Designing the Route - Theoretically

 

Designing the Route - Theoretically

Now you know where this crazy idea came from. But once it crystallised in the head and a decision was made to do it, it was no more crazy and random. Google Maps was opened and various cantons were mapped. Zoom in and find out which are the intersections lines between cantons and which are the closest stations to the canton boundaries. Zoom out to orient myself at a larger scale. Then the SBB app came into play. Finding out long distance trains touching all the cantons. Using the excel sheet to start typing the start and end stations, arrival and departure times, the platforms, changeover times and so much more. Go back to the map.

There were certain constraints that I had set for myself. Firstly, I had to start from Morges, where our daughter lives. We were reaching back from a short trip to Belgium the previous night at about 10 pm and I had only the next day, a Friday to try this out. Secondly, on weekends, many trains have a lower frequency and did not support this attempt. So Saturday & Sunday were out. Early morning on Tuesday was our flight back to India, so Monday was also not possible. Finally, I was not attempting a world record. I just wanted to see if I could try to get as close to it as possible with just my knowledge of the country and its trains. 

And that is when I found out about TSP - The Travelling Salesperson Problem. It is a classic, optimisation math problem - "Given a list of destinations and the distances between each pair, what is the shortest possible route that visits every destination exactly once and returns to the origin city? So my fun activity was already turning out to be a computational issue. Just an aside - TSP is used in real-life applications in Logistics and Warehouse operations. Now I was trying to apply TSP to a public transit graph and basically I had become a transport network optimiser.

The canton selection logic had to be fine tuned. Multiple options were considered and most shot down. Some saved time but missed cantons. Some backtracked adding to time but helped achieve the canton touch down. 

As I was searching for some ideas, I came across the Swiss Train Challenge website which I mentioned earlier. I got to know how that in 2018, École Polytechnique Fédérale de Lausanne (EPFL) researchers Emmanuel Clédat and Dirk Lauinger developed an algorithm to set foot in all 26 Swiss cantons in under 24 hours using only public transport. Their research was to convert the journey into a linear optimization problem and try beating the existing record by 25 minutes with a time of 16 hours and 54 minutes. 

The EPFL team was inspired by the journalism-launched Swiss Train Challenge started by RTS journalist Nicolas Rossé in 2015 that required maximizing tight SBB and postal bus connections across Switzerland's highly synchronized, decentralized rail network. Finally when they did try, their actual real-world attempt hit a delay out of Neuchâtel and finished closer to 18 hours. Initially it seemed like a good idea for me to just copy the tested route but what is the fun in that I thought.

So I planned my own. The more I dug in, the more I realised that to touch all 26 cantons, I would have to start from one edge of the country and finish at the other. And with the constraints of starting and ending at Morges which lies in the middle of the Geneva - Lausanne route, I knew I could not do 26 cantons in 24 hrs. So the next best option - try and do as many as possible. 

The first try got me as many as 21 in about 17 hrs. More cuttings on paper, more stations were visited, the map was revisited multiple times, till I felt I had got the magic figure of 24 cantons in less than 19 hours starting and ending at the same place. I still do not know how many times I reworked at nights even when I was travelling in Belgium. And finally, I realised that the plan I had frozen was the most optimal, within existing constraints.

Now with ChatGPT, Gemini and Claude, I stress tested my route outline. Each one of them came up with different ideas. All the three models found my idea not suitable because it was definitely not optimal anyway. Every time I would add the prompt that I had to start and end at Morges the plan would collapse. 

The AI models did not have to answer my wife and daughter. Only I had to. 

But working with these helped me refine my plan further. It pointed out some errors in my plan which really helped. It helped aggressively target the most restrictive bottlenecks to minimise backtracking. I had to abandon certain aesthetic and touristy routes. I did not consider ferries and buses thinking them to be too slow.

Redesigning The Loop

Now having completed a weak attempt, I came across multiple thoughts enroute which are more than likely to help me actually try the 26 cantons in better time. If I was to ever try this crazy stunt again, I will definitely consider some of these points. 

To achieve a sub - 18 hr time, I have to make multiple changes. Start from one end of the leaf node in Geneva or at Bellinzona to avoid long back tracking. Valais as a canton forces one to get cornered and we need to find a way out of that. Cross platform connections at small stations must be considered to avoid visiting major hub stations like Basel, Zurich and Bern, twice since transit within stations itself eats up time.

Wherever, there is an out and back journey, buffer time for them or else immediate return by bus must be considered. Ferries where feasible must be planned if they help cut some time. 

What finally mattered

In the end, the statistics tells me a lot. I alighted in 24 trains - 8 Suburban, 3 Regional, 7 Inter-Regio, 5 Inter City and 1 Euro City. I covered a distance of roughly 1,250–1,300 km in just under 19 hours with the double Gotthard crossing alone accounting for over 300 kms of that total, close to a quarter of my entire day's distance, just from that one there-and-back to Bellinzona. I spent 3 hr 26 mins either running or waiting for trains at the various stations. I completed the journey with a First Class Saver Day pass of CHF 99 and spent about CHF 20 more to buy some coffee and food.  If I had tried the same with no Saver Pass, it would have cost me around CHF 450 in Second Class and nearly CHF 950 in First Class. So having done it at one-tenths the cost confirms why this trip only makes sense with a pass.

It all started as a naive plan but as the days progressed it turned to a well-informed one. Finally after the journey was completed, one could draw real-world lessons that would definitely reshape the next attempt - if I have the will, energy, and intent to try.  

In the coming blogs I will tell you more about the trains, the engineering marvels and rivers and lakes which one would pass through during a trip in Switzerland.