Beginning last year, the Montreal metropolitan area begun the deployment of OPUS cards throughout the public transit system. The process was a long and painful transition during which half the gates to the metro stations were using the new system and half were still on the old one, resulting in longer than average queues. At the end of June, they finally completed the transition and stopped selling the old tickets. I should say almost completed, because you can still come across some of the old gates.
The concept of the OPUS card is fairly simple. Nothing the world has never seen before. An RFID card that can be used in the various public transit systems of the metropolitan area. The promise was interesting. Carrying a single card for all public transit needs. Unified, automated machines to purchase tickets, removing the need for interaction with other human beings.
It all started fine. Acting like a laggard, I waited until January to get a card. I could then get my monthly pass for the Montreal island, which served on a day to day basis. I then got train tickets to visit my family once in a while. Everything worked fine thus far. A few months later, they modified the train schedules and they were no longer optimal for my needs. I figured out the good old bus would do a better job. I then tried the machine to get the tickets for the bus. It wouldn’t let me purchase them. It felt counter-intuitive, but I made the line to speak to a human being. The woman did not know what was going on either and searched for a good 10 minutes to figure out what was going on. I could feel a queue of people behind me growing impatient.
She finally gave up and gave me a â€œsoloâ€ card, which is a single-use RFID card containing the tickets. The 1-card dream was over. I was already stuck with two. Worst part is, they both have conflicting signals, so I now had to pull out the card from my wallet to use it. So much for using RFID.
Time came by and this mitigated success held for a few months, that is, until I moved to a more central area. For the first month, I still had my monthly pass, but quickly figured out I was only occasionally using public transits because most of the places I need to go are within a walking distance. Having traveled for the major part of August, getting a monthly pass was certainly not worth it anyway.
So I went out to get tickets instead, which are sold in packs of 10. Went to the machine. Once again, the option was not available. Once again, I reverted to speak to get in queue to talk to someone. The lady told me there was a problem with my card and it had to be â€œprogrammedâ€ (one expression that makes me cringe when coming from a non-programmer) and that I had to go to the Berri-UQAM station to get that done. Quite off my route for the day, so I kept that for later.
A few days later, I did stop by the station and went with my card, asking for 10 tickets. I did not mention the previous story with the card having to be â€œprogrammedâ€. That did not feel right. It turns out I got a different response. There was a conflict between the types of tickets. Here is how it goes. Public transit networks are mainly municipal. STM handles Montreal, STL handles Laval, RTL handles Longueuil and a few other smaller ones handles smaller city agglomerations. The AMT is the umbrella organization handling infrastructure developments and operating above ground trains. All networks sell tickets or passes valid on their territory and the AMT sells tickets valid in zones.
It turns out the train tickets I previously bought were not only train tickets, but were zone 3 tickets. What prevented me from buying tickets all along was that those zone 3 tickets would have been valid tickets in the metro and most buses. The only problem is that those tickets cost about twice the price, so I’m not really interested in using them for anything else than the train.
Well, it’s a system limitation. Apologizing, the man at the Berri-UQAM station gave me a second OPUS card (from a single card promise, I now have two bulky cards and a smaller one). He also told me that to avoid the problem, I should purchase the zone tickets on solo cards instead of getting them on my regular OPUS cards if I don’t use them on a regular basis. Here is the problem with this. Solo cards are not distributed by automated machines, and the non-terminal train stations do not have human operators.
My big question is this: Is my use case so complicated that it was not even considered? How is the use case of a young man living in the city and visiting family in the suburbs around once a month a complex use case?
How did this happen? Well, it obviously is a problem of poor engineering work and misunderstanding of technology. They introduced a new technology to â€œsimplifyâ€ the interconnection between the transit networks, but did it in a way that prevents any major changes in the way they operate. Make-up on a monkey. Nothing more. They did not simplify the process. They unified it around a piece of plastic and made it harder for everyone to understand what is going on. Even thought carrying a few stacks of tickets was annoying, every time I used one, I knew what was going on. Guessing which one the machine will use for me was out of the equation.
The way this is handled in most cities is that they divide it into zones and you pay based on how many zones you cross during transit. Someone figured out that using the card going in and going out of transit like it’s done everywhere else in the world was too complicated for the citizens of Montreal, even though they had smiling people standing around the stations for months during the transition to help people. They probably would have made a better investment by actually training the permanent staff to understand how it works.
Deep down, there is a technical issue. Without input on where you are going when getting in the transit, or a way to know where you are getting out at, the system is stuck to blindly pick a ticket. The real solution is go get additional input to understand the route and bill appropriately. It did not happen, or it was completely ignored by administrators who did not understand the technical requirements implied by basic logic. To avoid this terrible randomness issue, the engineering solution was to prevent from purchasing conflicting titles on the same card altogether, and do it silently.
The AMT would have been a good solution: use the zones as a means for billing. Unify the currency, not just the payment method. Get rid of the accidental complexity imposed by the various networks without an overall vision. Especially the parts imposed by administrators in the suburbs who never actually used public transits and truly think it’s only for the poor rather than a valid mean for urban transportation.
There was never enough political pressure to give AMT the power they needed to unify the metropolitan networks. Worker unions in the various cities kept on fighting to preserve the jobs, or really just getting them re-organized. They did not bother finding a common financial model and kept on working the way they used to. All of it resulting in a system with more flaws than features. The political issue between the various networks ends up affecting us all.
Attempting to touch as little as possible of the statu-quo, all of the directors ended up agreeing on a common shopping list for the solution supplier. I will skip the open bidding process that typically comes with such projects, but here is a list of the requirements I would expect on such a list based on what I have encountered:
- Each network must be able to sell tickets and monthly passes valid on their territory.
- STM also has weekly passes, daily passes and 3-day passes.
- AMT sells zone tickets or monthly passes. Each zone may comprise one network or part of a network.
- CIT Laurentides and CIT LanaudiÃ¨re need to have tickets or monthly passes that are valid in both networks, which can be for the entire networks or only the south portions.
- Monthly passes for a single bus may be available.
At some point, someone should have realized that these are conflicting and that just asking for the card at the beginning of the trip could not possibly work. Instead of standing up and preventing a disaster, the supplier who could not afford to loose the lucrative contract simply accepted all demands. I don’t know the exact details about this contract, but I wouldn’t be surprised to hear that costs were well overrun.
There is a very simple analysis technique called the 5-Why. For any demand. Ask why to expose the underlying demand. Repeat 5 times. I’m pretty certain most of the complexity from the above list came because citizens complained that they had to carry too much change around and had to pay for every transit. If you have a unified card to pay with, change is no longer an issue.
Getting non-technical people to write down their wishlist is a terrible idea in any system design. It prevents the supplier from being creative and proposing a better, more efficient, solution. Instead, they must build poor systems to meet the illogical needs of people without the depth of understanding required to design a solution.
A bit of traveling around to see what is done in the other metropols of this world might have been useful as well. In my recent trip to London, I encountered the Oyster card. The concept is very similar to our OPUS, except that it was done a bit more efficiently.
- The card knows about pounds. Not tickets. You just load money on it and whichever transit you take will take what it needs.
- It knows about your route and charges differently accordingly. If you take a bus, then the tube, it will know it’s the same travel and not charge twice. I think it even caps to the price of a â€œdaily passâ€ at some point.
- Pricing on the tube is made based on the distance by asking the card to get in and get out. Short distances are less expensive.
- You can attach your credit card to it online and configure it to fill up when you go below a certain threshold, so you never have to worry about loading more money on the card.
Of course, this might not all have been possible in Canada. Our privacy concerns are typically higher than those of the citizens of London. London has so many cameras, they wouldn’t really need you to have a card to know your travel’s itinerary and bill accordingly.
I don’t know exactly how it used to be before, but the system seems a lot better to me. At least, they made it simple to use and abstracted away from the need of tickets.
Political issues just keep bleeding through the deployment of OPUS. Some details are not related to technical issues. They are just incomprehensible unless taken with some historical perspective. Coming back from visiting my family, I always had to pay for the metro in Laval even though I had a pass for Montreal and would spend most of the trip there. I could understand that. Even though the metro line connects, it’s actually a different network and I only paid for Montreal. They want to make sure the people living in the suburbs pay a bit more because their infrastructures end up being more expensive.
Now that I had tickets, I figured they would be good from Laval too. The tickets are the exact same price. That did not work as expected. The tickets valid on the same line are different depending on where you purchase them. I did not try, but from prior experience, I would believe I cannot purchase both at the same time. Once again, this is because the two cities could not agree on who would pay the bill for the extension of the network, which was estimated under 200M and ended up costing 750M.
Don’t these administrators think about the madness they place the citizens through with all their internal battles? Typically, when you invest a few millions to improve something, it’s a good time to clean up the multiple hacks imposed by the previously incapable system you are trying to replace. They took the hacks and turned them into specs. Now we’ll have to live with them for the 20 years to come.
Within weeks of completing the deployment of OPUS, thousands of Bixi bikes were placed around the city on hundreds of station. A complete revolution in urban transit they claimed. The creators did their homework and verified how similar projects failed in other cities and managed to avoid errors from the past. There are enough stations scattered around the city (about every two blocks) to make it useful and they attempt to balance the bikes between the stations (with moderate success). However, the company seems to be enjoying success with deployments in a few smaller cities and projects for London and Boston. There is only one issue. Those bikes can’t be used with the OPUS card. So much for standardizing.
Let’s see what the future will bring to us with the re-introduction of the tramway downtown, prolongation of the metro by 20km and the arrival of a train to the airport.