Suppose Money is Like Water Steven Strogatz has a intriguing column in the NY Times about a 1950s era “hydraulic computer,” which modelled the operation of a national economy using fluids flowing through a machine.  As the water circulates it fills or empties tanks, trips levers, and occasionally plots a graph of the level of a particular tank through time.  For example, when a tank called “Minimum Working Balance” fills up, it begins overflowing into a stream called “Income.”     (Click to enlarge.)  The thing has a series of input points where you can change something (modelling an external input of some kind) and see what happens as a result.
The commenters seem to focus on how charmingly obsolete the thing is, but my first reaction was:  What a great teaching tool!  Someone should create working online model of it, complete with all the rushing and gurgling sounds, that we can all play with on our laptops.
In a democracy, the greatest threat to national security is public ignorance.  The same is true of a democratically governed city.  That’s why as a transit planner, I’ve come to view explaining what I do as one of the most important parts of my job.

People learn a system better by playing with it than by reading about it or listening to lectures.  That’s why a science education involves so much time in the lab.  Playing with a system gives you a “feel” for a topic that’s seeps into your subcionscious thought processes in a way that book-learning rarely seems to do.
I’m sensitive to this because although I got a great graduate education in another field, I learned transit planning entirely by doing it, and it took me years to figure out how to explain what I was doing to an average citizen.  But such explanations are crucial, because in the absence of understanding a citizen will fall back on myths.  These myths are almost always false analogies to other things that a citizen does have a feel for, such as cars and traffic.  When I’m talking about basic transit concepts like speed and frequency, for example, I find that everyone understands speed, but it’s much harder to get them to understand frequency.  They can understand waiting time — they’ve waited for elevators even if they’ve never waited for a bus — but it’s often remarkably hard to get them to see how frequency percolates through a system, affecting so many other aspects of how (and whether) a transit system works.
Picture1The tape-and-mylar planning games that I use (described in this post, next to the photos) are a crude example of a tool for giving people the necessary “feel” for transit, but we need more subtle ones that cover other aspects of the trade.
Like many urbanists, I played for a while with Sim City when it came out in the 1990s.  Sim City’s model of public transit is worse than ignorant.  All you need to do is build a rail line and presto, trains start running!  In the real world of transit, operating cost swamps capital cost,* so if you don’t have a funding source to operate a service year after year, you don’t have a service.  (Again, this is a false analogy with cars and roads, where the capital cost, the contruction of the road, really is the largest expense, because the traveller pays to run the car.)  I suspect Sim City did subconscious harm in training a generation of inquisitive young people that transit is just about building things.  Some of those people are influential architects today.
The success of Sim City as a game shows how desperate people are for good models of how cities work, and if we don’t give them good ones they’ll use bad ones.  Who’s out there working on the transit dimensions of this problem?
Diagram of Phillips machine from Barr, N. (2000) “The history of the Phillips machine,” via New York Times.
* Generally, the cost of running transit is dominated by lhe cost of labor.  The big exceptions are (a) driverless metros such as Vancouver’s SkyTrain and (b) developing-world countries were labor is very, very cheap.

15 Responses to Suppose Money is Like Water

  1. Jason June 3, 2009 at 6:54 pm #

    There’s 12 copies of this machine in the world
    If anyone is keen, one is on public display in the economics building at Melbourne Uni. But i warn that it is not that interesting to see…
    However, I think the idea of building a ‘computer game’ that gave us a barrel of money and let us choose transport solutions would be a great tool. (not to mention a good argument against the caufield-footscray tunnel!). In a similar vein, I saw a very simple game that let users determine federal budget priorites on a couple of years ago.
    It’d be great if it was easy to use (right hand tool-bar, drag and drop tunnels, bike paths, pedestrian malls, elevated trains, freeways, BRT, light rail and new tramlines). It’d be cool if it ran in the browser, and users could save their designs and then browse, vote and comment on other peoples.
    The difficult part would be running simulated traffic through the designs, and measuring winners and losers, but that’s probably not necessary to start a debate. If it worked well i’m sure would be eager to host it, and it would attract a lot of attention!

  2. dale June 4, 2009 at 12:44 pm #

    It’s kind of surprising how little has emerged in the way of simulation models that everyone can monkey with (at least to my knowledge.) I used to play historical simulation games (known generally as “war games.”) The best online version, nowadays?, which simply made a digital version of the simulations done in the ’70’s. So state of the art, for simulations that have any value (as opposed to those silly shooter games), is still at the board, card, and little-cardboard-chit level. So much could be done!

  3. Ed June 4, 2009 at 5:08 pm #

    Anyone interested in transit simulation games might like to check out “Traffic Giant” by JoWood. Released in 2000, Traffic Giant is pretty old now but still available new or secondhand via the web – I got my copy some time ago on e-bay. A description and review of the game is at
    In the game, you create transit networks on preloaded city maps/scenarios with different populations, landuse layouts and densities. With a modest budget, you start out with your first bus line (locating the route and each stop on a city map) linking residential areas with other landuses. As patronage and revenue build, you can afford to add more buses to your line, establish a network of lines (it recognises passenger transfers), and on lines where buses can no longer cope with demand, you can upgrade to light rail or off-street suburban rail. You win when you meet mode share and other targets as a sole operator or in competition with another operator.
    Naturally the game greatly simplifies the real world and there are a lot of assumptions built in about how people move around and what makes a good transit system. What’s interesting is generating and watching the continual flow of passengers on and off the buses and through the system, and trying to get a regular rhythm going with the frequencies. It’s very satisfying seeing your buses and trams continually turning over passengers and generating revenue – but also frustrating to see them stuck in traffic. Some lines never seem to take off, and you adjust the route and stop locations to see if you can get the boardings up. I also like to see if I can serve the city with the minimum number of lines and with the minimum number of vehicles.
    For those looking to have some fun in playing around with and getting a feel for transit networks, Traffic Giant is one game worth having a look at.

  4. Multimodal Man June 4, 2009 at 9:34 pm #

    Great post. I’ve thought that a transit planning blog from an agency’s planning department would be very useful. I think the more planners can re-educate the better. Sim City has done a dis-service, never mind that most engineers think the same way. They don’t see alignment selection and service level provided as being very important compared to project delivered (beautiful stations, cost efficiencies in ROW acquisition, etc).
    One useful way a planner friend of mine has used to explain what frequency means is asking people to imagine they could only turn on the ignition in their car every 30 minutes.

  5. David Levinson June 5, 2009 at 6:13 am #

    The STREET project at the University of Minnesota has a number of free, open source transportation simulations (travel demand, traffic, signals) to be used in the classroom. Experiments in transportation are generally costly, so simulations are excellent teaching tools.

  6. Bob Davis June 5, 2009 at 11:29 pm #

    “The cost of transit [except for driverless metros] is overwhelmingly the cost of labor”. I rember reading one author who said that nearly every driver of a personal automobile is an “unpaid chauffeur”. One of the problems in “selling” mass transit is to get the message across that was used on (as I recall) Los Angeles streetcars “It’s work to drive, why drive to work?” Unfortunately for transit promoters, in the years since that slogan was created, automobiles have improved with automatic transmissions, air conditioning, power options, and stereo systems, while all the bus or train has is a seat that’s probably less comfortable than a 1937 PCC car, and this is assuming that it’s not so crowded that one has to stand.

  7. Jarrett at June 6, 2009 at 12:14 am #

    My impression from the historic reconstructions I’ve ridden is that early 20c streetcars were not especially comfortable compared to the higher end of low-floor buses and railcars today. But then, they had no competition.

  8. Jarrett at June 6, 2009 at 12:17 am #

    Thanks for all the great comments. I will follow up on the U of Minnesota lead and a box of Traffic Giant is now en route to me from the UK. Like Dale, I’m still struck by how underdeveloped this field is. Please keep sending any information you have on useful educational simulations of how public transit works.

  9. Nathanael June 6, 2009 at 12:09 pm #

    So, the usual argument against driverless metros is that they can only be done with fully grade-separated systems. (Of course, if you’re doing the grade separations *already*, you might as well make it driverless, right? But of course the main argument against it in *existing* cities with full grade separation — NYC Subway should arguably be driverless — is that you can’t break the union.)
    Now, given that tunnelling and bridging is really very expensive, does operating cost really swamp capital cost with a rail system with full grade separation in areas where rail doesn’t “own the grade” at ground level? And, more critically, given that capital expenditures only happen rarely and operating expenditures are recurring, *over how many years*? In other words, is a driverless subway really cheaper than a driver-run line with grade crossings, and if so how many years does it take to become cheaper? I’m sure it’s cheaper over a 50-year timeframe, but maybe not over the 4-year timeframe of the typical politician.
    Now, in Docklands Light Railway (London), they’re driverless, but they had to hire guards to make people feel safe. How does *THAT* cost analysis work out? I assume they get paid less than drivers — is labor still the largest system cost?
    One of the political problems is that there’s usually massive sticker shock at the capital costs. I’m very suspicious of the claim that operating costs dwarf capital costs in something like the NYC Subway or the Northeast Corridor (obviously they do in bus systems and other inefficient many-driver schemes), but unfortunately there is a severe reluctance to pay one particularly crucial type of capital cost: maintenance, renewal, and upgrading to modern standards.
    If full state-of-good-repair, renewal and update-to-best-practices (signal upgrades, ADA, etc.) funding was being paid every year by every mass transit system, would the operating costs still dwarf the capital costs? I don’t know, because it has never been done in the US. 🙁
    Personally, the games I play are mostly clones or knockoffs of Transport Tycoon or Railroad Tycoon: Simutrans right now. Even though they’re intended to model long-distance transportation…. they really don’t, but they provide a half-plausible model of short-distance transportation. The transport model is pretty unrealistic but it’s a lot better than SimCity; you do pay to run the trains, you can run at a loss, and people have places they prefer to go so you can get stuck with half-empty trains, and they don’t like waiting for trains. (It’s the population and industry growth model which is particuarly out of whack, and most of the games are tuned to a ‘it is possible to make money transporting people’ situation, rather than a more realistic ‘highway subsidy’ situation.)
    Traffic Giant is probably the best single public transit sim game so far, because it’s the only one where the goal is to serve lots of people rather than to make $$$, but unfortunately I can’t get it to run on my machine any more. It assumes a ‘steady state’ industrial and population model, which simplifies the design problems a lot, but within that limit it does quite well; it even includes the way it’s hard to rebuild ridership when you move bus stops or rearrange their lines, though I think the passengers are too forgiving. The ‘best strategy’ I’ve found for it is to experiment with bus lines, relocating them and rearranging them until you have a fairly efficient route moving lots of people, and then if it’s busy enough upgrade it to trams.
    In Traffic Giant, one oddity: although trains are in the game, it’s usually quite impossible to build decent train lines due to lack of right-of-way on the prefab maps (the game does not let you build street trackage or grade-separate or use eminent domain to knock down buildings, and the maps start dense).

  10. Jarrett at June 6, 2009 at 4:25 pm #

    Excellent comment, Nathaniel! My note about operations swamping capital is mostly based on the experience of bus, streetcar, and light rail operations. As you move into newly-built heavier urban rail, two things are different (1) capital is much higher, (2) ridership and fare revenue per driver are much higher, which offsets some operating cost. So I suspect the balance shifts but not to the point that operations aren’t the larger expense. (Of course there are all kinds of methodological questions you can ask about how we compare those two, since it depends on an assumption about life cycle of capital.)
    I don’t believe it’s practical to convert existing lines to driverless, but you’re right that even if it were technically possible the political obstacle would be spectacular. It’s also impossible to talk about how the NYC subway would stand relative to its capital cost; are you talking about the replacement cost to build century-old lines today? That would be astronomical. The builders of those lines didn’t have to worry about environmental impacts or the conditions of labor, so much was done far more cheaply than could be done today. That’s part of why I judge late 20c systems only against each other and never against older ones.
    Security is a big issue on driverless metros, but frankly I think it’s overblown. There’s very little a driver can do to prevent crime on his train. All a rider can do in event of crime is push the little button and tell the driver about it, and all the driver can do is call the police and hold the train at a safe location. On a driverless metro it works exactly the same way; you’re just talking to someone at the operations center instead of on your train. Vancouver’s SkyTrain has roving employees on the system who deal with issues and also check fares, but they’re fewer than one per train (and not directly related to the number of trains running) so the differential cost of running lots of off-peak frequency is low — which to me is the great payoff of driverlessness.
    Thanks for the referrals to simulation games. It sounds like an underdeveloped area to me, and I’d still like to hear more.
    Cheers, Jarrett

  11. Peter June 7, 2009 at 1:44 pm #

    Ah, SimCity. I was just complaining about this the other day with some of my planner friends. As a youngster I spent many hours building fields of residential tract housing, industrial parks, huge blighted and substantially vacant commercial districts, mega-highways connecting them all, and Godzilla.
    When I recently discovered that the original SimCity was released as open source, I had to download it and try it out. I knew that it was inaccurate, but it was nostalgia. Then I discovered exactly how inaccurate it was. “No mixed residential and commercial areas?!? WTF!” I did play it long enough to also notice that transportation was pretty much a capital expenditure with no operating costs. Sigh.
    Maybe your new US DOT readership could shake loose some funding for a project similar in scope to the DoD’s “America’s Army” video game? I can see it now: “Engage residents in realistic dialogue about the merits of density, reduced roadways expenditures, and increased public transit expenditures. ‘American Planner’ allows for sophisticated team play and role specialization. Play as a time-stressed project manager, travel demand modeler, or even as a community cycling and pedestrian coordinator. It’s your choice!”
    What do you say, Mr. Secretary? Future planners of America are awaiting your bold initiative.

  12. Peter June 7, 2009 at 2:05 pm #

    On a more serious note, I really like your mylar transit route exercise. I wonder if this is the sort of thing that someone could create a Google Maps “Mashup” utility for. This is a pretty simple model in that it’s coupling cost to a fixed length of a particular type of transit line, but it is, nevertheless, a very useful one. The need to physically place limited strips of mylar would be a powerful visualization for citizens who would typically have no idea what the relative costs are.
    Do you have other such exercises that you’ve done with people? Maybe I could figure out some mashups for a handful as my pre-gradschool summer activity…

  13. Alon Levy June 9, 2009 at 5:42 pm #

    SimCity does too have operating costs. I never played the original game, but in SimCity 2000, every year there’s a budget, which includes items such as schools, police, road and rail maintenance, operating costs for transit, etc.

  14. Zmapper June 9, 2010 at 9:28 pm #

    Ok, this post involves simulation games, so I’ll bite. I currently use Traffic Giant and SimCity 4. I also have a game called Mobility and SimCity 3000.
    As far as Traffic Giant goes, it does do a great job at simulating the operations of a transit system in a small town. You have no NIMBYS to collaborate with so you can do whatever you want.
    The game operates on a for profit model instead of a subsidy-based model so you are more inclined to serve that housing block then the sprawling subdivision. Even though I have never found a route that doesn’t make even a bare minimum of profit, density invites new challenges with the reward of more passengers.
    Given the choice between the 40 passenger, high floor bus found in the beginning and the first streetcar model, they will chose the streetcar. If the first streetcar model and the first streetcar model and the hydrogen bus are pitted against each other, the hydrogen bus will get the passengers.
    New rail tracks and rolling stock WILL cost some money compared to buying a bunch of buses but each streetcar can carry 150-200 people so it is worth it. The driver payroll is changeable but it will affect passenger comfort.
    As far as the problems with the game, I’ll break them down into 2 categories. Graphics and Interface problems.
    Graphical Problems:
    – Streetcars making funny U-turns
    – The lack of depots
    – The simplification of gender roles
    – Few Pedestrians
    Interface Problems:
    – Only house to job, school, shopping, and recreational locations. (No house to house trips)
    – Poor coordination between routes
    – Hard to make transit malls
    – Streetcars can’t run on private ROW
    – No city expansion
    If you want to demo it, you can download it here:
    Now contrary to what Nathanael said at the bottom of his post, you can demolish buildings. I haven’t used that feature much but it doesn’t look the best in my opinion.
    Seeing how long this post is, I’ll post my review of SimCity 4 later.

  15. Zmapper June 9, 2010 at 9:38 pm #

    Great, thought this was posted 6/4/2010 not 09! I had no intentions of opening up an older thread.