Urbanism and transportation


Moving around

People need to go to various destinations.  They commute to work or school, go shopping, run errands, visit others.  And there are many ways to fulfill this basic need: on foot, by bike, in a private car, or by taking some form of public transport.  It is important to realise that people use these modes in accordance with their utility.  Travel time (door to door), inconvenience and percieved safety are just as important as the monetary cost.  But their choice often impacts others, in one way or another. 


While ancient, walking is by no means obsolete.  Indeed, it is the "glue" that connects all other modes.  More seriously, as—unlike any other mode—walking has zero overhead, it is still competitive for short journeys, despite the low speed.  A 15-20 minute walk at 5 km/h has a range of 1-1.5 km, for reference.


Riding on two wheels used to be a very widely used way of getting from A to B within cities.  However, in most places it is very rare today.  The reason is obvious: the "official" way to use a cycle is to pretend to be a car, despite clearly not being one.  (This is called vehicular cycling.)  The result is that cyclists need to wear a lot of special gear (clothes, helmet), need to be physically fit, to significantly exert themselves, and not mind being passed very close by lorries, or being honked at by irate motorists. 

This all adds up and makes cycling have a very significant overhead—changing clothes, and getting a shower—in fact eclipsing the need to retrieve and store the bike itself.  Furthermore, having to move at speeds compatible with motor traffic is inconvenient, to say the least, as well as not feeling safe.

None of this is necessary, though.  Contemporary experience in the Netherlands shows that riding a bike can be as widespread as walking, if only adequate infrastructure is provided.  Because once a network of dedicated cycle tracks or paths is provided, people can ride at a more comfortable speed, in the same clothes as they would do anything else outdoors, and feel safe doing so.  Thus everyone (not only fit and bold i.e. young men) can cycle, with just the low overhead of storing the vehicle.

Public transport

This mode of transportation has significant overhead.  For one, the time between successive vehicles (the headway).  In the case of low-traffic lines, this can extend up to a quarter, sometimes even half an hour.  For another, the time it takes to move between to and from the station. 

An additional advantage of public transport is the variety of solutions available.  From a simple bus service to grade-separated heavy rail, public transport can cover any scale. 

It is important to mention that even between cities, many forms—such as high-speed rail—consume about an order of magnitude less energy per passenger-kilometer than cars.  Inside cities, the difference is even wider.


More formally, low occupancy motor vehicle.  This is really the odd one out, since its characteristics are so different from the above modes. 

The first major difference is the density of traffic itself.  Unlike the previous two modes, cars are a very inefficient way to move people around in cities.  The width of road required to move the same number of people as e.g. a proper BRT route is enormous.  This tremendous highway then becomes a barrier for all forms of transportation across it.  In effect, it cuts the city to isolated parts.

The second major difference is parking.  The other modes do not require a significant area of parking.  Cars, on the other hand, need a parking space of approximately 2.5m by 5-6m long.  All for moving a single person, in most cases.  This is fabulously wasteful of space, significantly decreasing the density of cities—the high density that is the very reason for cities to exist in the first place.

The third major difference is that of road behaviour.  The motorist drives a padded metal box over a ton in mass, at high speeds, and commands a hundred horsepowers or more by pressing a pedal.  Through simple inattention, they can kill vulnerable road users, at no risk to themselves.  Even when they are in collision with another motor vehicle, drivers mostly get away unscathed.  This great imbalance in risk and responsibility drives other uses off the road.

At the same time, the inevitable congestion severely hinders public transport from operating properly.  This leads to the widespread introduction of dedicated bus lanes, and even more separated rights of way, because motorists are notoriously bad at respecting the former.

To paraphrase, I could say the motor vehicle is the bully of the roads.

The fourth major difference is the effect on the environment, both local and global.  Cars are loud, and the constant roar of several lanes of traffic is not something anybody wants to be close to.  They also pollute the local air, with various byproducts of high-pressure combustion.  On a larger scale—that of a national economy—the dependence on (mostly imported) fossil fuels is detrimental, too.  On a yet larger–global–scale, road transportation is among the largest contributors of carbon dioxide emissions.

Now, some might say that the coming of the electric car will solve the problems in the last paragraph.  Indeed it will, but looking at the three paragraphs before that, I can't help but conclude that creating the electric car is solving the wrong problem.  The real problem is having so many cars, of any kind.

Perhaps the most surreal example is this Tesla ad for Hong Kong.  Just try to take in the bizarreness of it.  One of the most visionary companies in the world tries to peddle cars to a city where the people don't need cars (excellent public transport network) and don't want cars (few places to park them).  And it is just as well they don't, to keep their city dense.

The private motor car is detrimental to cities.

I have already enumerated several ways in which cars harm cities.  But there are several more worthy of mention.  One example is that cars promote sedentary lifestyles, resulting in widespread obesity and all consequent health problems.  Another is that the generally unpleasant infrastructure built for automobiles is often "separated" from the people by using so-called green space.  These mown verges of grass or unkempt collection of trees and shrubbery serve no other function than to act as a barrier—however, they consume precious land area, decreasing density.

And when you put these things together—lower densities, less direct routes, very unpleasant conditions next to multiple lanes of traffic—you realize that excessive car traffic, and urban design that caters for it, make all other modes of transport non-viable, further ingraining a dependence on cars.

Back to the top

Network or notwork?

Interactions between modes

Of course, the above modes do not exist in a void.  They can interact positively or negatively with each other.  However, the possibilities for such interactions show a very skewed distribution.

For instance, public transport can be mutually beneficial with walking and biking in many forms.  High-density walkable neighbourhoods, or—if there is a suitable network of bike paths—bike racks and/or rentals can reinforce each other perfectly.  Transit oriented development usually extends to a 800 m radius circle.  At a walking speed of 5 km/h, that's 10 minutes.  In a similar time, moving at 15 km/h (average measured in Copenhagen) a cyclist can cover 2.4 km.  If we allow for some overhead, that's still 2 km, allowing for a catchment area over six times as large.

Furthermore, public transport stations have a fairly high density, and create perfect sites for many small shops.  Indeed it is common practice to attach some shopping to stations, for mutual benefits.

On the other hand, private transport by car doesn't have any positive interactions beyond the short walk that is its overhead.  Once people get in their cars, they tend to just drive to their destination, not switch to public transport.  And the measures trying to encourage the latter—mostly free parking adjacent to stations—are exactly counterproductive.  They take the most valuable area, immediately next to the station, and convert it into a non-place.

The private car doesn't integrate with bikes, either.  For one, drivers just tend to look for a parking spot closer to their destination.  For another, once a bike ride is a necessary leg of the journey, obviously that means more than a short walk of 15-20 minutes.  At 5 km/h, that's about 1.5 km.  (If it were any shorter, we wouldn't bother and just walk.)  But when a bike ride of that length is necessary, often there are other possible endpoints within reach—often a public transit station.  Thus a two-stage commute of this kind is very rare.

As I already mentioned about cars, they also have further interactions, to the detriment of the other modes.  The wide roadways are uncomfortable to walk or bike along, are an obstacle to cross, and the inevitable congestion holds up public transport.  These negative externalities don't bother other drivers nearly so much—they just drive for a bit longer—which is much less of an inconvenience in the comfort of their own motor vehicles.

The car is a rural mode of transport.  It only has advantages over walking and biking (speed, range and comfort) when distances are significant.  It only has an advantage of low overhead compared to public transport where the latter is sparse due to low density.  And car traffic's drawbacks are much less of an issue in the countryside:

Thus the conclusion is obvious: the car does not belong in cities.

Back to the top

Between cities

This is fine for the daily commute and shopping, but what about journeys outside the urban area?  Unless someone has a stupendously long commute, these are by definition comparatively rare.  But, despite their significant differences from intra-city travel, they are still about utility: door-to-door travel time, convenience and cost.

This already means we can rule out walking and biking in the general case.  The remaining contenders are the various existing forms of public transport, cars and possible future modes.  Let's begin by looking at the easy cases.

Small towns and villages

These are the places that don't need any public transport system for the daily journeys, because everything is within reach by bike.  To recap, a 15-minute ride has a range of over 3 km, which means a town of 30 thousand can fit in this category without any difficulties.

This loose upper limit on population translates into a limit of potential travellers.  With a few exceptions (tourist destinations), the daily ridership will not justify anything beyond a coach (bus) service from one or two stations.  Indeed, this is exactly what such places already have—it is cheap, reasonably convenient, and overhead time is acceptable.

The Gray Area

This is the sticky situation, where the city is not large enough to fill a railway mainline with passengers, but a normal level of bus service doesn't cut it.  Normally, the solution is a railway branch line, or a passenger-centric variant, known in America as the interurban.

Since these lines don't need to be built for very high speeds or loads, are normally used by multiple unit trains with high tractive effort to weight, and sometimes can be single-track with some passing loops, construction costs are significantly lower than those of mainlines, and more comparable to country roads.  Still, multi-car EMU trainsets can carry many more people than buses, and sometimes also serve as a light metro line inside cities.

Between large cities

When both urban areas have more than about a million inhabitants, they deserve extensive public transportation networks.  Their large populations also mean that high numbers of passengers can be expected between them.  The question is, mostly, how far they are.


This category does not exist.  The two urban areas are actually one, just somebody identified them wrong.  Anyway, the reasonable thing to do is to connect the public transport networks seamlessly.  People will not be going very far, but connectivity is very important to keep travel times down.  Maybe throw in a few express lines.  The go-to example is the Ruhr area in Germany.

Very close

Unlike the previous example, the cities can be identified and are separate.  Still, they are close enough to allow unusually high levels of traffic: extreme commuters, and people in half-separate households going home every weekend, for instance.

The proper thing to do in this case is to connect their respective metro networks with an S-Bahn or a similar railway line.  Ideally, such lines can connect to the networks in multiple places, and even imitate metro lines in the city centres.  However, outside the centre, the main purpose of these lines is to provide a fast connection between the cities' respective networks.  A typical example would be Merseyside and Greater Manchester in England.


Up to about 200 km, normal railway services work just fine.  At a top speed of 100-150 km/h, and a fairly low number of stops, station to station the journey takes about 2 hours.  Given the amenities available, and that the door-to-door time is about an hour longer, this is perfectly reasonable.


Between about 200 and 400 km, standard rail speeds would mean the journey takes several hours.  This tends to become a nuisance, despite modern trains usually providing very comfortable amenities.  However, a full-blown high-speed train system (~300 km/h) might be overkill.  It is much easier and cheaper to build a system similar to the OBB Railjet, with a top speed a little above 200 km/h.


This is the gray area between 400 and 7-800 kilometres.  The perfect example would be the Los Angeles and the Bay Area (San Francisco), because there are no other large cities between them.

So far, I didn't mention air traffic.  The obvious reason is that it has an enormous overhead.  Even between international hubs, there cannot be multiple planes per hour, meaning long waiting times.  Checking in and out seem to take an eternity.  Airports, with their tremendous needs for space, are relegated to the outskirts of cities.

According to this site, the average gate-to-gate time between SF and LA is 84 minutes, of which 26 is spent on the ground, only 56 in the air—and as you can imagine, an even smaller fraction of that is at full speed.  Add to this the time to check in and out (I found a minimum time of 20-30 minutes for checking in).

Thus the strictly measurable overhead runs to about an hour.  The time lost in planes leaving more rarely, and to the airports being on the outskirts, can only be guessed, but we must keep them in mind, too.  Remember, people travel door to door.  But back to our average 84 minutes gate-to-gate plus check-in, totalling about two hours station to station.

The ~550 km between LA and SF would take about two hours by high-speed train, too.  However, since the train stations need much less area than an airport, they can be placed much deeper into the city, with better connections to the public transport network.  Thus while station-to-station the competition is down to the wire, what actually matters is door-to-door time, where the train wins hands down.  Furthermore, high-speed trains can run about every five minutes, while there are only 40-odd flights per day.

Very far

On continental scales, about 1000 km, air traffic starts to win out.  Even there, it is less convenient and more expensive.  It also has lower capacity: a Boeing 747-400, the most common variant, provides seating for 416 in its typical three-class configuration.  For comparison, the Eurostar has 750 seats (first and second class).

What about...?

There are also various suggestions for futuristic systems with ridiculously high speeds.  Probably the best-known today is the Hyperloop proposal (numbers on page 6).

First, capacity: its cost estimate refers to 40 capsules to go along with the LA-SF double tubes.  With a one-way trip time of 35 minutes, plus 5 for dwell time at the terminus, this gives a round-trip time of 80 minutes, thus a minimum headway of two minutes.  The same value is mentioned by the paper, although I can't grasp where the 30 second during rush hour comes from.

Anyway, a two-minute headway for vehicles of 28 people makes a meagre 840 passengers per hour per direction (pphpd), at 100% capacity.  Even with 30 seconds headway (requiring 160 capsules), the allegedly commuter-oriented system has 3360 pphpd, less than the absolute minimum even for tram (streetcar) lines (page 3).  Bus rapid transit systems have capacities of tens of thousands of pphpd, too.

Furthermore, if usage patterns really reflected those of intra-city public transport, then the morning rush hour would see, over an hour, 10-15% of the total daily use.  Thus the capacity outlined in the paper is hopelessly inadequate—by an order of magnitude or more.

Second, travel time.  Obviously, 40 minutes one way is very impressive, and there can't be anything wrong with that, right?  Sure, but this speed is completely pointless.  Since people travel door to door, we need to consider the time it takes for them to travel between the stations and the endpoints of their route.  If we allow for about half an hour in each city, which is actually a bit tight, we get a door-to-door time of 100 minutes.  This is the problem of the last mile, just on a larger scale.

So, if the travel time is dominated by the time spent inside the cities, traveling between the stations and endpoints, we might as well loosen our belt a bit.  For instance, if the station-to-station time would double, to 75 minutes, the total would be 135 minutes—or two hours and a quarter.  That sounds tolerable, especially given the distance covered.

More importantly, that allows the speed to drop from just below 1000 km/h to circa 500.  That is the design cruising speed of the Transrapid, a maglev train.  This technology has already been implemented, in the Shanghai Maglev Train.  Incidentally, this provides an illustration of the original problem: while the 30 km track from the airport is covered in 8 minutes, the city centre is a further 20 minutes subway ride away, thus ridership levels are low.

Back to the main line of thought.  The Shanghai Maglev trains provide seating for 440.  It would be trivial to make longer trains of the same kind, and to run them with a headway of e.g. 10 minutes.  This would easily allow several thousand pphpd—incidentally, still lower than traditional high-speed rail.

In conclusion, intercity traffic needs to keep in mind the conditions inside the cities.  Never forget that people travel door to door, and thus extremely high speeds are completely unnecessary below 7-800 km, above which the main contender is air traffic.  And that often the right problem to solve is how to move the people more efficiently inside the cities.


Often, when several modes of transportation are compared, rail is labeled as expensive, as opposed to roads.  A simple comparison of fares and gasoline prices seems to support this, but this is completely the wrong comparison to make.

Obviously, road traffic is heavily subsidized by various levels of government, building roads from taxes.  On the other hand, generally rail infrastructure and operations are heavily reliant on farebox income.  Let's conduct a thought experiment.

Suppose every passenger would pay the same amount as they do today, but only half of it would be the actual fare, the other half would be added to their taxes.  They would end up paying the same amount.  The extra taxes would be the income of the railway company, so they would get the same income as they do today.  However, the actual fares, paid per journey, would be halved.

Since different modes of transportation fill a utilitarian need, they are very good substitutes of each other.  Thus if rail fares decreased by half, it would be a more appealing choice, and more people would use it instead of driving.  These extra passengers would also pay the (now halved) fares.  Thus the railway's revenues would increase, while its costs—mostly unrelated to the number of passengers per train—would stay the same.

That is to say, the railway has tremendous fixed costs, but additional passengers on vehicles that run anyway (marginal costs) are negligible.  Even if the additional passenger traffic is so high that extra trains have to be put into service, the costs of the infrastructure—track, stations—and the wages of station staff are fixed costs, therefore amortising them across more passengers is beneficial.

This line of thought can be iterated until we hit either of two blocks.  The simpler is that the line reaches its maximum capacity.  In this case, the solution is to stop, and ask whether it is sensible to improve the infrastructure.

The more complicated one is when fare revenue falls below the expenses of fare collection, i.e. the wage of ticket office staff, the costs of electronic faregates, etc.  Even if the two values only approach each other, this case can apply.  Fare collection often slows the ingress and egress of passengers to/from the station.  Also, by this point the fare is a rather trivial sum, obviously.

Thus the question is whether we should bother with fare collection at all.  Probably fares are so low anyway, that making them exactly zero would not have too much of an effect on ridership levels.  Anybodywho didn't use it already probably had some reason other than cost—comfort or travel time.  Public transport being made free would not change that.  And anybody who had any inclination to use it for travel could already afford to do so.

Environmental impact

Electric trains are about 20 times more energy-efficient per passenger-km than cars.  If we also take into account that some of this energy can come from renewable sources, we get an even better picture.  And even the part which is today still generated by burning fossil fuels, is generated at plants where large and heavy equipment to improve efficiency, and to decrease emissions of e.g. sulfur dioxide, is feasible.

This contrasts sharply with the car, which has an engine much too small to use e.g. cogeneration (running a steam engine with waste heat from internal combustion engines) and simply cannot fit washers to scrub the byproducts of high-pressure combustion. 

Electric trains are also much less noisy.  The traditional source of much noise, at joints between rails, is completely eliminated by countinuously welded rail.  Due to the hardness and smoothness of the surfaces involved, steel wheels rolling on steel rails cause much less noise than rubber pneumatic tyres rolling on asphalt.  Even on curves, if the right amount of superelevation is used for the speed, there is no flange squeal.  And the main source of noise for high-speed rail, the air being forced out of the way, simply cannot be helped.  However, a single countinuous train causes much less disturbance than the same number of passengers driving cars would.

Back to the top