Dan Luu writes:
Here in Vancouver, the expert consensus is that the probability of a magnitude 9 earthquake is something like 0.1% to 0.4% per year. . . .
When we looked at automotive safety in 2020, we saw that almost every car manufacturer does the minimum necessary to score well in crash tests and no more, assuming no one is cheating. Since then, it’s been found that there has been major cheating for crash tests. In 2024, Toyota was caught cheating, with a number of cheats, including cutting away a panel inside the vehicle that would cut the test dummy on impact.
After Toyota was caught, it also came out that Mazda, Honda, and Suzuki were cheating test results as well. These are all Japanese manufacturers because this came out when the Japanese government investigated cheating (it’s unknown how much cheating would be found if other car companies were investigated).
Should we expect building safety any different? Let’s start with what we know and can verify. . . .
Assuming the building code is designed correctly, buildings in Vancouver that are built to the modern building code are designed for “safe egress” after a severe earthquake. . . . However, from talking to a civil engineer who used to do some building inspections, it’s not uncommon for companies to try to cheat inspections. . . . one thing we know is there’s a moderately high rate of attempting to cheat inspections, some of which is caught by inspectors. . . .
If you talk to trades in Vancouver about how this works, the builders give contractors timelines and budgets that are impossible to meet without severely cutting corners. . . . it is also the case that their buildings often have serious issues due to cut corners.
And he’s got an example:
The building I [Luu] am living in now is one such building with serious issues. Within 30 minutes of looking at the place, I found a double digit number of issues, ranging from minor “fit and finish” details to significant safety issues waiting to happen. I still rented the unit after seeing this because the property management company that’s renting the unit out cuts a lot of corners, including on market research, so the unit was severely underpriced and, because Vancouver’s rental vacancy rate was very low at the time, I might wait a year before finding another unit that roughly met my other criteria.
That’s funny, the bit about them cutting corners on market research. I just hope there’s not a level 9 earthquake when Luu is in the building!
He continues:
After I rented the unit, the building had an entirely predictable catastrophic failure that impacted something like 1/3 to 1/2 of units, forcing people to move out for months (maybe 8 months?) as the impacted units had to be torn apart to be repaired. The failure also caused half of the elevators to be inoperable for the same time period, so I suppose it’s sort of lucky that so many people had to move out, otherwise the wait for the elevators would’ve been very long. If you look at projects done by mid-tier builders, like the builder of “my” building, they do seem better than the bottom tier builders. Instead of every building having catastrophic problems, it’s only some fraction, perhaps 10% to 25%? And, in terms of what the builder cares about, the catastrophic problems often show up after the initial warranty period, so the builder should not see increased insurance rates on future buildings and generally won’t have liability for the problems, unless a degree of negligence that’s very difficult to show can be shown (I would be very surprised if it’s shown in this case, even though the issue was easily preventable). Unlike the worst builders, mid tier builders seem to know how to cheat in ways that make them money instead of losing them money.
Interesting multidimensional analysis here. This reminds me a bit of cheating in science: The best scientists don’t cheat, but as you go down the scale you get different rates of cheating. Mid tier and lower tier scientists don’t necessarily cheat; often they can find niches to do their work–and, depending on where they’re working, they can still make useful contributions, in the same way that a non-cheating mid or low tier builders can still construct useful buildings, if the financing is set up appropriately. But, when they do cheat, the mid tier scientists might get away with it (I’m thinking of pros like that voodoo guy from Ohio State), but the low tier scientists like these bozos at Harvard might get caught. On the other hand, financially speaking, the Harvard fraudsters are hardly low-tier, and indeed they’re so well connected that even after the fraud story came out, they received fawning news coverage, so maybe this relates more to Luu’s general point, that if there are many benefits to cheating and few consequences to being caught, that lots of unscrupulous but rational people will be motivated to cheat.
Luu continues:
If we look at top tier builders that are the most reputable builders, it’s easy to look at some units and see that fit and finish details tend to be better, but fairly funny issues still occur at a higher rate than I’d hope for as a potential buyer. For example, here’s the a photo of the one window that opens in a unit from a builder that’s reputed to be among the best in Vancouver (it’s possible this reputation is undeserved, but for people looking at buying, reputation is generally the best thing most people have to go on). The window can only be opened a bit because the path the window would open to is blocked by the balcony railing of the adjoining unit.
I love that parenthetical there. Luu is so careful in how he writes!
And this:
If we look at the Japanese car company cheating scandal in more detail, I think we can see similar forces in play (although Japanese cars are built to a much higher standard of quality than Vancouver condos). On the surface, some of the cheating, such as cutting away a panel that would cut a crash test dummy, seems like malfeasance. But most of the cheating was incorrectly reporting that a test was done according to the parameters when the test was either done in a more challenging way or in a way that would be very unlikely to materially impact the test. For example if a test was supposed to be done at 50 km/h, but it was done too fast, say, at 53 km/h, an engineer would report that it was done at 50 km/h. In another case, for an impact test, a sled that’s too heavy was used (it was the standard weight for North American crash testing, which is heavier than the standard weight for Japanese crash testing).
It doesn’t appear that there was some kind of master plan in place to cheat tests in the most efficient way possible in order to save money or anything like that. It looks more like engineers didn’t want to have to deal with doing extra work. In some cases, this meant lying about how the test was done in a benign but still fraudulent way. In other cases, this meant cheating the system in a not-so-benign and fraudulent way. The whole scandal seems like it could be filed under “the banality of evil”. People seem to just be trying to make their day a bit easier without much care for what the putative goal of crash testing is, making cars actually safe.
Luu concludes:
If you look at the overall situation and ask, what are the odds that these builders are not cutting corners and cheating building codes in safety critical areas where they’ll only be exposed if an event that probably won’t occur in my lifetime, I think the the answer has to be that the worst builders are definitely cheating and it’s plausible that reputable builders are cheating as well. . . .
When we step back and look at what the worst builders are doing, it seems like they’re doing what engineers at Japanese car companies did, but more. If they had some kind of well thought out master plan for cutting costs, they wouldn’t cut costs in ways that caused them to get sued and pay out much more than they would’ve saved by doing a standard slapdash job, but there’s no plan, so their cost cutting frequently costs money instead of saving money on top of creating higher future insurance costs and producing a bad reputation that reduces what future units sell for.
What’s it all mean? I agree with Luu’s take:
When I see public discussions about this kind of thing, people often say the problem is greed. That doesn’t seem right to me, in that, if the developers were extremely greedy, they would take unethical shortcuts that make them more money. Instead, they’re taking unethical shortcuts that sometimes make them money and sometimes lose them money. If you wanted to phrase it in terms of a cardinal sin, you could maybe call it sloth, but I don’t think that’s really right either, at least as the term is used colloquially.
So much bad behavior, in so many realms. seems to me to be explainable by laziness.’
Also, a few years ago I was told that NYC is likely to have a major earthquake sometime in the next 1500 years, and when that happens it will cause lots of damage.
Tangential at best, but I immediately thought about earthquake insurance in Anchorage Alaska. I lived there for 12 years and when I insured my house, insurance against earthquake hazards was a separate rider you had to buy (standard insurance did not cover such damage). Anchorage is particularly earthquake prone with a large scale damaging earthquake in living memory.
I looked at the geological maps which should the degree of earthquake susceptibility in different areas of Anchorage – it varies a lot depending on whether it is near the shoreline (very prone to severe damage) or on the hillside on quite stable ground. The surprising thing was that the price of earthquake coverage was uniform throughout the region. There was no differentiation based on location. This was even more surprising given that house location is pretty fixed and verifiable, unlike things like driving habits and auto insurance coverage. I have also lived in wildfire prone areas (and lost my house to a forest fire) and insurers did not vary their rates depending on an inspection of whether there was defensible space around the house (that may have changed by now, but wasn’t the practice back in 2000).
This doesn’t relate to the building practices referred to in the post, except in the most tangential manner. After all, if the insurers don’t care about real exposure to hazards, then perhaps builders won’t care much either. The explanation of why insurers are so un-careful about setting their rates has always perplexed me (though I suspect it has something to do with lack of competition, although that really doesn’t explain why they wouldn’t care, only why they might get away with it).
Dale Lehman and I currently live in a zone that almost never has earthquake tremors. Although I have never been to Alaska, I did live for several years in an area that is prone to earthquakes: Mexico City. Word has it that “People get accustomed to it” in Anchorage.
From my experience, people can live with semi-regular mag 5 earthquakes but once they are over 6, and they come with damage, then people’s anxiety really ramps up for a very long time, not helped by a long period of after-shocks that comes with earthquakes at those magnitudes.
Andrew:
Speaking of earthquakes and Vancouver, here’s a link to Kathryn Schultz’s bone chilling (and Pulitzer Prize winning) 2016 New Yorker account of what the “The Really Big One” will be like when it arrives!
https://drive.google.com/file/d/1txjo83qCcCfovG7kxsMU98Hijlt6LUHy/view?usp=drivesdk
Even NASA which certainly has a code of very exact rules has had disasters. Apollo 1, Space Shuttle Columbia, and Space Shuttle Challenger were grim events. Apollo 13 was a close call. Bad things can happen! We can not expect flawless performance by any human being.
Earthquakes are interesting. I wonder if a meeting of architects, engineers, and public safety officials from Tokyo, Anchorage, and Reykjavik might yield worthwhile ideas for all. Is there such a forum already?
On the point of the post: I think it is greed, not laziness. Almost all house purchases are made without really understanding the quality of the work (I’m speaking of myself and the vast majority of people, who like me, don’t really understand construction details). Much of the work is hidden behind walls. The exact type of pipes and wiring used is hard to gauge. There are codes and inspections, but these are imperfect. This is a classic case of asymmetric information, and when it is present, we should anticipate that those with the better information (builders, in this case) will abuse that at the expense of consumers – unless we devote resources to preventing such outcomes. The optimal solution is always to devote such resources, but only imperfectly.
In the case of earthquake- resistant construction, this is particularly severe (I think) because the quality of construction is even harder to gauge and the codes are more variable and somewhat vague.
I can’t help but think of the statistical analysis analogy. The analysts’ forking paths are difficult to monitor and the data is often not available. There is asymmetric information between the authors and the readers. If people are greedy, we should expect similar outcomes to those above. We can and do devote resources to address this: peer review, editors, potential retractions, reputational effects, etc. These are imperfect and it is optimal for them to be imperfect (though I would argue they should be less imperfect than they are – there is no reason to expect the current situation to be optimal). These results derive directly from greed, regardless of whether laziness also plays a role.
I think to improve things would require more directly attacking the incentives that greedy individuals face. For construction, it might be more legal liability, though I think there are legal issues that would make that a costly change (e.g., determining who was at fault when an earthquake destroys a home). For academic research it seems simpler to me. Increase the reputational costs of falsifying data, distorting analysis choices, or selective and overstating results and less of it should occur.
The other – and better – solution is to breed less greedy researchers. Exactly how, I’m not sure.
> I think it is greed, not laziness
I agree with what you say, but both these words feel like they simplify to the point of possibly obscuring the situation. Maybe it’s like the honesty and transparency isn’t enough thing — there are probably contractors out there trying hard and who are operating at their financial limits that are also producing bad work.
It’s probably cleaner just to evaluate the outcome than ascribe some moral character to it like greed or laziness.
Vancouver felt a nearby 5.1 earthquake about a year ago and it is my impression that the risk is more in front of peoples’ minds since then. I know of at least two departments at UBC that are currently working to increase their buildings’ resistance to earthquakes. I guess some things easily slip under the radar when other things are perceived to be more pressing, whether that is to deliver a building or to publish. All the more reason to have good control systems in place to not let it be forgotten.
I can’t tell without more details, but this may not actually be cheating:
“cutting away a panel inside the vehicle that would cut the test dummy on impact.’
In the 1990s, I used to visit car companies often, as SGI workstations were used to visualize crash simulations and then over time, SGI supercomputers replaced Crays at enough lower cost to drastically raise the number of simulations done.
Ford showed us a sim of a Taurus crash, where simulated dummy pitched forward, bounced off airbag and …
and skull was crushed by metal strut of roof.
This showed up as bright yellow flash highlighling extreme stress, and it only happened at a very narrow range of crash angles, so unlikely to have found it in more limited nubmer of real crash tests.
The fix was to weaken the strut, in fact something commonly done to metal in that era to create more “crumple zones,” which auto engineers would often mention to me..
https://en.wikipedia.org/wiki/Crumple_zone
The Computer History Museum (which Andrew has visited) has an exhibit on this, including a real crashed Taurus Ford kindly donated to us:
https://www.computerhistory.org/makesoftware/exhibit/car-crash-simulation/
For what it’s worth, engineers I talked to cared about safety. The Swedes were especially intense. Saab engineers worried about moose collisions.
Searching around, it sounds fair to characterize the panel cutting as cheating
> In some side impact tests, Daihatsu says, “the inside lining of the front seat door was improperly modified.” Test engineers, Reuters reports, cut a small notch in the door “to minimize the risk in testing that the door interior could break with a sharp edge and cause injury to an occupant.”
That is kbb citing Reuters citing who knows what. Though there it’s called a notch, which seems like a different thing than a panel, so I guess you’d have to dig farther.
Yes, if they altered the car just for test, that’s cheating. If they incorporated the change into the design, that’s nirmal evolution.
When you need a construction job done what are you supposed to do with multiple bids? Some decades ago I solicited bids for a small bathroom remodel. Low was c. $5,000, high was c. $15,000. Okay, now what? Ceteris paribus you choose the low bid, but it ain’t paribus by a long shot. I can only imagine the subcontractor bidding headaches with a high-rise apartment building.
Was this before or after you finished working on St Paul’s Cathedral?
Had a buddy doing some work recently where he had to get contractors and whatnot. The consensus from people we knew involved in more construction was it’s better to pay a low price for low quality work than a high price for low quality work.
I assume within the contractor world people can kinda work out the reputations more easily, but with projects like these, the building only happens once, so it just comes down to guessing.
There was a required inspection that had to be done by a private company. As I understood it, the inspection failed, but there was apparently an exception that if the problem was too expensive to fix, then everyone could just sign off and ignore it.
The economist in me feels compelled to point out that laziness is monetized as the value of leisure time (as opposed to work time). However, the key point is that this isn’t often industrial scale cheating!
The analogy of the building code being inadequate and it being ok because the pricing was also low reminds me of this great paper:
https://diegogambetta.org/wp-content/uploads/2022/06/the_ll_game.pdf
I have to say that the abstract of that article is one of the worst I have ever read (I haven’t read the article – it may actually be excellent, so I am only commenting on the abstract). To begin with, the empirical observation about Italy is unclear to me. I’ve never lived in Italy but I’ve visited a number of times and I can’t make heads or tails out of what behavior the article is referring to. Then, the discussion of the equilibrium in the game and its contrast with the prisoners’ dilemma also seems opaque to me. For the record (apology for putting AI output in), Chat GPT offered this translation of the abstract into plain English:
The authors describe a situation they’ve often seen in some Italian institutions: people promise good-quality work or services, but end up delivering something worse. Outsiders see this as cheating, but insiders actually get used to it—and even depend on it.
Surprisingly, people in this system don’t get upset about poor quality. Instead, they may feel uncomfortable with or even avoid people who consistently deliver high-quality work. There’s a kind of unspoken agreement where everyone provides lower quality, and that becomes the norm.
Unlike typical situations (like the idea behind the Prisoner’s Dilemma, where people act selfishly to get the best for themselves), here people are still acting in their own interest—but in a strange way. They prefer a system where everyone, including themselves, produces lower-quality work, because it removes pressure and expectations. It’s easier and more comfortable socially.
To keep this system going, people develop subtle social rules that discourage high performers and protect the low-quality status quo.
The main takeaway: poor outcomes don’t just happen because of a few bad actors—they can persist because groups of people collectively settle into, and even defend, a culture of mediocrity.
Perhaps this is an example of how AI might improve academic writing?
I like it because it’s funny, rather than because it’s a particularly high quality article. That said, I have seen a fair few complaints about it, usually from southern Europeans annoyed at being badly stereotyped.
I think he should distinguish between company and individual. The individual cutting corners due to laziness is benefiting from cheating. The “company” (or its shareholders) benefits if individuals cut corners in ways that increases profits directly.
Also, it’s unfair to paint Japanese car manufacturers as the main cheaters. I don’t know about crash tests but when it comes to cheating EPA regulators with “defeat devices”, it has emerged that cheating has no boundaries! It started with VW but here is Fiat Chrysler: https://www.reuters.com/business/autos-transportation/exclusive-fiat-chrysler-nears-plea-deal-us-emissions-fraud-probe-sources-2021-10-26/. Here is a probe into Bosch that may involve Ford and GM: https://www.reuters.com/article/business/environment/german-prosecutors-may-widen-bosch-emissions-probe-idUSKBN1HR2FT/
A couple points:
1. The US and Canada have very strict building codes for earthquakes. Building quality is far lower in places like Turkey and Armenia despite even higher danger. So it feels a bit first world to complain about corner-cutting in Canada.
2. There is no such thing as making a building resistant to a 9 on the Richter scale. Areas that are on fill will shake like jello. As the saying goes here in the land of the tremors (I live near Seattle), do you know what to do when the number 9 hits? Put your head between your knees and kiss your ass goodbye!