5 ms·

"This was a couple of software engineers who put this in for whatever reason" This was a system. A complex one. No way it was done by a couple of guys at 2AM,

by kabdib 11y ago

"This was a couple of software engineers who put this in for whatever reason" This was a system. A complex one. No way it was done by a couple of guys at 2AM, or whatever. This has to have been inter-group collusion, and management is in CYA mode and shoving individuals under the bus. VW has gone from being a cravenly stupid company that is going to be fined billions of dollars, to a place where I suggest it is in every engineer's best interest to resign, and the earliest opportunity they can manage.

maxerickson 11y ago
Say the detection for the emissions testing is something like a flag in the software that says the car is on a dyno, or some other factors that the ECU has legitimate reasons to keep track of. In that case, the defeat system is just a quick if else predicated on those state variables and swapping between a set of engine parameters that is emissions test safe and another set that provides better performance. I don't mean to insist that this must be the case, I'm just trying to illustrate that the cheat could happen at a level where much of the complexity has already been abstracted away.
- kabdib 11y ago
  Okay, imagine you're working on the hardware of the engine. Pistons, bolts, lubrication. One Friday afternoon you leave work depressed because the emissions suck, they've sucked for months, and they won't get better without a major redesign. On Monday morning, the emissions tests are suddenly passing. "Ha, just one of those things," you say, "I thought we had emissions problems, but I guess we don't after all. Let's ship the engine and have a party!" There is no way that this happened.
  - maxerickson 11y ago
    I'm not arguing that it was a few people working in secret, just that it would be possible for the changes to be done at a high level (in the engine system), without pulling in all the people that were involved in developing the vehicle (or even the powertrain).

by kabdib 11y ago

maxerickson 11y ago

Say the detection for the emissions testing is something like a flag in the software that says the car is on a dyno, or some other factors that the ECU has legitimate reasons to keep track of. In that case, the defeat system is just a quick if else predicated on those state variables and swapping between a set of engine parameters that is emissions test safe and another set that provides better performance. I don't mean to insist that this must be the case, I'm just trying to illustrate that the cheat could happen at a level where much of the complexity has already been abstracted away.

kabdib 11y ago
Okay, imagine you're working on the hardware of the engine. Pistons, bolts, lubrication. One Friday afternoon you leave work depressed because the emissions suck, they've sucked for months, and they won't get better without a major redesign. On Monday morning, the emissions tests are suddenly passing. "Ha, just one of those things," you say, "I thought we had emissions problems, but I guess we don't after all. Let's ship the engine and have a party!" There is no way that this happened.
- maxerickson 11y ago
  I'm not arguing that it was a few people working in secret, just that it would be possible for the changes to be done at a high level (in the engine system), without pulling in all the people that were involved in developing the vehicle (or even the powertrain).