Rogue Software Changes

  • Comments posted to this topic are about the item Rogue Software Changes

  • This argument suggests that the management structure of VW was completely incompetent rather than being simply greedy, immoral bastards, who were solely concentrating on profit (and bonus) maximisation. Simply on the grounds that the entire management structure benefitted handsomely (and least in the short term) and because managers tend to be clever people, I am disinclined to believe this argument in this case.

    I find it as plausible as the accusation that 19 people could learn to fly commercial airliners in 3 weeks using light aircraft and then be able to fly said airliners into skyscrapers.

    However, truth is often stranger than fiction, so I may be surprised yet. I wouldn't have thought it possible that a commercial aircraft could simply disappear and yet it happened not too recently. I must watch Thunderbird again.

  • If I want to ensure "quality" in a project, I implement a process to ensure this quality. This is not something that will happen automatically, magically or similar.

    Software Projects by their very nature need to be pragmatic to be in time / on budget (or something resembling either) - corners will be cut. You also have people involved at various skill levels etc, and often these people will not be involved in running / maintaining the software in the long run. This is particularly true if you outsource and / or offshore (parts of) your project. It is naive to expect some developer to be compliant with your rules and regulations if these mean more time / effort with no reward. If you can get by not being compliant and probably at the same time save effort, than it would be naive to expect people not to cut corners.

    So the basic and old paradigm still reigns: You need to specify each and every requirement - functional and non functional. You need to define how to measure this, and how to verify the requirements are met all the way to implementation and cecks of these verifications. If you choose to not do this at a management level, you cannot blame some developer.

    And really we're talking automotive industry here. It's not financial services, but the basic quality assurance should be even more deeply ingrained. And if you have a compliance organisation, which is probably involved in all sorts of detailed design and production activities to keep the company out of liability issues, it has to be that organisations core responsibility to ensure compliance in each and every aspect of bringing cars to hundreds of markets around the world.

    So yes: There have to be some developers having implemented the software the way it is, but really these engines with this software have been used in dozens of models throughout Volkswagen Group. I am not sure how "abstract" interfaces can be if all sorts of mechanical and elektro-mechanical are involved in inerfacing an engine with the rest of the car. Fact of the matter is: you can't even start a modern engine unless it gets signals from all sorts brake, exhaust, drivetrain etc systems. So I don't think it is possible to just drop an engine into a different model and tightening the screws is all that is required for it to work.

    So for each of those models, it is safe to assume we have dozens of test and development cycles, some of which probably required customising the engine software.

    Therefore while there may be some level of "fault" with developers (although based on the "who profits" / "follow the money" principles it is probably safe to assume, this is not something that has been exclusively devised by some developer group ...), all of the responsibility is with compliance. And this is true whether compliance checks have been ignored due to negligence or intent.

  • I don't buy it.

    I certainly do not believe that the shifting of a normal in the case of the Space Shuttle is equivalent to deliberately changing the way an engine works so it has two modes; one normal operating one and one for reporting.

    Any technical person knows that you report on what is.

    My personal opinion (which may change as we find out more) is that it was known at some managerial level. Remember that this has occurred across the group. We are not just talking VW cars here but also SEAT, Skoda and Porche. Clearly, this is not isolated to a single team which makes this cultural. Also, they have form: VW were fined for a similar event in the 1970s.

    Gaz

    -- Stop your grinnin' and drop your linen...they're everywhere!!!

  • I worked for six years with a world-famous motor manufacturer, in their R&D department. Unless a hell of a lot has changed, there is no chance whatsoever that management weren't aware of this. The people doing performance special models could have possibly got away with it but even there, the decision would have been taken at a pretty senior management level. In my day, admittedly some time ago, almost every car manufacturer did a little cheating, by putting in a very lean strategy when the car door was open. (the 'man from the ministry' who ran the tests used to keep the door open so he could see the readings on the machine). This seemed perfectly legal since, if the car door was open, why would anyone want full power? This got 'found out' around 1991, the test equipment got less bulky, and everyone stopped doing it.

    Best wishes,
    Phil Factor

  • My opinion on this just from my own experience is that some level of management knows about something. As you get higher to the top, the information tends to get watered down to a point where knowledge of what exists is nonexistent.

    In example, my boss is just a copy of me, but he manages now. His boss is a copy of him, but he has managed longer and has not been like me for some time. His boss may run the company on behalf of the shareholders. Somewhere between me and the shareholders, someone dropped the ball. But, someone surely knew what was going on.

  • The software controls a physical machine. Are you saying that the finest German engineering goes into crafting every part of the VW except the emissions system which is handled by Joe the intern? Are you saying that this problem was able to walk through every level of QA that VW has without anyone noticing? VW is a massive company -- are you saying no one anywhere noticed anything was wrong?

  • I still haven't seen anything that says the other manufacturers aren't doing it. If anything VW are cleaner than the average.

    I can imagine a senior manager insisting that system 'X' be delivered with mutually contradictory features and the engineers produce something that convinces him that he's getting it. Trying to convince a senior manager that the chocolate teapot they have promised to the board is actually not going to work is a thankless task. The Dilbert principle comes into play.

  • [font="Verdana"]Its insane that it was due to some intern etc.

    But on the other hand if any requirements come to us; to cheat the world by even the company we are employee of as a developer or manager, we should have a tendency to set apart our self from such nonsense.

    Proper regulatory checks should be enhanced and even functionality of each item (SW) should be independently verified.

    [/font]

  • I too am sceptical that management didn't know. If it genuinely didn't know, it needs to explain to the world, how the rogue developers were able to keep this information so effectively concealed. It hasn't done this so far.

    Of course the other side of the coin is that, if the management did not know about something as important as this, how effective could they be in running VW? Perhaps VW just ran itself, in which case what is the justification for the large numbers of highly paid people employed.

  • Management throwing people under the bus. Again. Nothing more.

    --------------------------------------
    When you encounter a problem, if the solution isn't readily evident go back to the start and check your assumptions.
    --------------------------------------
    It’s unpleasantly like being drunk.
    What’s so unpleasant about being drunk?
    You ask a glass of water. -- Douglas Adams

  • Can they do it? Yes. Should they do it? No.

    Saying the tests were 'too stringent' says they don't understand why the tests are set up that way. And if they're being evaluated on meeting those tests then of course they're going to be tempted to fudge the software.

    They're not nearly the only company that's been caught making special code to get specific test results regardless of real world performance either.

    Not to play on words here but you get what you measure. If developers are measured on meeting the tests then they're going to meet the tests in whatever way they can. If management figures it out but doesn't change the criteria then nothing is going to change. It's like giving developers a bonus based on bugs found.

    Changes should not originate from the developers. Suggestions and proposals do. Then they're evaluated and a decision is made whether or not to implement them. It's not a case of "It's better to ask forgiveness than permission" when it's got this kind of visibility.

  • Maybe not quite as bad as the VW situation, but Peugeot were caught a few years ago manipulating the odometer on vehicles. Every so often, it would add a few miles to the odometer (e.g. after opening and closing the door a certain number of times).

    When caught, they explained that "it more accurately represents the true wear and tear on a vehicle".

    Of course it was nothing at all to do with the fact that people would have the vehicles serviced sooner/more often, and that, based on the mileage, they may replace the vehicles sooner, and that owners would think they were getting more mpg than they really were :hehe:

    What is so difficult to comprehend is that Peugeot didn't have the common sense to realize that an eagle-eyed owner would sooner or later spot the deception... in fact the same thing goes for VW.

  • Management bears complete responsibility for ethics. Any notion that they do not is preposterous.

  • Hmm... it sounds like this is a case of what happens when people are managed to metrics instead of actual results.

Viewing 15 posts - 1 through 15 (of 33 total)

You must be logged in to reply to this topic. Login to reply