Sidewalks and driveways: is it time to throw the curb to the curb?

So a few months back, I saw a Tweet regarding the issue of driveways that made me think, and I put it aside in order to write on the subject later on. The time I think has come, so here it goes.

First, here is the Tweet:

Link

Later on in the discussion, this image was brought up as an example of the horror of driveways:

Proposed townhouses with massive driveways

That's a pretty strong statement. Driveways are the most anti-urban thing, really?

Reasoning behind the claim

The claim is essentially based on a North American conception of the street and integral mode separation, wherein motorized vehicles and pedestrians need to be strictly separated (and sometimes cyclists are added in there too). In such a perspective, an area without sidewalk is necessarily pedestrian-hostile and thus anti-urban because it forces pedestrians in an area made for cars: the street. I've already spoken about this conception in an earlier post.

What is the link between that conception and driveways? The problem here is that driveways represent an intrusion of cars into the pedestrian corridor and breaks it off in disconnected pieces. On the other hand, if cars are parked on the street instead, then the pedestrian corridor is safe from intrusions by cars.

Schematic representation of the scenario with on-street parking

Schematic representation of the scenario with driveways

Of course, cars intrude on these spaces only rarely, it isn't a commercial parking lot where cars stay only 30 minutes or so, and they do so only at very low speeds, so how bad is it really? There is also the issue that pedestrians are then asked to walk between parked vehicles and moving vehicles, though, to be fair, vehicles parked on a driveway tend to be some distance away from the sidewalk.

I think what mainly compounds the problem is the curb cut. Sidewalks are by tradition built slightly elevated from the street (the curb), as a result, the sidewalk is lowered at least partly to the level of the street to allow cars an easier access to the driveway. Since North American sidewalks are so narrow, it often means that the pedestrian corridor will not be level, but will have many sections where it is lower and in a slope, making it much less comfortable for pedestrians.

Example of curb cut

Why have a curb anyway?

If the curb cut is so annoying for pedestrians, one might ask why we have a curb anyway. It's not like this tiny elevation is going to do anything to stop cars with 150 to 300 hp from going on the sidewalk. Not one month goes past without some car somewhere "jumping the curb" and hitting pedestrians or, more rarely, houses. If the curb is meant to protect pedestrians from cars, it is failing badly at it. At best, it keeps cars parked on the street from parking on the sidewalk... sometimes. Way back, when the vehicles in the street were horse-drawn, it may have been a significant obstacle (and indeed, major Roman streets did have sidewalks), but horse-drawn vehicles weren't that much of a threat to pedestrians.

In the end, perhaps the prime reason that sidewalks were invented wasn't about safety, but about cleanliness. When roads were unpaved and horses and beasts of burden were in common use, the road itself could be extremely dirty. Muddy when it rained, with pieces of feces from animals, not the best area to walk in. Sidewalks were often built first, often in wooden planks (in which case they are often called boardwalks) in order to provide a clean way of walking around. To avoid accumulation of water in the absence of sewer systems and to avoid dirt from the street from being blown over, they were elevated. The road was still for pedestrians, but those who wanted to get around without fear of being dirty on arrival could use sidewalks

So with universal street paving, the replacement of beasts of burden by motor vehicles and rainwater collection systems, the reasons for the curb's existence don't seem all that significant nowadays. If we get rid of the curb, then we don't have curb cuts and driveways can be way more tolerable for pedestrians. And indeed, I will point out again Japanese streets that are built without a curb. The Japanese also hate on-street parking and make massive use of off-street garages, parking lots and driveways:

Commercial street in Takayama, note the small water channel to drain the street, a traditional method that works well

Recent developments in Yokohama, with no sidewalk and pedestrians owning the street, driveways are massively used and often decorated, not disturbing pedestrians in the least

Another residential street, note again the parking in the form of driveways to the left

One last example from Tokyo, note the garage doors to the left, again, they do not make it less comfortable to walk thanks to the absence of a curb

But for residential streets to work in that way, it is necessary that they be built primarily for pedestrians, not cars, and car drivers need to feel that they do not belong there so that they drive slowly and carefully, like in a parking lot. So when pedestrians and cars have to share the same space, that space must be NARROW with fixed objects close to the edges (trees, walls, fences, etc...). Too often, I've seen residential streets in North America being built wide without sidewalk, probably with the thinking that if cars and pedestrians have to share the pavement, then it needs to be wider to provide greater separation, which is exactly the opposite of what we should do.

On arterials where we want to channel motorized traffic, use separation is still required. If you want cars to go fast (and I define fast as anything above 30 km/h, about 20 mph, and I'm probably overestimating it), then you need to clearly separate vehicles and pedestrians. But again, the curb is a terrible way to do so, it is not much better than a line, only useful to drivers to align their trajectory when going fast.

The sidewalk acts like a big line to aim a car's trajectory...

However, as it is low to the ground, it is invisible to a driver looking to the side

The Japanese, again, seem to have this down better than us. When they have sidewalks, they favor fixed obstacles at the limit between car space and sidewalk, and if they have a curb, it is often only the slightest of curb or even just present at the border between sidewalk and street, to provide an indication for cars when they're driving too near to the sidewalk. That way, curb cuts have little to no impact on pedestrians who still have access to a flat, continuous path.

A Japanese sidewalk with a curb cut for a parking lot. The sidewalk and street are separate by trees and planters otherwise, there is also a tiny curb, but the pedestrian corridor is far enough that it is largely unaffected by it.

Japanese sidewalk in Akihabara, note how it is essentially flat with the street, but with small poles separating the two, a much more effective separation

Sidewalk in a recent area of Yokohama, again using trees and small poles to separate the sidewalk from the street, the sidewalk being almost level to the street. You can see parked cars in parking lots, which entries again require no curb cut, simply an absence of pole or tree.

Example of a curb used to separate street and sidewalk, but the curb itself is just 15-cm wide or so (6 inches), it can be interrupted for drains or for driveways and parking lot entries (yes, a few eccentric Japanese own Jeep SUVs in Japan). This case is particularly interesting because of how similar it is to North American examples (sidewalk width, manhole, etc...)

In defense of driveways and parking lots

Some may ask why bother with making curb cuts less disruptive, when we could just not need them by avoiding driveways. The attraction of driveways is actually that it is a very spatially efficient way of inserting parking in an area. Parked cars require space when immobile, yes, but they also need maneuvering space so the driver can drive into that spot and leave it. Like on-street parking, driveways use the street as a maneuvering space, but unlike on-street parking, it requires only about 2,5 meters of width on the street (8 feet) versus 6 meters for cars parked on the street (20 feet), and you can have a driveway deep enough for two vehicles. So that means that driveways can pack nearly 6 times more cars than on-street parking in the same street length... by sacrificing the front yard of course.

This may not look good, but it manages to pack 12 parking spaces in a 25 meters x 35 meters lot (85 feet x 120 feet), to do the same with on-street parking, you would need nearly 100 meters of road (330 feet). Like the aesthetic or not, it remains very efficient, all without having expensive underground parking.

Driveways well-integrated in front yards in Japan, leaving the street uncluttered with cars for the benefit of pedestrians and cyclists

Again, people may ask why bother with packing parking spots efficiently and affordably... Well I agree, but unfortunately, in many places there are still strict minimum parking requirements which limit the number of units that can be built to the number of parking spots that can be built (sometimes with a 1 to 2 ratio). In that context, being able to pack more parking more efficiently can allow for more density, and doing it cheaper makes dense housing less expensive to build. Of course, it's not the best outcome, not at all, but it may be the best we can do in many areas, a necessary first step for more sensible urbanism.

Due to Houston's strident parking minimums, without driveways, townhouses like these would be impossible to build and possible density would be half or less of what is currently being built there.

Another alternative might be to have small parking lots in residential areas, which allows to reduce the curb cuts required and provides great spatial efficiency. The parking lot could also be hidden by trees or small fences to reduce their visual impact on a neighborhood. If one is under-used, it could also be converted to a building.

Conclusion

Yes, driveways are anti-urban... in a certain model of urbanism in which all streets (bare perhaps a few exceptions) are built primarily or exclusively for cars, with pedestrians relegated to small sidewalks. In that instance, protecting the sidewalk from frequent curb cuts and from cars intruding into their narrow corridor does make sense. However, we must keep in mind that this isn't the only model. The Japanese and other Asians provide a different model where residential streets are primarily built for pedestrians and cyclists, so that car drivers feel unwelcome there. In that model, driveways and other off-street parking allows the street to remain uncluttered and welcoming for humans. This is the concept of "shared space" that some have been experimenting with but that have been in use for decades in many Asian countries.

Shared space in the Netherlands

Maybe we simply need to review entirely how we treat pedestrians in North America, and Europe too. We keep building ugly, narrow sidewalks of segmented concrete blocks everywhere when most of the rationale for their existence (muddy, filthy streets, poor drainage of rainwater) is gone. Maybe we should just build a lot less of them and build shared streets instead where cars are allowed but unwelcome, but when sidewalks are required because of street width and vehicle speed (arterials), they should be built wide and to a good standard of quality so as to be welcoming and comfortable to pedestrians.

Why is Montréal so much more affordable than Toronto and Vancouver?

So, many people have heard about the million-dollar homes in Vancouver and Toronto. Many economists are talking of an housing bubble in Canada, and that it should pop anytime soon. However, the situations of Vancouver and Toronto are quite exceptional. Montréal on the other hand still has a balanced and stable housing market, with much more affordable prices. The average house price in Montréal is 325 000$, versus nearly 600 000$ in Toronto and over 800 000$ in Vancouver.

The website Numbeo which allows to compare cost of living between world cities largely concurs:

Montréal is the left column, Toronto is the right column, housing is around 50 to 65% more expensive in Toronto

Montréal is the left column, Vancouver is the right column, housing is about 60 to 100% more expensive in Vancouver

So why is Montréal so affordable versus the two other big Canadian cities? I have my own idea on this.

First of all, here is an idea I don't share: "Montréal is cheaper because it is much poorer". That is not all that correct. Yes, if you look at the average, people in Montréal are poorer, but at the same time, Statistics Canada says that the median family income in Montréal is essentially the same as in Toronto and Vancouver. Montréal has less rich people though, and the population is overall older, so more retirees with lower income, so the average is lower, but when looking at families with couples both in working age, overall Montréal isn't poorer.

Factor 1: less demand

It is a simple fact that Vancouver and Toronto are growing faster than Montréal and are bigger draws on population than Montréal. Much of it may be because of the cultural and linguistic status of Montréal which is a French-speaking city (with a significant English-speaking minority), and so always a bit more isolated from the rest of North America. It is also the fact that Europe is stagnant while Asia is growing, so the West Coast of North America, which deals directly with Asia, performs better economically than the East Coast.

However, this impact shouldn't be overstated. Yes, Montréal isn't growing as fast, it doesn't mean that it isn't growing at all. According to official statistics, Montréal's metro area grew by 1,2% per year between 2011 and 2014, versus 1,6% for Toronto and 1,3% for Vancouver. A bit further back, between 2006 and 2011, both Toronto and Vancouver grew at a rate of 1,8% per year versus 1,0% for Montréal. For a worldwide comparison, Montréal is growing at roughly the same pace as London and Los Angeles, and faster than Paris, New York and Chicago. 

However, it is important to note that neither Toronto nor Vancouver are the fastest growing big cities in Canada, both Calgary and Edmonton have grown faster in recent years. Furthermore, these cities are also much richer, being submerged by oil money. Nevertheless, neither is anywhere near as expensive as Vancouver in terms of housing costs. If housing demand and wealth were the dominant factors, then they would have much more expensive housing, that is clearly not the case.

So, lower demand is a factor, but not as much as one would think, Montréal is also growing, albeit at a slower rate, and faster-growing cities do not exhibit higher housing prices than Toronto and Vancouver.

Factor 2: geographical constraints

The cost of land is crucial in determining the cost of housing. All types of housing "consume" land, and the price of the land will be part of the price of the housing. The value of land is also the result of supply and demand, if there is a shortage of land, then land prices will increase, pulling up the value of housing, especially land-intensive low-density housing. In a metropolitan area, in general, land nearest to downtown is more expensive, because it is more desirable. People who live in a metropolitan area want to access the services and opportunities it offers, and for that, they have to live at a reasonable distance from concentrations of such, and the biggest concentration is downtown.

Montréal's downtown area is located on an island, which itself is on a river, but the island is well-equipped in terms of bridges and roads, with some commuter rail too. As a result, it is able to "sprawl" in every direction, and once passed the river, we are in flatlands, perfect for development. Let's use an arbitrary number and talk of a 20-km radius around downtown for the most desirable land.

Downtown is the blue circle, areas in red cannot be developed as they are waterways, Indian reserves or else

Most of the area in close proximity to the downtown area is thus liable to be developed.

Now, let's look at Toronto:

Toronto's 20-km radius

Toronto's downtown is located right next to Lake Ontario. As a result, it cannot grow to the southeast, depriving it of nearly half the area within 20 km of downtown.

Now, let's look at Vancouver:

Vancouver's 20-km radius

Vancouver is even worse off. The downtown area is located on a small peninsula. To the West, the Ocean, to the North and North-East, mountains. Vancouver residents are proud of the sights they have, of being a "sea to sky" city, but in terms of urban development economics, it's absolutely terrible.

So as lands in proximity to downtown are rarer in both Toronto and Vancouver, it should be expected that land prices would be much higher as supply is constrained, especially as populations grow. Now, you can compensate by using higher density, less land-intensive housing, but here, we come to the next point...

Factor 3: housing supply and typology, both existing and new constructions

First of all, in terms of housing supply, all metro areas manage to build enough housing for the growth in population... of course, housing shortages can restrain population growth by forcing people to go elsewhere for affordable housing, but people can squeeze by through sharing apartments and the like before they have to leave.

Anyway, the 2011 census provides both the number of private dwelling units and the number of dwelling units presently occupied by permanent residents. This isn't perfect evidence, but it can provide a hint of the vacancy rate of housing.

Share of private dwellings occupied by permanent residents, be careful, the graph doesn't start at 0 in order to illustrate differences better

So there doesn't seem to be an oversupply of vacant housing in Montréal versus Toronto and Vancouver, though pied-à-terre are more common in the last two.

As to comparing population and housing growth between the 2006 and 2011 censuses, here is the data, first for each city, then for the entire metro areas:

Rate of growth of population and housing in the core city, for all 5 years (not annual rate)

Rate of growth of population and housing in the metro areas, for all 5 years

So it's interesting to see that Toronto is building very fast, at a rate outpacing population growth quite a bit. Montréal sees a lot less construction, but still builds a bit more than population growth. Vancouver is a more sensitive case, with population outpacing housing growth a tad. If anything, Vancouver does seem to show a city where population growth is truly constrained by housing, whereas it doesn't seem to be necessarily the case in Toronto.

Maybe the problem is not so much quantitative than "qualitative", meaning not just the number of housing built, but also the type of housing. As I've said many times, the higher the building, the more expensive the square foot to build. So low-rises are cheaper to build than mid-rises, which are cheaper to build than high-rises, for a similar unit size. Likewise, higher densities and less land use result in cheaper prices. So, following those informations, we can imagine that a graph of density versus unit cost of housing types would probably look a bit like this:

Housing typology by density and cost per unit

SFD: Single-Family Detached

SFSD: Single-Family-Semi-Detached, smaller units and land consumption results in lower prices

SFRH: Single-Family RowHouses

LRA: Low-Rise Apartments

MRA: Mid-Rise Apartments

HRA: High-Rise Apartments

Of course, this goes by North American conventions (single-family detached can be very dense and small in Asia or Mexico, but rare over here) and is variable with land prices. I already spoke of the different housing typologies of the different cities in another post. Here are the graphs for a refresher:

Percentage of housing units per type in the main city

Percentage of housing units per type in the metropolitan regions

Percentage of housing units per type in the suburbs only

So again, Montréal has a great amount of low-rise apartments, the type of housing that is most affordable and yields decent density, Toronto is a city of extremes, with plenty of single-family housing and high-rise apartments, but little in between (the "missing middle" housing options, thanks to LetsgoLA for referring to that website). Vancouver is more balanced, with a particularity of a great number of low-rise apartments in its suburbs.

But, I'm going to push the analysis a step further here. Existing stock is all fine and well, but I think the price of NEW housing is much more influential. The market for housing is just the market of units for sale, not all existing units, as most people are not looking to sell their homes or properties. As a result, since all new constructions are for sale but only a small proportion of older units are, the impact of new constructions on the housing market is vastly disproportional to their proportion relative to the total amount of units. For example, if I have the right data, the census says that on average, Toronto has built 37 000 units per year between 2006 and 2011, and the real estate association speaks of only on average about 85 000 real estate transactions per year. So though the new constructions are only 1,9% of the housing stock each year, they represent about 45% of real estate transactions if we assume new constructions are sold the year they are built.

I have already spoken of my belief that what limits the value of current housing is the cost of the alternative of building new housing (1, 2, 3, wow, I need to stop repeating myself). If new housing is very expensive to build, then it won't directly lower housing prices of the existing stock, it may do some filtering, but only if housing grows much faster than population. If prices ever fall and new housing is expensive to build, then construction will stop dead in its track until prices increase again.

So let's see what kind of housing each city is adding:

Typology of new constructions for core cities

Here is the data only for the core city. What is glaring here is that neither Toronto nor Vancouver are adding low-rise apartments in any kind of amount. Vancouver is adding quite a bit of duplex units... but these look like duplex conversions as Vancouver has lost nearly a thousand single-family detached homes over the 5-year period and gained about 3 000 duplex units (for a net increase of only about 2 000 units). Almost all of Toronto's and Vancouver's housing growth comes from high-rises... the most expensive housing option per square foot! Meanwhile, Montréal is still adding quite a bit of low-rise apartments, which are the cheapest housing option, if it's possible to build.

Now, let's look at the entire metro area:

Typology of new constructions for entire metro area

Here, we can see that both Montréal and Toronto keep building plenty of detached single-family homes in suburbs, but Vancouver isn't. In a context of high land prices, single-family detached homes can be supremely expensive, but the new units are built far from the center for the most part, and so can be relatively affordable. Again, apartments with more than 4 stories are massively being built in both Toronto and Vancouver, and they are very expensive per square foot. In fact, Montréal is building more low-rise apartments than Toronto and Vancouver together, despite a smaller rate of construction and half as much people as Toronto and Vancouver together.

Finally, let's look at suburbs alone:

Typology of new constructions for suburbs

Even in the suburbs of Montréal, low-rise apartments are being built in great number. In some close suburbs, like Longueuil and my home suburb of Boucherville, low-rise apartments have been over 60% of new units built. Meanwhile, Toronto suburb are still single-family land, with three quarters of new constructions being various types of single-family houses, and most apartments are high-rises or mid-rises. Vancouver's suburbs are very varied in their constructions, from what I see from real estate sites, it seems to me that where low-rise apartments are built, they tend to drag prices down around them, even for high-rise apartments, but that is hard to show.

So, what does this mean?

Well, if I'm correct and the construction cost of new housing is what imposes a "ceiling" on the price of existing homes, then it is to be expected for prices in Vancouver and Toronto to stay quite high, as their new constructions are almost only high-rise apartments, which are extremely expensive to build (though they still help slow the growth of prices). Much of the rest of the existing housing stock is detached or semi-detached houses, and in a context of severe land shortages, these become extremely expensive because they consume a lot of land.

Montréal may be showing another way here for affordability, with the typical and traditional emphasis of housing on low-rise apartments of 3 or 4 stories. This is a type of housing that is quite affordable to build and can reach decent density levels even in suburban areas with 2 parking spots per unit required, and quite high density levels in cities. These are being built even in suburbs, offering entry-level homes for 200 000$ for decently sized units (1 000 square feet or so), which helps draw pressure away from both single-family homes and apartments by cannibalizing the entry-level market of the first and the top-level market of the latter.

So, in other words, this is what affordable housing looks like:

In an urban setting, with low parking requirements, these achieve about 150 units per hectare, or 60 units per acre, with just 3 stories (plus the basement)

Also in an urban area

This is in a farther suburb, density of around 50 unites per hectare, or 20 per acre, but each unit is about 1 000 square foot in size and accommodates 2 or 3 bedrooms

Another case of suburban low-rise apartments that provide adequate size for an affordable price

This is in Longueuil, and a suburban area too, they have huge parking lots behind them, so density is only about 60-80 units per hectare (24-32 per acre)

Last example of suburban low-rise apartments, these can be had for about 190 000$ according to real estate listings

 On the other hand, these...

 ...and these...

... and these...

...are certainly dense and contribute to walkability and urbanism, but they're not particularly cheap to build, and thus not particularly affordable to buy. It would be one thing if Toronto and Vancouver already had maxed out their core area and suburbs with low-rise apartments and row houses and that the only way to add units would be to reach for the sky, but that is far from the case, most of their core cities are covered by low-density single-family housing. Montréal has the advantage that zoning in most of the city, if it's not keen on high-rises, will at least typically allow low-rise apartments, and urban planners in the suburbs are open to the idea of allowing plenty of low-rise condos (and pressured to).

I'm not saying that building high-rises is bad, not at all. Better for richer people to pay premium prices to live in high-rise, ultra-high-density housing than for them to compete with the middle-class for lower-density housing, pushing their prices up. The problem is not building high-rises, the problem is the lack of low-rise and mid-rise construction in Toronto and Vancouver, especially near the downtown core.

Now, this doesn't necessarily mean that the low-rise condos of the Montréal metro area are perfect, far from it. I've pointed out before they are often built in unwalkable areas, or areas with poor walking environments and that Montréal should be more open to the idea of allowing more TOD around subway stations and mid-rise and high-rise in existing low-rise areas. Still, as far as housing affordability goes, I think Montréal's approach is much more effective, because the housing that is built is cheaper to build and still gets great density.

Conclusion

So, anyway, that is my take on which Toronto and Vancouver are so expensive, while Montréal remains affordable. Toronto and Vancouver both are cursed with badly located downtowns that can only sprawl in a few directions, which exacerbates land shortages, and both cities are attempting to satisfy the housing demand almost exclusively with high-cost high-rise constructions, which may help keep price increases under check but cannot directly lower housing prices, not unless there is major overbuilding leading to massive filtering. What they need to do is embrace low-rise and mid-rise apartments and upzone their core areas for it.

Attached or detached: townhouses and density

One of the big differences that is immediately obvious to anyone visiting Japan is the almost total absence of what is a clear favorite of European and American planners and urbanists: the townhouse. This form of high-density family housing where houses' lateral exterior walls are joined is clearly beloved by a lot of them.

Random street in London, the UK is particularly fond of townhouses that they called "terraced house", nearly 70% of the population lives in attached homes

The presence of these buildings built wall-to-wall is common to almost all European cities, it is typical of European classical urbanism.

Even though the Japanese have plenty of high-density areas full of single-family houses, these areas almost always steer clear from townouses. They do have a few of them, mostly in older areas, and some developers build apartments in townhouse format rather than in the typical format.

Apartment building built in a townhouse format in Sendai

But in most cases, the Japanese avoid shared walls at almost any cost, even in areas where it would seem to make a lot of sense:

This is an image from Osaka

These houses have been built at the same time, probably by the same developer (traditionally, the Japanese built their own homes, but in recent years, developers have taken to building entire neighborhoods of single-family houses and selling them), yet despite their proximity, they are not attached, with the smallest of space between each of their exterior walls.

The basic logic of the townhouse

There is a simple logic behind townhouses and attached buildings in urban areas. The biggest attraction point of single-family housing over multi-family is the promise of one's own personal yards, access to a small portion of land that is shared with no one. The back yard is the single most important selling point of single-family housing. The front yard is still important in preserving privacy and offering a customizable façade for the house, so that's another big selling point.

If we look at a single-family house schema, we'd get this:

Basic schematic representation of a single-family house

Now, considering in general the depth of the plot is imposed by the street network and that there is only one household on that plot of land (let's ignore accessory units here), then the only way to achieve higher densities is to play with width, to make the plot narrower.

A typical house tends to be wide but shallow, a way to increase density is thus to rotate it, so the shorter side is facing the street.

This achieves nearly twice the density of the house above

But people can look at this and say: "The back yard has value to the owner, the front yard also... but what about the pieces of land on the side of the building? They are too narrow and small to be of any use, they're not really visible from the street and they don't really help much for privacy". So a seemingly obvious solution is to simply get rid of them, just build homes wall-to-wall.

For urbanists focused on "feeling" or on the look of areas, the presence of buildings wall-to-wall reinforce a feeling of enclosure, of unbroken lines of buildings forming walls for the street, which they frequently seek as it evokes European cities.

The problem with townhouses

Townhouses seem to make a lot of sense, but they have one big flaw: the lack of exterior walls on which windows can be installed.

It is a factor that may be often forgotten, but windows are crucial for proper living quarters. It's not just about the view, but about air flow and natural illumination. In some jurisdictions, it is downright illegal to have a bedroom without a window to the outside, and even when it is legal, people avoid them like the plague, with good reason. Even a window opening on a wall within a few feet of it is considered much more desirable than no window at all.

Of course, in the past, before efficient house heating, building houses side-to-side reduced heat loss in winter and made it easier to keep living areas warm. However, today we no longer have problems with heating, so that advantage is much less significant than in the past.

If you look at the picture of the Japanese homes, you can notice windows on the side walls, even with the tiny amount of space available. I also have seen such windows in Tokyo Hotels, for example:

Photo from the inside of a Tokyo hotel room

With the window opened

A picture taken right out of the window, showing what space there is between the two walls

So, visually let's compare a small detached house with a townhouse of the same size:

The blue bars indicate where windows can be installed, the red bars show where they can't

In this case, supposing a 7m x 7m (23ft x 23 feet) detached house with a 5m x 10m (17 ft x 33 ft) townhouse, the detached house has 28 linear meters of walls per story that can be equipped with windows, of which 14 meters are the front and back, with better air flow and light. The townhouse on the other hand has only 10 linear meters of walls per story that can be equipped with windows.

This also means that the detached house can have more bedrooms for a same floor area as bedrooms need to be located next to an exterior wall with windows.

There is a major difference between Japan and North America that may explain partly why the Japanese favor small detached homes over townhouses. In North America, parallel streets are often separated by 60 meters (200 ft) or more, as a result, lots tend to be very deep. In that context, reducing width is essential to get some density. Meanwhile, Japanese streets are built much more tightly together, separated only by 35 or 40 meters (120 to 130 feet), so the Japanese don't need to build narrow townhouses to get high density in single-family areas, the lots can be wider because they are a lot shallower:

Left: townhouses in a typical North American street grid, right: small detached homes in a Japanese street grid, both images have the same density

Left: a random neighborhood of Minneapolis, right: a recent random single-family area in the small city of Sano in Japan, yes, the images are to scale

This is true for townhouses, but also for apartments. Apartments built in depth with little frontage will also struggle to find ways to have many bedrooms because of the lack of windows. This is why Montréal, which housing stock is primarily made up of low-rise apartment buildings built side-to-side, struggles to have more than 2 bedrooms per unit in that type of housing, as it is very hard to design an apartment that has three bedrooms with a townhouse-style floor space with windows only on the narrow front and back.

In the past, people have found a way to go around this problem by using a fat, reversed L- or T- shape.

Apartments with a T-shape in Montréal, separating the walls in the back of the buildings to create more exterior walls and opportunities for windows

Another example in Montréal, more of an reversed L-shape here

Now an American example, Philadelphia, with its traditional rowhouses with L-shapes

The result of this effect is that it is a very clashing experience to live in a detached home and in a townhouse. The lack of windows on the townhouse can be downright claustrophobic for one used to living in a detached house, like the younger generation is.

A compromise: the semi-detached

The semi-detached is a compromise between the attached buildings and detached buildings. With just one shared wall, the semi-detached can still have plenty of windows while reducing the lateral buffers between buildings. This can provide for greater density while maintaining some of the attraction of detached houses, it can also provide a way to deal with the deep North American lots by allowing for the construction of very deep houses or apartments, increasing building coverage to the lot that would hardly be possible in an attached context, and increasing the possibility of building more bedrooms per unit.

Semi-detached houses in the UK, where the style is quite common in more suburban areas

Semi-detached houses in Boucherville, double the density of typical bungalows, but each unit is still wide enough that it doesn't necessarily require side windows

A multi-family type of semi-detached, each floor has two units built in depth with windows mostly on the side of the building, as a result on a 30m-deep lot (100-ft), the buildings are 15-m deep (50-ft) and the apartments manage to be just 5m-wide (17-feet) yet have 2 bedrooms each, which would be impossible in an "attached" situation without making one bedroom just 2m-wide or less (around 6 feet)

Typical condos found around Montréal, which are basically two semi-detached triplexes built in one building, providing big (1000+ square feet), affordable housing options in suburbs

Conclusion

If attached buildings and townhouses represent a very efficient spatial organization and are typically associated with urban areas, especially in Europe, they are not necessarily required for density and good urbanism. Though they are also well adapted to the typical deep North American lot, townhouses' lesser access to air and light may be a turn off for a generation born in suburbia who may crave more density but who still remains attached to a detached lifestyle (pun not intended). This may also explain why I personally find residential areas in Japanese cities more attractive than European cities, due to better aeration and light access, as I was raised in a North American suburb and am more used to the more open, less enclosed feel of them.

Oh, and finally, a last point about why the Japanese skirt around attached buildings: remember that the Japanese think that buildings have a limited life expectancy, that they should be torn down and replaced every few decades, whereas the European/North American approach is to consider buildings to be built to last at least a few lifetimes. The Japanese approach of tearing down and rebuilding is much easier to do with detached buildings than with attached buildings where tearing down a building without affecting the ones on either side with which it shares walls and which belong to different owners can require a lot of attention and care.