New York’s MTA Genius Challenge
I don’t like the word “genius.” When people use it unironically, what I hear is “we haven’t met many smart people, so the first one we meet looks like a genius to us.” Math academia is very good about excising the word from anyone’s vocabulary. It drills you on the idea that you’re not Manjul Bhargava or anyone of that caliber, and if you are, you’re judged by what you’ve proved, not how theoretically smart you are. The tech industry uses the term more often, alongside related terms: rock star, 10x engineer, ninja. Most of it serves to convince coders that they’re masters of the universe, that all of them are above average and half of them are in the top 10% of coders.
New York State just issued a call for proposals for a $1 million grant, dubbed the MTA Genius Transit Challenge. I sent in a request for more information, and haven’t gotten a response yet; when I do, I will probably apply, if the specs and timeframe are within what I can give, but I doubt I will get it. My suspicion is that the state is looking for a tech company to privatize something to. Governor Andrew Cuomo wants someone to tackle one of the following three problems:
- Rail signaling, in context of how to maximize the subway’s capacity in trains per hour.
- Rolling stock maintenance schedules: the state isn’t saying what the ultimate issue is, but presumably it is reliability.
- Cell service and wi-fi underground.
I doubt that the tech industry is capable of doing much on the first two issues, while the third one is a solved problem (as in cities like Singapore and Boston) that just requires installing wires. The first two issues have a lot of potential improvements, but they come from the transportation field, including service planning.
Unfortunately, the panel judging the grant is tilted toward people in the tech industry. Only one has background in rail transportation: Sarah Feinberg, former administrator of the FRA, whose background prior to working at the US Department of Transportation is in politics and tech. Two more are academic administrators, neither with background in transportation: SUNY Chancellor-elect Kristina Johnson, an engineer with background in energy and 3D graphics, and Daniel Huttenlocher, dean and vice provost of Cornell Tech, whose background is in IT. The other five are in the tech industry; one is a professor who studies networks, with some applications to car transportation (congestion pricing) but not to rail. Missing from the panel are people who worked on ETCS, people who have developed driverless train technology, and professionals within the major rolling stock vendors.
The biggest tech fixes in New York area outside the three areas identified by Cuomo. One, train arrival boards, is already in development, with planned opening next year.
But an even bigger fix is speed: the subways in New York have permanent slow orders at some places, not because of deferred maintenance but because of past accidents. There is a railroading tradition, in the US but sometimes also elsewhere, of using slow orders to mask underlying safety issues, even when the accident in question had very little to do with speed. The subways in New York today are getting even slower, for a combination of legitimate reasons (temporary signal upgrades) and illegitimate ones (inexperienced crews assigned at the busiest times).
However, the solutions to these problems often combine many different viewpoints. Speeding up the subway involves ending the slow orders (which involves signaling, but isn’t exactly tech), improving scheduling to reduce delays at merges (which involves service planning), reallocating crews (which involves labor relations), and coming up with ways to reinstall signals with less impact to operations (which is itself a combination of signaling tech and service planning).
American tech industry titans like to think of themselves as omnicompetent; Elon Musk’s bad ideas about transportation, from Hyperloop to elevator-accessed tunnels for cars, stem from his apparent belief that he can understand everything better than anyone else. This is not how good interdisciplinary work happens; the best examples in science involve people who are specialized to the two fields they’re combining, or people in one field collaborating with people in another field. A governor that understood this would empanel people with a wider variety of fields of expertise within the transportation industry: service planning, civil engineering, signal engineering, local labor relations and regulations, rolling stock maintenance. There would be one tech person on the panel (among the existing panelists, the professor studying networks, Balaji Prabhakhar, seems the most relevant in background), rather than one non-tech person.
This sort of self-importance especially appeals to Cuomo. Cuomo is not managing the state of New York; he is running for president of the United States, which requires him to be able to say “I did that” about something. Solving big problems requires big money; reducing costs requires local tradeoffs, such as reducing construction costs by using more disruptive cut-and-cover techniques. That’s how you run a good government, but that’s not how you run a cautious political campaign for higher office, in which the other side will pounce on every negative consequence. As a result, Cuomo is hoping to solve problems using tech innovation without spending much money; but the parameters of his plan seem to guarantee that the panel can only solve small problems, without touching on the most fundamental concerns for people riding the subway.
Elon Musk’s idea of elevator-accessed tunnels and car sleds is indeed insane. One of the first destinations he mentions is LAX airport. But why would you want to bring your car with you to LAX just for the privilege of paying $15 a day to leave it parked in the garage?
There is, however, one insight in the proposal that is easy to miss but is true. That is: Underground finished spaces are extremely expensive to build. Mezzanines, platforms, escalators, etc, and all the systems to support them, are often the most expensive parts of any subway to construct.
If you can minimize these spaces, or even avoid building them entirely, then might not subways be a lot cheaper to build? Imagine a subway system where the elevators themselves ARE the platforms. Elevators depart the surface level timed to arrive at the subway level at the same time as a train. Doors open. Passengers move directly between the elevator and the train. When passenger loading and unloading is done, the train departs and the elevator returns to the surface.
So you’d need less (no?) finished underground space, which saves money – but you would need more elevators – so would this be a win overall? Not really sure. And if there were too many people wanting to get off at a given station, the train would have to sit there and wait while the elevators make a second trip. Alternatively you could leave some elevator queuing space, but then that basically amounts to having a platform, and then you’re going outside of the footprint of the bored tunnel and building finished underground space.
There are probably other reasons this wouldn’t work, but who cares! It’s fun to daydream.
You’re sort of reinventing the large-diameter TBM here, in which the platform is within the bore. Musk’s plan involves narrow diameters because he genuinely thinks cost is proportional to cross-sectional area, but to get any real capacity out of cars, you need a larger bore.
Elevators and subways mix really poorly. Subway capacity, even at an individual outlying station, is at least a full order of magnitude higher than elevator capacity, probably more than one order of magnitude. At Nation, there’s an elevator, which I use a lot since it’s the best access point for me. But it has queues, especially when I’ve just gotten off a train (since many people get off the train at once, whereas people who are getting on a train enter the station at a uniform distribution), and it can easily delay me 10 minutes if there are a lot of people at rush hour.
More elevators then? One elevator per train door. Most transit elevators I have ever used are also quite a lot slower than they need to be for factors like passenger comfort. Something like the high speed elevators used to access sky lobbies in skyscrapers could possibly cycle more than once in a reasonable time at a busy central station. And of course there is also the option of just making the elevator bigger. Then again at some point the costs of more, bigger, and faster elevators approaches or exceeds the cost of just building a conventional station.
I’ not sure that’s enough at the busier stations, either. A normal-size elevator holds maybe 10 people, and at the busiest stations the on/off volumes are higher. At Gare du Nord, I counted about 16 on and 16 off per door at rush hour; the RER B runs trains with 4 door pairs per car. Not to mention, once you’re drilling 32 elevators per platform you might as well just blast a cavern, or (better) cut and cover a station.
The only subway I ever used regularly was in the Metro Rail in Buffalo, and few used the elevators because (1) they are very slow and (2) they smelled heavily of urine or (3) occasionally were “out-of-order”. As a layman I would think escalators and yes, stairs are much better. I usually when for the stairs. It was good exercise, and even fun because with a good pace you be the last person off the train but the first to reach the top, at street level. What about ramps, like at GCT? Or even spiral ramps. Obliviously space is a big concern, but I could see how ramps in some situations might be workable.
As for Gov. Coumo, he is making a fool of himself concerning rail transport, including his big call to “privatize” Penn Station or have the MTA or Port Authority take over… when those organizations have their own big issues. He proposed a pretty pragmatic plan to improve Penn Station’s concourses last year which could dovetail with Amtrak’s announced intention to turn manage of the above track level infrastructure to a private company like at airports… but then goes off the rails with clearly poorly thought out and uniformed pronouncements. In all fairness a lot of rail improvements have or are occurring during his governorship including important track, signaling, and station work Upstate on the state supported Empire Corridor, but he doesn’t either fully acknowledge or want too face the far greater work and money needed to repair, modernize, and expand passenger rail transport in NYS.
Perhaps some of the issues with the governor could have been alleviated if NYSDOT had a much larger rail division with a higher profile that could advise the governor, make long-term plans, and advocate for clearly necessary big ticket items like replacing Amtrak’s dual-modes. Decades ago NYSDOT had a very large and experience rail division, big projects were completed under its watch, but then it was downsized under the current governor’s father, Mario Coumo. One result was the failed HSR Program of Gov. George Pataki involving the Turboliners. Poor management resulted in project failure and a waste of public money. At least Coumo has avoided a failure that big so far, in fact it was his administration that finally scrapped the rusting Turboliners. It should serve as a lesson.
So maybe this isn’t really an option for locations with very high demand like Manhattan or central Paris. Those places clearly have enough demand to the point where the expense of building stations, platforms, mezzanines, escalators, etc is warranted.
But if the costs of tunneling really can be brought down as much as Musk suggests, this could be an alternative to make the costs pencil out to build a full subway in medium-density areas where currently surface tram or busway are the only options right now. (Like the Tvärbanan in Stockholm, or Before you say “Forget it, just take space from cars and build a tram/busway.” I should first remind you that even in places where this is politically possible, the process of rearranging and rebuilding streets, and building grade separations where necessary, is actually very disruptive and quite expensive – and even then you have to stick to existing wide arterials, which (as you have noted in the past), often miss the center of an activity node, and sometimes lines will have to turn or jog from one street to another which slows things down.
Since this would be for medium-density areas, the infrastructure would be designed for medium ridership. Think: two (large-ish) doors per (short-ish) car, with three- or four-car trains. Trains could be something along the lines of the Bombardier Innovia. Assuming an elevator takes 30 seconds to travel up or down, and you have 30 seconds for loading and unloading at the top and the bottom, that allows for headways down to 2 minutes.
Also since the only surface disruption would be the shafts bored for elevators, it wouldn’t cost much more to put the tunnels deeper. And if you go deep enough, you don’t have to think as hard about the things that usually make tunneling unpredictable and expensive like subsidence. The only problem is, at greater depth, the elevators take longer to travel, so minimum theoretical headways are reduced.
There may well be no situation that this could work. But hey, I enjoy a fun mental exercise. And, as much as we may despise the notion of a ‘genius’, the notion of disruption is real, and it could potentially apply to transit and transportation. Most often, it comes from mundane existing technologies being assembled in a new way (see: iPhone. Touch screen PDAs with cell service had existed for years prior to its introduction, but Apple was the first one to make it work.) TBMs, automated metros, and elevators are similarly very much existing technology.
I am not sure if Musk intentionally styles himself as a ‘genius’ or the Messiah or whatever, or if it is just the press and his fan base. He does seem to enjoy being compared to Tony Stark. At any rate, most of what he actually does, including these efforts to make tunneling cheaper, are really fairly mundane matters of industrial engineering, done at scale. (Steve Jobs, with whom he is also often compared, had a similar forté of user experience and industrial design.) For tunneling, the main approaches that I have heard are: 1. More TBM automation, 2. Ways to remove the spoil more quickly, 3. Modifications to the machinery so that it can continue to bore while installing the liner panels. Number 3 is a bit more of a stretch than the other two but even that doesn’t sound too far-fetched to me.
Anyway I am really rooting for them to succeed at improving TBMs. Not because I like the cockamamie scheme for 125mph Car Sleds, but rather because of the other benefits that cheaper tunneling could bring. Rather than just dismissing The Boring Company because of Car Sleds and urban geometry, transit advocates should instead engage in a mental exercise to start thinking about what could be accomplished if tunneling (particularly, deep tunneling) were much cheaper than it is. Perhaps applying the cheaper TBM technology to the existing concept of Barcelona-style large-diameter tunnels is the answer, or perhaps there is some other way of combining existing technology in a new way (like elevator-platforms) to make it work.
I dismiss the Boring Company because Musk doesn’t seem to understand modern TBM technology, and is flagrantly uninterested in where advanced could really be deployed.
In lower-capacity environments, like Vancouver, the cheap solution is to build shorter platforms. The Canada Line platforms are 40-50 meters. Most of the Canada Line was built cut-and-cover (and Vancouver won’t do it again on a major street – the plans for Broadway are to go under 10th Avenue, one block away, and just build entrances on Broadway). But the downtown parts were bored, and shorter platforms really did help reduce cavern costs.
I believe that anyone really thinking about it will decide that a reasonably large volume of space is needed in an underground station, regardless of what we might wish to be true. The absolute minimum space, for a station which is completely polarized, load-wise (i.e. vast majority of traffic arrives in AM rush from one direction and in the PM leaves in one direction – other traffic minimal) – such a station could get away with 1 center platform big enough to handle (un)loading 1 trainload of passengers; this is pretty much the minimum, and you’d need to have some mechanism to prevent additional people from overcrowding the platform in the PM waiting for the train home. On the flip side, a central station experiencing high traffic in each direction could require upwards of two full side platforms with room for more than 1 trainload of passengers each, plus a mezzanine level capable of holding at least another 2 trains worth to buffer the throughput.
The idea of having whole platforms made up of “just in time” elevators, or whole platforms which are themselves “just in time” elevators is pretty ludicrous, First of all, it would be fantasyland to imagine that these things would remain in working order to any dependable degree, let alone be in fact “on time”, the failure of which would play havoc with schedules and congestion (like with full elevators/platforms of people going up and down hoping that they can get off onto a train instead of having to go back up again or vice versa in various ways). Even aside from that, the volume of excavation necessary for all those elevators, or especially for the whole-platform elevator shafts would be equal to or greater than a large cavern and could be structurally challenging. If one is going pie in the sky, might as well go for some sort of cross between conveyor belts and little mobile lounges that pick you up on the platform , then roll over to the opening for the conveyor belt and hop on (literally so that the belt doesn’t have to stop) and continue the loop on top to unload and take people back down, etc.
Or, you bite the bullet and have big enough platforms/mezzanines underground with some combo of escalators, elevators and stairs (preferable all three) to go up and down.
As orulz (and many others, I’m sure) said, we could use an improved method of doing the excavation and tunneling, so that these underground caverns can be affordably constructed.
Perhaps Governor Cuomo’s prize should be for that – an equivalent to a squad of robotic sand hogs.
Some kind of “portable” TBMs that could travel between station caverns in the main subway tunnel and could work at enlarging the tunnels at the station sites to turn them into caverns big enough for platforms and mezzanines. Obviously not as simple as that, but maybe the $1 million winner has the proper actual approach and the ways to handle the underground faults and water incursions and alien artifacts that might come up too.
The vertical part of the trip could use some improvement too, but I think that is manageable with current tech and maybe even reasonably attainable especially with a bit of effort in project management and quality control.
600 foot long things, in Manhattan, are not cheap to build.
650 feet tall.
Almost a billion dollars. It’s not all marble tile and bronze lobby art.
Did you know that the 10x thing apparently came from a study where they tested people for all of 30 minutes?
I’ve grown to grudgingly enjoy programming as a career but dear god the tech industry needs a good shock to its system and a fair dose of humility.
For improved maintenance (problem number 2), they might be looking for a remote diagnostics system like the one that GE Transportation are supplying to DB. Does mentioning that make me a Genius?
The part about train refurbishment is odd. Working out whether it would be cost effective would need inside knowledge about individual train types (physical condition, maintenance costs, spare part availability problems, etc. all ,leading to a cost-benefit assessment for refurbishment versus new trains). It’s hard to see how outsiders can do more than raise the possibility of refurbishment, which the Governor has already done in the challenge.
At TransportationCamp New York this year, one of the discussions in the last block concerned the coalition-building effort that was required to pass Measure M. To be clear, I am sure that there is much to the final package to be picked apart, and I am certainly not qualified to comment on that having never ventured to SoCal, but I definitely got the sense that the key to its passage was a bottom-up, granular approach. As an antidote to the well-known, failed top-down approach taken by Cuomo, the tech industry, etc. to the political transportation challenges of Greater New York, is there anyone working bottom-up with the mass transit operations workforce? At the happy hour, another attendee remarked to me that a key reason for the high unit capital costs in Greater New York is the alliance between organized blue-collar labor and white-collar engineering associations.
Going forward, I think that the unions can either be reformers’ friends or enemies. With every day that passes, I am more and more firmly of the belief that the central challenge of New York transit advocates is to get labor back on the side of the public interest to combat the consultants’ need to justify their own existences. Anecdotally, of the many family friends I have talked to over the years who are past and present MTA track/signal forces, most get the current system is broken. It’s about making clear how organized labor rank and file benefits from reform: bringing construction costs in line with peer cities will get more stuff actually built. I think there is a certain distrust between some transit advocates and transit labor unions. The existence of this disconnect is counterproductive and unnecessary; as has been demonstrated repeatedly by Mr. Levy’s cited data, other first-world cities with strong labor cultures build a unit of infrastructure far cheaper than does New York, which strongly suggests that blue-collar compensation does not suffice to explain our cost problem anyway.