Slashdot Mirror
EPA Finds More VW Cheating Software, Including In a Porsche (nytimes.com)
schwit1 writes with this news from the Times that Volkswagen's emissions scandal just expanded to include more expensive vehicles with larger diesel engines, including Porsche, and Audi sport-utility vehicles. According to the article: "The Environmental Protection Agency said on Monday that it had discovered emissions-cheating software on more Volkswagen and Audi cars than previously disclosed and, for the first time, also found the illegal software in one of the carmaker's high-end Porsche models. The German carmaker disputed the claims, however, saying it had not installed defeat software on the models in question that would 'alter emissions characteristics in a forbidden manner.' The company pledged in a short statement that it would cooperate with the E.P.A. 'to clarify the matter in its entirety.' The latest findings by environmental regulators put significant new pressure on Volkswagen and its new chief executive, Matthias Müller, who was previously the head of Volkswagen's Porsche division. E.P.A. officials indicated the latest violations were found during testing performed by federal regulators and their counterparts in California and Canada. The implication is that Volkswagen did not provide the information."
However VW denies the vehicles have software designed to cheat tests.
Instead the company says that cars with the 3.0 litre diesel V6 engines "had a software function which had not been adequately described in the application process".
If VW wants to get past this scandal, they really need to adopt a full-transparency, maximum mea culpa stance right now, and this kind of statement does not appear to be helping. If there's a software function that seems to the EPA to be cheating on emissions tests, well, if it walks like a duck and quacks like a duck...
Oh no... it's the future.
Plus, the DMCA can be swung at any uppity asshole who dares poke at your software; which makes it a safer place to hide regulatory...irregularities.
I don't understand why they are even trying the 'oh, just a few lowly rogue engineers' excuse.
If it were true, that might absolve them on the defeat-device issue; but it'd be pretty horrifying on the 'software validation processes for a life-critical component going into millions of vehicles' front. I sure as hell wouldn't want to be in, near, or on the same road as, a car whose ECU firmware was built under such a sloppy process that some engineering peon could secretly slip easter eggs into it.
Aside from that, there's also the minor issue that the affected engine doesn't exist in isolation: VW has been making diesel engines for decades, has multiple product lines for various purposes, has ongoing R&D efforts, and so on. Are we supposed to believe that nobody raised an eyebrow when the revision N+1 engines suddenly started turning in far better NOx numbers than the revision N ones; and none of the mechanical engineers had touched anything and the software guys would only look away and mumble something about 'optimized the firmware'? Are we supposed to believe that the R&D people working on refining existing designs or creating new ones aren't wondering why their advanced prototypes are getting worse NOx numbers than years-old production models?
If you just have a single product; no predecessors, no successors, maybe you can rig the demo without alerting anyone not involved in the rigging; but if your rigged product is an adaptation of a prior version? Then you have to explain an impressive discontinuity in performance between the current design and the prior model; and somehow explain away why the research guys can't do as well as you can(or find research guys so dense that they only use EPA tests and don't wonder why none of their tweaks appear to change the results). That is a great deal harder and less plausible.
I doubt that there is only one guy.
Things like this usually work like this:
During a routine meeting:
Upper Management: Oh, there is a new requirement for the US market. The test procedures have been updated. To be able to sell our cars at the US market, we need to make sure that during a test that will cover this and this and this scenarios A B and C, we must not emit more that amount x of stuff. I don't need to remind you how important that market is for us.
Middle managment: Ok, I'll draft a spec. "When A, B and C, emissions must be below x." I'll pass it to the engineers.
Later at a brainstorming meeting:
Engineer 1: Guys, we need to reach x in situations A, B and C. So, how could we reduce emmissions?
Engineer 2: We could load that set of motor parameters into the engine controller
Engineer 3: But wouldn't that cost us performance/milage/acceleration and/or increase engine noise, driving comfort,
Engineer 2: Yes, but this spec says we only need to do it when A, B and C, so this wouldn't cost us anything in normal operation mode.
Engineer 1: Ok, make it so.
bickerdyke