Conspiracy Theory

Is it possible this was some sort of simulation/development code that got left in by mistake? No, I don’t think so either, but I wondered if they’d tried that as an excuse.

Maybe it’s the truth, but they’re too embarrassed to admit it, for damage to the image of German technical competence?

I think you overestimate the number of people that needed to be involved.
This could all be done inside the engine management software team.

It would only take one senior manager to have the idea and get them working on it.
Someone probably sold the idea to the workers with a flawed logic like:
“The test is artificial and it’s putting our cars at a disadvantage. We think the competition are getting around it by doing something like this. So we should too.”

There was a whole spate of similar incidents with graphics cards and CPUs having special modes when they detected standardised tests.

There’s something toxic about the software developer mentality that leads to this kind of “optimisation.”

How many people were in on the JFK assassination? Or rather, how many people were in on the Oswald assassination?

Sounds like the theory described in this post applies:

https://crookedtimber.org/2015/05/26/libor-for-the-universities/

So “everyone cheats” gradually becomes something that everyone knows. It’s not a conspiracy because no-one has to work hard to keep the secret. The only people who would even understand a casually blabbed secret are other engineers already rendered cynical by a lifetime in the industry.

That is until one day some unfortunate bastard finds themselves under a spotlight having to explain the same thing to outsiders who don’t share the same cynicism, and suddenly it all looks very very bad.

Of course for this theory to work the problem must be endemic to the entire industry. Looking forward to seeing how that plays out over the next few months…

RE how many people were needed: I dunno, but maybe not all that many. The software that lets the engine be adjusted to beat the test is pretty legitimate – the whole point of this fancy computer-controlled engine is that the computer adjusts settings to balance performance, efficiency, and emissions, so adjusting to minimize emissions or to maximize performance are all part of its normal capabilities.

So, it comes down to a couple of things:
1) How hard is it to detect the emissions testing?
2) Given a software module that detects emissions testing, how many people would have to collude to install it, and to let it impose strong emissions controls (and, even more significantly, relax or abandon emissions controls the rest of the time)?

It seems to me remotely possible that a few senior software people could do (2) – though all the incentives suggest they would inform their bosses in the fancy suits what was going on.

My experience with these things says that not only could a small team of maybe three people get away with something like this, I’m surprised that they didn’t play it off as a software bug or an artifact of using, say, code designed on a slightly different embedded platform not doing exactly what was expected when coupled with a new generation of hardware interfaces. Often things like this are written by a very small group and tested by that same very small group (which causes trouble when people leave or retire, as so much of what makes an embedded system smart ends up being locked up in a couple guys’ heads.)

If you weren’t specifically looking for something like this, you could get through a lot of testing cycles internally without seeing it. “Does this code actively defeat testing protocols” isn’t exactly something a VW engineer would be expecting to test for if they weren’t in on it.

= = = But it has always been my belief that, by and large, complex, dangerous conspiracies involving many people simply don’t happen. The more danger attaches to a criminal conspiracy—and here the danger seems in the worst case scenario actual dissolution of the company—the more the conspirators must be benefiting. Why would they do it otherwise? = = =

While the logic and the first-order human behavior analysis in this statement are impeccable I am personally always brought up hard by the counter-example of Richard Cheney’s OVP. 40 people, only two of whose names and titles we ever learned, did something in that office for eight years. At the end of which time they spent 10 weeks shredding records at three locations and then disappeared into the night. Who were they, what did they do, what affect did they have on the direction of US policy and government actions? We do not know: none of them have talked, nor have any of the hundreds who must have interacted with them. I’m not sure how that situation fits with the theory in the OP.

But certainly in the private sector it is possible to keep things quiet for long periods of time [1]. Particularly in heavy industry there is typically a culture of us-vs-them, inside-vs-outsider that operates unwritten from top to bottom of an organization without any central control or prompting. If you’re not part of the team, if you don’t work at the plant, if you don’t wear the blue jacket we don’t talk to you period.

fn1: over 100 years in one case I observed.

“I’m surprised that they didn’t play it off as a software bug or an artifact”

Yeah, the line I hear most often when I am prosecuting undergrad plagiarists is, “that was an early draft–it was never intended to be submitted for the assignment! My final draft didn’t have any of that stuff in it, and I just attached the wrong file!”

It would have been easy to say something similar here–“it was an early version of the software! It was in beta-testing and somehow went wild! etc. etc.”

I’d agree with Metatone @3 above, its probably a surprisingly small number of people. Software development for complex systems like engine management is surprisingly compartmentalized with teams of people working on particular modules within the broader suite of functionality. It could have just been a few people assigned to program ’emissions test’ software and they came up with a clever solution.

All of the sensors for data input and other hardware features were already present. The software was designed to interpret the data and basically “if x, then Y.”

I’d go on to assert that that it wouldn’t surprise me if very, very few people in senior management really had any idea aboutt this – by no means does this absolve them, but it strikes me more that the software folks came up with a solution that solved a problem. Management asked ‘is it fixed’ and they said yes – without necessarily providing all the detail. I admit to speculation here, but the hirearcy of these design systems is long on answers and short on details…its the same type of system that hides GMs inginition system problems for +8 years. Expect a lot more heads to roll before all is said and done.

Note also that the emission cheat really only mattered for beating US emission standards…it wasn’t needed for Europe since the engine already met those relatively more lax criteria without the need to cheat….perhaps those touting the clean-fuel chops of Germany and the EU would comment why this is so.

I would assume (hope?) that the chief architect of the engine program would go to sleep with his stoichiometry and trade off diagrams floating through his mind. When the firmware guys popped in and said they had a solution that violated the thermo and pchem models he asked no questions?

I tried the blockquote tag (probably a first for me) in the previous post and hoped that it would include my link. It didn’t. Here is the source of the blockquote.

It would only take some test code that someone later decided to enable in production. I don’t see how an engineering manager gets his or her team to implement regulation cheating software, and also gets the person responsible for emissions testing to go along. There are too many people involved and those groups probably don’t work together directly. Last minute, we can’t pass this test, let’s flip the switch, maybe.

To @9 I’d add 3) does VW do its own testing of the emissions system, which would have revealed that they met EU standards but no time US ones? That they don’t, I just don’t believe. That this tweak was done to meet a VW test by people with no idea of the legal implications, I can.

oldster 09.24.15 at 12:32 pm

“right–this is what makes me think a lot more people were in the know. They had to send out armies of people saying, “our engine delivers better performance and lower emissions than anyone else, and it doesn’t use the normal urea trick!” Of course at the level of the ad-copy people and the car-salesmen, they’re not going to give it a second thought–sure, our engineers are brilliant.”

As I understand it, VW was claiming that they had some engine tech which did what nobody else could. If it was for some, but not all, engines in in VW vehicles, then the other engine program managers would have wanted that Right Now, and upper management would have wanted to roll that across the company.

Keep in mind that the days of the engine swap or even the standalone engine management computer are gone. The entire package is an integrated unit controlled by integrated software. The headaches Boeing & Airbus have had with the 787 and A400 projects are larger in scale but provide public record of the difficulties such tightly integrated systems can have when one part of the software mis-functions. Heck, even the sound system designer needs to know how much noise she will get from the engine under condition x and is probably reading engine parameters to know that.

Hi Belle, this isn’t on topic, sorry, but I heard some forensic stuff about war today, and the fact that 85% of those who die in wars now are civilians, and it made me think of your earlier posts about NPCs, which are a while ago now – so I wrote about it in my own blog. So thank you.

@26 and the excerpt from the NYT story on “cheating software”.

Now my eyes have been opened: the real culprit is excessive government regulation!

Why, these corporations were *forced* to fight back against the job-killing, heavy handed, socialistical mandates from Washington that were killing jobs, socialistically.

I mean, sure: maybe they broke the law. But it was pretty much entrapment.

VW hit the trifecta.

evil: We are going to lie to our customers in the US about engine performance for
short term financial gain.
stupid: They thought they would get away with it in perpetuum.
Crazy: it was irrational to endanger the future of the company for advantage in one part of their market.

This seems to me like a symptom of contemporary corporate culture. Mortgage origination companies gave mortgages to the financial equivalent of dead people, and no one said anything. Investment banks bought those mortgages, which everyone knew were shitty, and no one said anything. They packaged them into shitty securities, and showed them to ratings agencies, who also knew they were turds, but pronounced them gold.

Many people at tech companies (Apple, say) must know the conditions under which their products are manufactured. Many people at energy companies know their pipelines are under maintained and leaking. Many people at mining companies know their tailings and sediment ponds, etc, are poisoning people. Many people knew the tires they made would fail at high speeds. Many people knew their peanut butter factory was infested with rats and the peanut butter itself comtaminated with salmonella. Most of these things, if not all of them, are crimes, yet no one says anything until the crimes are exposed, catastrophically.

As to why, it seems pretty clear. People at senior levels reap huge financial rewards by skirting the laws and regulations. Everyone else understands the personal cost of speaking up. That could come in the form of retaliation, but even that’s not necessary. If you speak up, and the company suffers, you probably lose your job anyway, along with a lot of other people. The crime drives the revenues and the profitability and therefore creates the jobs. No crime, less revenue, lower profits, fewer jobs. It is all financial incentive, though the incentive is more greed at the top and more fear at the bottom.

I was actually just looking at the U.S. conspiracy laws the other day. If it is a conspiracy the prosecutors do not have to determine who did what, they must just be able to charge the whole lot with conspiracy.

I haven’t trolled in like days. Weeks, even!

Conspiracy Art!

Learning From [Mark] Lombardi …best site I could find. Google image also.

Toscano on Lombardi, from Cartographies of the Absolute:

As Robert Hobbs notes, ‘Instead of simply solving crimes, Lombardi’s work often intensifies their mystery’.102 Lombardi’s rhizomes eschew any hierarchy of responsibility, and instead depict networks of sometimes only loose association, never coming together in a simple solution, some kind of cognitive or political epiphany. His ‘structures’ are painstakingly neat, their immediate visual effect is one of ordered complexity; but cognitively, and politically, they are nothing if not messy: ‘his brilliantly detailed drawings actually make things harder to understand, not easier. Looking at the endless miasma of names, institutions and locations, his charts are more about obfuscation than revelation … Lombardi’s drawings are like a pointillist work, best viewed from afar. From a distance you can see that a system has been revealed, but the closer you get to it the more invisible it becomes.’

Even if endowed with a decent knowledge of the BCCI scandal or the networks in which Roberto Calvi or George Bush (I and II) operated, it would take a considerable investment of intellectual labour to make any ‘sense’ of the drawings, to specify their structure and project a coherent narrative onto them. No doubt, this was something that Lombardi was aware of when elaborating his practice. In that respect, the drawings are intentionally opaque. In later works, for instance, the viewer isn’t even provided with a legend to explain the difference between a solid line, a dotted line, and the squiggles that intervene in some of the lines of connection. Lombardi’s obsessive passion for inquiry is writ large, but it is also evident that he judged that the results of this research could not be presented with the kind of direct communicational economy endorsed by Tufte. Thus, as much as Lombardi’s work is about the actual conspiracies revealed by his drawings, it is also about the very gap – the perhaps unbridgeable gap – between lay viewers and the activities of, and collusion between, the ‘overworld’ and the ‘underworld’.

( You don’t even understand that you can’t understand.)

ZM,
It is a very bad scandal. Probably the company will not continue in my opinion.

I can hardly imagine such a big company disappearing in a puff of smoke. And if somebody were to wish for it, as a punishment, I’d say that is exactly the wrong way around. One of the fundamental problems of accountability in our current economies is that when fraud happens, a company often pays a fine amounting to somewhere south of 10% of the profits the fraud made possible while the people who did it rarely face any personal consequences*. Surely destroying the livelihood of thousands of innocent workers isn’t an improvement on that approach; quite the opposite.

*) Few things are more exasperating that seeing CEOs claim that the success of their companies is all down to their personal performance, but that they had no idea what any of their underlings were doing when something went bad.

#24 – this. It’s a case of the dog that didn’t bark.

The problem of reducing NOx emissions to meet the 2007 EPA standards was so difficult that VW pulled its diesels of the US market. (NOx is a symbol meaning a combination of nitrogen oxide (NO) and nitrogen dioxide (NO2), which are pollutants that cause acid rain and smog and other bad things.)

Every other company that makes diesel passenger cars solves the problem with an exhaust treatment system that uses urea (ammonia plus carbon dioxide) to break the NOx down into harmless emissions – water and nitrogen gas.

VW claimed that its “clean diesel technology” didn’t produce the NOx in the first place, eliminating the next for the urea exhaust treatment process. This was cheaper and more fuel-efficient and required less maintenance. What VW was claiming was that it had invented an engine that burned diesel fuel differently from every other diesel engine in the world.

VW introduced its “clean diesel technology” in 2009. An effort to develop a non-NOx producing engine in two years would have been a huge project – hundreds of people in labs, offices, and testing facilities working round the clock.

Was there a project like this? If not, then every engineer and manager in the diesel engine department must have known that “clean diesel technology” was a con. And if so, then the engineers and managers must have known that it failed, so the con must have been even more obvious.

“Clean diesel” was marketed as a major breakthrough with huge implications for VW in terms of profits and image as a green company. There must have been many presentations at every management level, right up to the board of directors. What was the budget for it? And given the size that the budget should have been, wasn’t the board presented with progress reports? Wouldn’t the brilliant engineers and managers who brought this major breakthrough to fruition receive awards, bonuses and promotions?

But if none of this happened, it must have been obvious that something was wrong.

Alex SL,

an associate professor from uni wrote about the company in the paper and he did not think VW would continue, which is why I got the idea the company won’t continue

” There is a marketing word for the kind of trouble that VW is now in but I do not dare write it in this fine publication.”

There is no reason to keep the company, as we should be investing less in cars and more in public transport and active transport infrastructure, turning roads into linear parks and public spaces , and for necessary cars and trucks move to electric driverless vehicles – one expert said once we move to driverless vehicles we’ll only need about 30% of the current vehicles. If we improve public transport and active transport infrastructure then hopefully even less than than 30% will be needed, so streets will be much more pleasant being narrower and without exhaust fumes and too much traffic.

Maybe someone who understands more about diesel engines than I do can answer a couple of sets of technical questions.

1) Diesel exhaust fluid, a solution of urea in water, has been widely available for some time. It can greatly reduce emissions of nitrogen oxides: it is sprayed into the exhaust gases, which then pass over a catalyst in which urea is converted to ammonia, which then reacts with the nitrogen oxides to form nitrogen and water. Presumably (yes I realise this may be a weak point, but is there any other interpretation?) the Volkswagen software ensured that a great deal of exhaust fluid was being sprayed in when the engine was being tested, and much less at other times. Question: what is the advantage, in terms of engine performance, of reducing the spray rate of exhaust fluid? I suppose there must be some advantage, else otherwise why would VW do it?

2) VW may have fudged the particulate emission rate as well – if they didn’t, the rate must be acceptably low even when unfudged, and if so is there any justification for worrying about particulate emissions? But particulates can’t be removed by urea treatment, you need a special filter. How would it be possible for VW to fudge that?

Cheney to the world: Your silly international conventions change nothing. Schäuble to the Greek government: Your silly elections change nothing. Winterkorn to the EPA: Your silly regulations change nothing.

We may not like the implications, but it’s not as though we haven’t been warned.

No Lederhosen in Wolfsburg. Btw, I’m surprised nobody mentioned this yet, VW is 20% state owned and the government (of Niedersachsen) is represented in the governing board.

“In other words, in the two cases of actual rule-breaking you’ve identified, there have been actual negative consequences for the rule-breaker.”

What were the consequences for Cheney again? I missed that part.

@Willimam Berry. It might be more accurate to say that we don’t actually know that it’s a conspiracy because we don’t know how many people were involved and at what level. I know that in some contexts a conspiracy can be alleged between just two people (in criminal cases for example) but I think the point in the OP was rather different. We don’t know yet (though it seems unlikely) if this was large and organised enough to make us change our opinions about how far the kind of conspiracies featured in books and films actually exist in real life. Until there’s more evidence, I remain a firm believer in the cock-up theory.

Doctor Memory may have the right of it, especially if combined with the notion described in #7 by Exon (and earlier d-squared). It isn’t the same type of conspiracy if you believe that ‘everyone’ is cheating. Then you don’t feel you are really hiding anything. So long as whatever engineering unit you are working with all buys into the idea it isn’t really a conspiracy so much as a systemic corruption.

I definitely believe in systemic corruptions! (See banks, and banks, and the big financial companies).

We will hopefully know more as the investigations ramp up, but if this ends up being a large group of engineers and managers who really knew they were actively cheating, we may very well have to revise our ideas about what conspiracies can last.

#50 – good question. I have no idea. Also, where are the patents? There should have been patents for the “new technology” and patents are public record. Were there any?

Jake@62: again, you may well be right. Sources, please.

I will also concur with the various people above: the realities of modern software development not only makes it possible to pull off something like this with a small team of people inside a larger organization, it practically encourages it.

The ideal of large-scale software engineering is a system of loosely coupled black boxes: your team never looks at the code another team produces, you merely do integration tests of your code against the APIs they provide. Does their API produce syntactically correct output when your component provides input? Yay, your regression test passed and you can get on with your life. It’s hard enough having a deep understanding of your own segment of the system; if you had to be able to reason a priori about the structure of the entire thing (millions of lines of code, in turn running on top of millions of lines of other code, on top of a turtle, and turtles all the way down) you’d quit and become a yak farmer instead. Many of us do that anyway. It’s a weird business.

“But it has always been my belief that, by and large, complex, dangerous conspiracies involving many people simply don’t happen.”

I skipped over the comments to post this, I’ll go back and read them later. I’m sure someone earlier has made the same point I’m going to make.

But Jesus Christ, this is ignorant. Its the sort of statement that makes me take everything else the blogger writes less seriously.

https://en.wikipedia.org/wiki/Manhattan_Project

https://en.wikipedia.org/wiki/Bletchley_Park

At some point in these threads someone has to point out that the whole idea of “conspiracy theory” was created as part of a CIA directed propaganda campaign (the CIA is not supposed to conduct propaganda campaigns against the American public), when they started to get worried that people were asking too many questions about how involved LBJ and the CIA itself was with the the first Kennedy Assassination. Here is another link:

http://utpress.utexas.edu/index.php/books/dehcon

The book reprints the now declassified original memo behind this, which is well argued and makes for interesting reading.

However, before then, it just wasn’t considered to be particularly strange or controversial to suggest that powerful organizations might be acting from motives, or doing things outright, that they didn’t really want to become public knowledge.

Sorry for the multiple posts at once. I forgot to point out that right after the news about Volkswagen came out, it was revealed that scientists at Exxon discovered that carbon emissions were causing global temperatures to rise in the 1970s. The company then started funding a campaign to convince people that there was no scientific proof that carbon emissions caused temperatures to rise:

http://insideclimatenews.org/search_documents?field_related_project=41124

I actually think this is bigger news than what Volkswagen was doing, though the cases are similar.

Layman @ 37
Thanks for saving me the time of writing a less-succinct version of that response.

I’m struck, reading this post and others on the topic, how we finally seem to be waking up to the fact that when case after case makes the news, the isolated incidents start to accumulate and cohere to tell an increasingly hard-to-deny narrative.
Despite how badly most of us seem to want to believe each revelation exposes the independent actions of a bad actor or two, rather than a problem with the system (i.e. the system encourages and rewards decisions and behaviors that we claim we want to discourage), it’s harder and harder to take this rationale seriously. I think there are a lot of smart people out there who realize that even the worst-case scenario for cheating is not much worse than the outcome they can expect from 100% compliance with regulations. Especially when losses are paid by the company, and NOBODY EVER GOES TO JAIL.
It doesn’t help, of course, that most major media outlets tend to report each case in isolation, and continue to portray managed settlements (for instance) as providing a reasonable degree of security that illegal behaviors won’t continue. Despite plenty of evidence that they do no such thing. How many times has Pfizer promised to always do everything right, no matter what, if we just give it another chance?
Personally, I’d like to provide everyone a basic income and remove protections for employees who are just following orders to keep the kids fed and clothed. Including the implicit one of knowing your coworkers all have food and clothes to afford, as well, and can also generally be counted on not to raise a fuss. In my admittedly-utopian daydream world, the result might be companies behaving ethically, in order that they can attract and retain talented people who expect to be able to not have to quit due to pressure to do illegal and/or unethical things. Unfortunately, I don’t have the means to provide everyone a basic income.)

@Dr memory, it’s not the software side that makes the ‘small group of coders” theory unlikely. It’s the engine side, where hundreds to thousands of engineers struggle to increase performance, and know perfectly well that the engine can’t do what it is claimed to do. And that includes former engineers who became high level managers.

They might not know exactly what trick is used, but they know tricks are used. The surprise (if there is any) is not about the cheating, it’s that the cheating is so straightforward. If you asked, they would have guessed the cheating worked more subtle. React to the small band of rpms used in the test, acceleration profiles as used in the test, etc.

The competitors are denying cheating, and I bet this is what they mean. That their tricks are not as obvious cheating. And they know because they caused the department in charge of test-cheating, and asked what tricks they used. If you ask someone in engine development, whose in charge of fudging the tests, they can provide name and phone number.

Embedded developer here. I see a lot of uninformed speculation on this page about corporate conspiracy and culture of complicity. Metatone @ 3 and Terra @ 9 have the right answer. Controlling an engine requires complex code that seeks different goals at different times (performance, efficiency and emissions). For the developer, the problem is to guess what state is best to control for depending on the input of the driver, and they probably have a suite of tests against which they evaluate different control algorithms. To some extent, any test is artificial and it is not hard to see why a developer might view it as an arbitrary choice to force it to optimize emissions if, say, the engine is connected to a test harness.

So much of this comes down to penny-pinching. It was always a horror in the car manufacturing business (think of the decades of Chevrolets–trucks, especially–with brake, tail and turn signal feedback because Chevrolet redesigned the electricals to save fifty cents a unit, or Sir William Lyons, when the first XK-E was being costed out, choosing standard brake linings over premium to save 25p per unit on a 150mph luxury car), and the power generation business makes car manufacturers look like pikers, and the food industry makes the power generation business look like amateurs.

My guess is that this may be a combination of not meeting advertised mileage figures without the kluge, or perhaps having to pay more for full-time emissions equipment, but I’m reasonably sure it comes down to lost Euros per unit, whether in component costs or losses through marketing.

As for how to do it–dead simple. Not all emissions testing is done just by the EPA on rolling roads, and not all stations in the states use dynamometer testing (very expensive equipment), so I don’t think the front wheels turning/rear wheels not turning would be foolproof. The simplest way would be to detect that the OBD-II port was in use, which is standard practice in all air quality stations in the U.S. Plugging in the test connector could just swap to a different parameter map. Easy stuff to do.

Diesels, wrt new emissions standards for nitrogen oxides and particulates, have been a real headache for engineers. If those nitrogen compounds are formed at temps above 1200 deg. F., and the fuel doesn’t ignite efficiently except at combustion chamber temps above that point, then the problem has to be solved at the exhaust gas end of the process, and that means expensive particulate traps, reduction catalysis, and exhaust gas recirculation schemes and/or NOx adsorbers and urea injection (to reduce NOx to ammonia). It may also be that with the emission systems running full-time, they would not meet the EPA’s standards for component lifetimes and couldn’t be certified in the U.S. That would be a driving reason for creating the scam.

Uh, i dont think its even illegal in the EU. I think i did even read an article suggesting this is widespread month ago. No matter either way the spread betwen real life and standard test performance is increasing a lot accross all manufacturers. Cant help but wunder if this would have been far less of scandal in perception if it had been done by GM.

The poof disappearance of Anderson, Enron, and Lehmann was a bit easier than the full disappearance of Volkswagen would be. It’s relatively easy to repurpose a bunch of office buildings and a bunch of white collar bankers/consultants than it is to repurpose a lot of factories and specialized engineers.

I think Zamfir in 72. nails it. It is well-known for example that automakers everywhere “game” mileage standards. Often in perfectly legal ways – e.g. by allowing the driver to choose between various programs including one called something like “green” or “eco” or whatever that they expect is almost never going to be used but that they can use in their testing to establish some amazing mileage for their product. So I’m thinking the reason nobody at VW blew the whistle on this is because nobody thought there was anything to blow the whistle on.

Then again I learned from the paper of record that this is somehow about German national character. Or something.

The NYTimes says such a scandal was all but inevitable:

“The governance of Volkswagen was a breeding ground for scandal,” said Charles M. Elson, professor of finance and director of the John L. Weinberg Center for Corporate Governance at the University of Delaware. “It was an accident waiting to happen.”

http://www.nytimes.com/2015/09/25/business/international/problems-at-volkswagen-start-in-the-boardroom.html?hp&action=click&pgtype=Homepage&module=second-column-region&region=top-news&WT.nav=top-news&_r=0

The NYT says Volkswagen’s culture is less than optimal, for example

One measure of [former autocratic Chairman] Mr. Piëch’s influence: In 2012, shareholders elected his fourth wife, Ursula, a former kindergarten teacher who had been the Piëch family’s governess before her marriage to Ferdinand, to the company’s supervisory board.

The case study for business school students around the world is being composed as I write.

#61 spot on – no patent, then the ‘invention’ is a scam.

As far as the software process controls, I suspect that in a regulated business, each change put into the master source will be reviewed by other engineers, approved by QA, attached to a documented change request, and possibly subject to random review by a 3rd party.

That’s what we have at my office (money transfer stuff). Hard to believe they have a bunch of cowboy coders at VW. And if they do, then their shoddy practices constitutes willful negligence.

As an embedded software engineer myself, I can imagine a scenario in which it is possible to create separate software components, whose combined effects might well be what is occuring. The individual programmers would not realise see the final series of control results. It would be difficult, but not impossible.

I think the point is well made in that more than one person who would have had to know or instructed the intent. That by definition is a conspiracy.

You only need a small number of good-intentioned engineers/managers for this to occur. The tech industry is rife with such misguided indiscretions (leaving a jeep vulnerable to wireless hacking through its entertainment system, for instance, or Lenovo shipping spyware in their laptops). There is no sinister plot. Ttwo levels away in the corporate hierarchy all they hear is “our smart engineers solved the problem”.

Here is the interesting statement by BMW:

BMW says:
The BMW Group does not manipulate or rig any emissions tests. We observe the legal requirements in each country and fulfill all local testing requirements.

In other words, our exhaust treatment systems are active whether rolling on the test bench or driving on the road.
/BMW

At the current level of scrutiny, I assume that the statement is literally true. It also leaves room for plenty of cheats and manipulations, as long as it is not the direct detection of a roller bench. Some of their cars have been shown to exceed test values by many times in road tests, at least ballpark similar to VW.

We have to get to the bottom of it simply for that reason–we have to know how stuff like this happens.

You are assuming they must have known but wouldn’t it be amazing if they’d simply created a culture where everyone worked in harmony to commit a massive fraud like this without talking about it directly?

Also, it could have been a freak accident (unlikely but possible).

But it’s something that would reveal a lot–because it is absolutely challenging to some ways of thinking about how bad things get done collectively. And we need to know how bad things get done collectively because the worst things are generally done collectively.

Another interesting point is the upcoming Euro6c norm, which updates Euro6 to include road tests. I put a europarliament document below.

As you can see there, everyone involved knows and takes for granted that road tests are a severe addition to the norm – which means they also realize that bench tests paint a misleadingly positive picture.

The introduction date for road tests is in 2017, explicitly to allow manufacturers to replace their dirty models on a normal, non-accelerated timeframe. See page 8.

It reads to me as if regulators were implicitly OK with gaming the bench test, as a form of relieve from an otherwise too heavy norm. ‘Too heavy’ as in, they could not find politic support for that norm if it had included road tests early on.
They might not have realized how blatantly the test was gamed, but they knew tricks were played.

Matthew Exon at 7 is right that this fits D^2 scandal model. A course of action that insiders started to find normal, and suddenly they are forced to explain it to outsiders. At which point the card house falls down.

http://www.europarl.europa.eu/meetdocs/2014_2019/documents/envi/dv/nec2_land_/nec2_land_en.pdf

@Warren, BMW might not (yet?) have been found cheating as blatantly as VW. But some of their vehicles with smaller engines have been road-tested, with similar deviations from the bench test. They have gamed the system almost as hard, but perhaps with less explicit instructions in the software. Same for PSA, or Volvo.

That said, urea injection does seem to work better in real life, and everyone was switching to that anyway. Question is, does it work good enough for renewed levels of scrutiny?

@Bill Benzon

So, did Gov. Christie himself order those lane closures as political retribution or did he just create an atmosphere in his office where people did such things on their own initiative?

Working towards the Führer?

ZM,

I wouldn’t mind switching to a pretty much car-free economy, personally. But unless I misunderstand you you are talking about wiping out, from one day to the next, the economic basis of one of the largest federal states of Germany … in an unprecedented move given that all previous comparable cases ended with a slap on the wrist. Again, I don’t like all the cars on the road either, but how is that fair? And how is it remotely realistic from a political perspective?

The solution to a company doing something wrong can’t be to give it a slap on the wrist while those responsible walk free. But it can’t be to destroy the company and thus the livelihoods of thousands of innocent employees, of tens of thousands of their family members, of hundreds of thousands of their suppliers, retailers, etc., in short the economy of an entire region, either. The only solution that is fair and works is if you identify the people who were individually responsible and throw the book at them.

Alex SL,

My opinion was formed by the article I mentioned. I found another article that quoted an expert saying that with driverless cars we would only need 10% of the current amount of cars, so we don’t need so many car companies any more.

Germany has a very good economy, so no doubt they will be able to manage the demise of VW.

I think it is worse when companies are allowed to continue when they have done something wrong, like the banks that were too big to fail.

I suppose if you identified all the people who did the wrong thing and prosecute them that would sort of work — except if this is a conspiracy it is probably going to be difficult to identify who did what as the company would not have kept proper records of this wrong doing.

I hope President Trump arrests Winterkorn in Anguilla just after Angela Merkel’s picked squad of Bundespolizei has arrested Dick Cheney in Wyoming and dragged him to the Hague.

But in fact, Winterkorn will not suffer any more than Cheney will. News reports say that he is in line to leave VW with a severance package worth about $67 million US. And if a lot of workers on the assembly line lose their jobs when VW sales plummet, what does that matter to him? It’s simply capitalism at work.

Oldster #113: The relevant German unit of small hard guys in balaclavas is the GSG-9. They might be able to pull it off. Skorzeny did.

I’d guess that the competitions engine management systems (petrol and diesel) all have logic in them which detects test-like conditions and optimizes emissions for those conditions, and detect everyday driving and optimize power or economy.

VW’s crime was explicitly detecting the test, rather than always going to emissions mode in all test-like conditions.

“PS: it’s uncanny to see another Layman.”

I contain multitudes.

“This is a win?”

Surely they thought they wouldn’t get caught, in which case yes, selling 11 million cars you would not otherwise have been able to sell is a big win. Bonuses all ’round!

I’m with NuckJ @12 (possibly others I don’t have the time to read more).
The engineers could be teasked with the strictly technical tasks, generate software with the following easily selected capabilities -presumably most are only for testing/development. A few easy to flip software switches are then available to management, and the lower level people who wrote the actual software need not know what settings they actually selected.

The NYT business section today has an interesting article on the VW corporate culture that allowed this to happen:

“That such a thing could happen at Volkswagen, Germany’s largest company and the world’s largest automaker by sales — 202.5 billion euros last year — has mystified consumers and regulators around the world. But given Volkswagen’s history, culture and corporate structure, the real mystery may be why something like this didn’t happen sooner.”

http://www.nytimes.com/2015/09/25/business/international/problems-at-volkswagen-start-in-the-boardroom.html?ref=business

I don’t think the excuse that it was an “internal test mode” washes. That would only need an on/off switch that could be flipped from the external control computer. What we know indicates is that the software had test-detection built in, which requires that you code in parameters that would imply the car is undergoing a regulatory test.

ZM,

So what do you do if all major banks have done something wrong? Destroy the world economy and start again from the neolithic? And extending the logic, do you destroy an entire country whenever its government has committed a crime?

I am afraid that your vision is just the flip side of certain US judges’ view that corporations are persons. They aren’t, and consequently they shouldn’t be treated like persons. Instead they are made up of persons, and only some of those persons have responsibility for any particular thing that happens.

This is due to the method used to check exhaust emissions.

When Illinois started testing years ago a measurement device on a probe was physically inserted into the cars exhaust pipe.

The last few times our Subaru was tested a literal black box was place on top of the car’s hood.

It is hard to understand why emissions testing by the EPA is no longer a physical process. Checking thousands of cars a month by the State of Illinois is one thing.

Exhaustively checking a dozen or so test cars at the actual point of emissions
should not be that daunting a task. Of course it is being done by a government agency.
Maybe that explains why stuff like this is allowed to happen.

@ The Wolfman, #125
VW has apparently 404’d the Clean Diesel page you tried to link. Their homepage has nothing obviously about diesel or fuel efficiency – but also has no other references to this debacle, such as a statement of apology on the home page.

It does feature that the Golf family of vehicles is the 2015 Motor Trend Car Of The Year. This of course includes the now disastrous Golf GDI diesel. I wonder if they will have to forfeit this and other honors, and stop using them in advertisements?

Warren: Yeah, that was the point. That the VW clean diesel page can’t be found. I thought it was funny. I’m grateful to The Temporary Name (is this a temporary pseudonym?) for having tracked down the original.

#126-
“Remember in that world, the sole purpose of the company is to increase shareholder value for the investors. ”

Not so. VW is managed under the German system known as Mitbestimmung (co-determination). This means that VW’s unions have equal representation on the powerful “supervisory board” that oversees the “board of managers.” In addition, the German state of Lower Saxony owns 20% of VW stock and has corresponding representation on the board. The remaining seats are appointed, effectively, by the Piech family, which owns a controlling block of shares through a foundation.

The result is that VW does not try to maximize value for shareholders. It tries to maximize jobs and sales. As the NYT article I linked to above points out, VW has 1.75 times the number of employees that Toyota has, but produces only about 10% more cars.

Maximizing jobs is an obvious union goal. Maximizing sales has the benefit of increasing the number of jobs at suppliers, whose employees are also union members, and many of which are also in Lower Saxony – so that’s something the union and government board members care about. And the Piech family doesn’t need to maximize wealth, but its leading members get a lot of prestige in Germany by making a German company the largest car company in the world. So the goal of maximizing sales is one that all parties to the co-determination governance could share.

The goal of maximizing sales contributed to this scandal because VW made the decision that it had to make big inroads into the North American market at any cost, because otherwise it could not grow to be the largest car company in the world. That was the goal, not profit maximization.

Looks like VW hardly made any money from cheating that way. Just a little more consumer comfort due to shorter maintenance intervalls or some money saved on a larger tanks for that additive. Maybe they severly misjuged how large the fine would be if they get caught, maybe based on the prior similar cases with Trucks?

Whistleblowing is not exactly encouraged by German employment law, but it still seems all rather irrational.

I think a certain thing needs to be emphasized here:

The actual issue with the cheating had to do with just how much cheating it was. Regulators can be pressured on a little shave here and there, but VW’s differential was both qualitatively (as in no magical pollution control w/o urea solution engine) and quantitative (as in, if there are enough of those, in realistic numbers, in Los Angeles or Salt Lake City, then they are very real air hazards that even the rich suffer).

The idea that cheating is bad, just so, is simply everyone rubbing salt in Germany’s public face for past slights, real and/or imagined.

Stephen,

“Alex SL: it seems that ZM is in favour of exiling a large part of the population of London for the eco-crime of not producing enough food to support themselves, but importing food instead. That being so, I don’t suppose the devastation of the economy of Niedersachsen would bother her too much.”

I have already had this argument with you in the past I believe.

I have never said that I am in favour of exiling a large part of the population of London as they consume too many resources.

What I am in favour of is the population of London decreasing their consumption of resources and producing more resources in London — such as by increasing allotment gardens which provided a lot of fruit and vegetables in wartime and the immediate post-war period. They can also have rain tanks to catch water, and have renewable energy technology to produce energy, and develop a circular economy to recycle all the metals so they don’t use metals from other countries too.

@142
Oh for heavens sake, why don’t you engage with what ZM’s saying? Oooh, 12000 miles away in Australia, must be irrelevant.

CT has good authors but lord some of the commenters are boring.

ZM, good luck with convincing Londoners to take their living standards back to what they were in the war and its aftermath, 70 years ago. I don’t think they’ll buy it, but I might have underestimated your powers of persuasion.

Thanks for the explanation, that way it is indead a new quality of cheating that goes beyond the more cynical suggestions ive come accross before. Just hope this is limited to vw. Theres been an article that suggests Bosch has “warned“ vw about using that software in a way that rather sounds like a suggestion to use supplied software to cheat with legal coverage and bosch supplies pretty much every diesel manufacturer.

@Val and ZM: You guys rock.

Don’t let the neo-lib old-boys’ club of (some of) the commenters here get you down.*

*Not that there is any danger of that, mind you.

Stephen,

I have no difficulty in believing that the population of London is unable to grow sufficient food to feed themselves within the city boundaries. I never suggested they could do this. What I said was that they should grow a greater proportion of their food supplies in London with allotment gardens and having fruit trees and herbs on the grass beside the footpaths and so on — once you get electric cars the pollution even on heavy traffic roads won’t spoil the fruit.

” get you to question your underlying belief that each area should support itself from its own resources without relying on resources from outside.”

I do not assert this either. This is as it would be unfair because a country like Australia with a lot of land and relatively small population has a lot of resources, but I think somewhere like Singapore needs trade due to being overpopulated from colonialism and 20th C mobility like London.

You have to be historic and take into consideration the facts.

I hope you enjoyed your orange and coffee, I guess its a pleasantly warm Autumn day over there if you’re wearing a cotton outfit.

We have a different problem with imports of oranges — we can grow oranges in Australia but the imports are often cheaper so you see orange orchards bulldozing their trees which is a great waste in my view. Although where I live we have stonefruit orchards not orange orchards — but the federal government a few years ago decided to let New Zealand apples in despite the risk of fireblight, and it is not good for the orchardists at all and if the orchards are subdivided for housing it will be a misuse of good agricultural land.

I am sure there are plenty of nice things you can grow in England, like elderberries and damson plums and currants, and linen is a very elegant fabric if you either enjoy ironing or don’t mind crinkles.

Australia does not manufacture much electrical goods anymore, as that sort of manufacturing got offshored some time ago. We do have a lot of metals that we mine and export that might be used in electrical goods.

Anyway — my point was not that Londoners should be forbidden from ever trading with any other country again due to past wrongs and over-consumption.

My point was that due to the nature of the world’s economy being interconnected and globalised we should accept re-distribution between nations to reduce inequality, in the same way as we accept re-distribution within nations to reduce inequality.

This being the case, people in advanced economies should reduce their consumption to responsible levels.

I do not need an electoral mandate from Londoners since they already voted in this Conservative government with David Cameron and a few days ago UK’s Conservative government signed up to fully implement the post-2015 development agenda by 2030.

One thing the government — who you already elected so they already have an electoral mandate — signed up the UK to do was move to sustainable consumption and production. So everyone in England has fortunately already agreed to reduce their consumption to sustainable levels in the next 15 years — happily it does not rely on my powers of persuasion from here in Australia.

(I didn’t see this post until today. Sorry!)

126: I don’t think the excuse that it was an “internal test mode” washes. That would only need an on/off switch that could be flipped from the external control computer. What we know indicates is that the software had test-detection built in, which requires that you code in parameters that would imply the car is undergoing a regulatory test.

Nope. I figured out what they did.

The trick is that all the sensors in a car are basically very simple. The oxygen sensor, for example, simply detects the amount of oxygen in the exhaust pipe, just downdraft from the exhaust headers. It does not know nor care about any other chemical component in the exhaust. And there aren’t that many sensors controlling any given engine. In platonic ideal form, a modern car has: a crank sensor, cam sensor, oxygen sensor, voltage sensor, mass air flow sensor, and a few more perhaps depending on the car. Those are all the sensors needed to control the engine proper, which effectively controls the exhaust output.

The software to track those sensors and make the data available to the master program were all written long ago and only slightly updated from time to time. Otherwise, they’d be risking the proliferation of bugs which would cause warranty (and safety) problems.

The master engine control, in turn, has (correctly!) lots of modes to adapt to different driving conditions, not to mention a startup mode which is heavily polluting. Adding an extra mode would be easy peasy and would simply appear to be a different driving mode.

Emissions testing is done in place, with the tires on rollers, so all they’d need to do is detect the situation that exists when the tires are going round and round but the car itself isn’t moving. For that you’d need an accelerometer. Nominally, adding an accelerometer would be expensive and suspicious, but if you already had one, you’d be all set. Just sniff for the full stop condition from the accelerometer but acceleration on the engine and say, hey, we need to be in mode X.

Airbag systems have built-in accelerometers. Airbag systems have to have them, or otherwise there would be no way to detect an accident occurring and trigger the airbag explosion. If the airbags are wired to communicate their actual positional data to the main computer for other reasons (detecting an ongoing accident and attempting to maximize safety by shutting the car down), then the master system already has the positional data.

The master system just looks for engine acceleration with no actual movement of the entire car and says, boom, let’s switch into Mode X, clean mode. (That particular mode would actually be helpful in certain weird circumstances: the front bumper was up against a brick wall, say, and someone stomped on the accelerator. Or the parking and standard brakes were all fully engage. In that instance, the computer would just radically reduce horsepower output, minimizing damage both to the car and the wall. (Or the brakes.) Aand tah dah!, distinctly reducing the amount of pollutants.)

To add the mode in its entirety would probably take 500 lines of assembler, tops, and far fewer lines of high-level code. To tweak an existing safety mode would take far less. Then the programmers would have just needed the running parameters for that mode to minimize pollution.

Nothing much to it, and no switch needed. As soon as the accelerometers sniff significant forward movement a mode change is engaged, so actual driving performance is never impacted.

No big conspiracy needed – this is just a couple of programmers, the manager in charge of the coders, and the diesel car VP.

max
[‘They’re probably all doing it, because they’ve already done almost all the work, and not being able to deliver the numbers needed to roll out new models would cost some people their jobs.’]

ZM @151
People behaving as if they have a responsibility to other humans and/or the ecosystem that supports life might seem nice enough in theory, but do you realize this might mean fewer and/or more expensive avocados for many parts of the world?