceived complexity of traffic increases, so does the learning to manage complexity, which would, after some delay, decrease the perceived complexity of traffic.
- R5: of course, with an increase in learning to manage complexity, the number of train in circulation could increase, thereby decreasing the number of people waiting for the next train and then reducing too the risk of incidents.
B3-B4-R5 form that ‘shifting the burden” archetype where there’s a strong incentive to reduce the perceived complexity by reducing the number of train (short-term, symptomatic response) which reduce the possibility of train controllers to learn how to manage complexity (longer term, better, response). Further, that short-term, symptomatic response is a “fix that fail” in that it worsen the situation by increasing incidents, thus trains late, perceived complexity of traffic and thus increasing pressure on controllers to further reduce the number of running trains.
Hopefully, there’s an unintended beneficial consequence for myself: being blocked in a train is free time to read more systems thinking books!
What to do? Well, I think one of the leverage points resides in the traffic controllers increasing their learning from complexity, but they would need to be aware of the situation first.
Also, I haven’t modelled the security measures that further makes controllers wanting to reduce the number of trains, but it’s acting in a similar way as B3. Yet, for the same reason, more people waiting makes for more incidents and thus a decreased security.
To me, the solution to less incidents (thus improved security) is to have more trains, which would mean more complexity, but traffic controllers would get a chance to learn from it, thereby making them able to sustain a dense enough traffic in case of incidents.
I’ll try to have that essay covertly sent to Transilien for their consideration…