I’ve long heard and even held the refrain “They Don’t Make Them Like They Used To” with regards to good old American craftsmanship and ingenuity. There is a compelling notion that the old ways are best, and that the future is forever evaporating into cheap air.
The thing that shook me out of this nostalgia was a scene in Frank McCourt’s “Angela’s Ashes”, which described the family tearing apart their house’s walls to collect firewood. Apparently a lot of houses were built cheaply (family of five able to tear open walls to get to precious, flammable studs within), and treated shabbily, even then.
I also thought that people built things sturdily in the past because they were at the base of accumulating knowledge about structures. The same accumulation that allows us to build so cheaply and efficiently today. The first office building with elevators (Equitable Life Building, NYC 1870) at all of 7 stories had walls on the first floor over 5 feet thick just to support the weight above. It would be almost 30 years before the steel framed skyscraper made such support concerns superfluous.
As another example of this, the Brooklyn Bridge (1883) still stands across the East River, sturdy as ever, as the longest suspension bridge in the world for years after its construction.
But I found out while reading Henry Petroski’s book on failure that the Brooklyn Bridge was built six time more sturdily than it needed to be, even by engineering calculations of the day, because the design staff did not the trust the cable supplier to provide adequate cables for the bridge. What I always thought was a naive design was just a quality assurance issue.
The quaint refrain from the beginning of this article is the one that got me asking, what data is there to show that they don’t build them like they used to?
I noticed that the Census keeps a tally of “Year structure built” with each decade. The quality of these tallies varies, at least in the way that they are presented to the public. It seems the Census is still incapable of presenting more that two decades in their data portal, and that they shut down their old data portal with every update of the new. What’s worse, the Census has begun putting questions like “Year Structure Built” in their American Community Survey (ACS) questions, and not in the regular census long form. This would be fine if the ACS provided reasonable numbers, but as you will see the data has difficulties.
As far as I can tell, the Census Housing Survey has been collecting data since the 1940 census. This is probably why every census has “Built before 1939” as its starting year. In 1940, the starting year was “Built before 1859”, which is just badass. The overall accumulation of housing in the 70 years since then is impressive, more than tripling in that time
The answer to the question I started with seems to be at hand. All we have to do is see how many houses disappear form the housing stock from each decade, and compare slopes to see which is the sturdier of them all.
The problem is, the Census data add houses to some age groups between censuses, frustrating me no end. I wanted to post this anyway, as a valiant example of negative results. People don’t post those often enough.
Here’s the negative result : The census “year structure built is Useless for studying structure durability over time”
At least the proportions decline over time, which is what we’d expect in a world full of 3D printers but no time machines.
This showed that indeed, houses built before WW2 constituted great deal of the housing stock to be had until the 1970s. The great depression of the 1930s basically shut housing starts down, and here’s the long term implications of that. A famous reason for the suburbanization of the 1950s was that there were no houses for people to go upon the return of family life after the war. This is why houses from 1940 to 1950 are the lowest performers of them all. Not many were built at all, in the age of national rationing. Maybe people thought the old houses were built well because they were the majority of what was around for so long.
I wanted to look at the last shred of data that the census of housing was going to give me, so I wrangled the data from 1940 into this graph, showing the distribution of housing builds between the 1850s and 1930s. This showed pauses and absolute declines in response to WW2 and the Great Depression. Its hard to say specific things about those events without knowing the background rates of declines, which, using Census data, is not easy.