3 ms·

The Micheal Barr Group did a very high-profile autopsy on the code. The upshot is not good for Toyota.

sitkack 12y ago
Floor mats fit Gladwell's narrative much better than even mentioning software. His writing sounds like news, but it is not news, it is revisionist entertainment by omission. A half story, is half truth, is a lie.
- ArkyBeagle 12y ago
  It's an essay. An essay is whatever you want it to be. Floor mats was a sort of strawman erected ( by someone other than Gladwell ) in the history of the subject. He's illuminating the schism between how engineers think and how ... people ... who are not engineers .... think. Always a fun topic, IMO.
  - sitkack 12y ago
    I am an engineer, and I don't think in the picture he paints. An essay is obviously opinion meant to sway opinion, that is not how Gladwell writes. Both Toyota and GM acted maliciously in covering up flaws in the products they sold. For Gladwell to spin the folksy, aww shucks it is just the floor mats, you still have breaks is just as disingenuous as GM telling teenagers not to hang so much shit from their key chain. http://www.safetyresearch.net/blog/articles/toyota-unintended-acceleration-and-big-bowl-%E2%80%9Cspaghetti%E2%80%9D-code http://www.safetyresearch.net/blog/articles/toyota-unintende... > Jean Bookout and her friend and passenger Barbara Schwarz were exiting Interstate Highway 69 in Oklahoma, when she lost throttle control of her 2005 Camry. When the service brakes would not stop her speeding sedan, she threw the parking brake, leaving a 150-foot skid mark from right rear tire, and a 25-foot skid mark from the left. The Camry, however, continued speeding down the ramp and across the road at the bottom, crashing into an embankment. Schwarz died of her injuries; Bookout spent five months recovering from head and back injuries.
    - ArkyBeagle 12y ago
      I am also a (software) engineer. I work on stuff that's much like the throttle controller on the Toyota without being directly automotive. I don't think it's an opinion piece. It's just not journalism. People find Gladwell infuriating because he somewhat defies categorization and tries to take a very neutral, nonjudgemental tone. I said "essay" because that's as close as I could get. I don't buy "acted maliciously". I don't agree that it was covered up. Malice implies intent; there was no intent. Indeed, that's the problem. There really isn't, to my knowledge, a fixed protocol for these things outside of whatever management or PR protocol you follow and then possibly the courts. Toyota as an entity believed that it was acting responsibly. From the article: "The engineers were right. A series of exhaustive investigations by federal regulators, with help from NASA engineers, established that the perception of an electronic failure was almost certainly illusory. The problem was caused either by the fact that some people put in poorly fitted, nonstandard floor mats or by the fact that drivers were pressing the accelerator thinking that it was the brake." For one, Gladwell isn't the one saying it was the floor mats. The team of Federal regulators, with help from NASA engineers said that. Toyota took this seriously and launched an investigation. Prior to Micheal Barr's analysis, I'd say it was formally undecided whether there was a software problem. Even then, the defects found were not a smoking gun. This is new territory. The Barr Group is one of the first of its kind.
  - SilasX 12y ago
    >It's an essay. An essay is whatever you want it to be. I really can't say I like the trend of exploiting ambiguity to obscure the line between fact and fiction.
  - davidgerard 12y ago
    "Gladwell" is a synonym for "ignorant glibness". He doesn't understand anything he writes, and keeps right on going just as if he does. (Continued execution past a stack overflow.) Everything he writes should be assumed to be on the way to "not even wrong".
lambdaelite 12y ago
Is there a public version of their findings? Toyota hired Exponent to do an independent analysis of their electronic throttle and released that report to the public. It'd be interesting to compare Exponent's findings with the Barr Group. Exponent didn't find any plausible failure modes for the Toyota electronic throttle systems—their study looked very comprehensive, including a static code analysis and hardware-in-the-loop testing on the software side. They also did extensive hardware failure testing. It sounded like the Barr Group is claiming that the software was failing their quality tests, but it didn't sound like they were proposing a specific hardware or software failure that would lead to an uncommanded acceleration.
- ArkyBeagle 12y ago
  Right. No smoking gun. There were multiple scenarios described in which such a failure was plausible. The story is nowhere as cut and dried as in the Therac or the Arianne cases. Here are his Embedded Systems Conference slides: http://www.barrgroup.com/killer-apps/ http://www.barrgroup.com/killer-apps/
  - Aloha 12y ago
    Here is a copy of his testimony and slides he did. http://www.safetyresearch.net/Library/Bookout_v_Toyota_Barr_REDACTED.pdf http://www.safetyresearch.net/Library/Bookout_v_Toyota_Barr_... http://www.safetyresearch.net/Library/BarrSlides_FINAL_SCRUBBED.pdf http://www.safetyresearch.net/Library/BarrSlides_FINAL_SCRUB...
pnathan 12y ago
Somewhere in my computers I have a PDF of Michael Barr's testimony. I read through it (< 300 pages) one evening with a friend. Both of us had worked in an embedded device company that dealt in life and death type devices, such as cars, traffic controls, etc. So we are familiar with the tools and rigor needed to design and develop safety critical systems. It is a matter of great professional interest to us to observe what transpires in legal cases relating to software quality. Some of the key overall findings: The management was at fault, by picking people who were not experienced in developing real-time devices to lead software development (many software faults were those of inexperienced developers, and of developers inexperienced in real-time control). The lead developers were at fault, following extremely poor coding styles. The end code, of course, was badly done, as you might expect from management who didn't understand and leads inexperienced in the field. Toyota, bluntly, messed up in breadth and depth when it came to software quality control. If you can find an unsafe coding practice listed on the web, Barr found it and detailed the precise ways it was unsafe. Huge interrupt service routines, complex logic in the wrong places, huge cyclomatic complexity. It is understood in the world of safety and real-time software that certain designs of code are to be encouraged and others discouraged, based upon painful experience. It is not a handwavy "different strokes for different folks" field. The information from the Barr transcript can be used as a comprehensive manual of things to not do.

Floor mats fit Gladwell's narrative much better than even mentioning software. His writing sounds like news, but it is not news, it is revisionist entertainment by omission. A half story, is half truth, is a lie.

ArkyBeagle 12y ago
It's an essay. An essay is whatever you want it to be. Floor mats was a sort of strawman erected ( by someone other than Gladwell ) in the history of the subject. He's illuminating the schism between how engineers think and how ... people ... who are not engineers .... think. Always a fun topic, IMO.
- sitkack 12y ago
  I am an engineer, and I don't think in the picture he paints. An essay is obviously opinion meant to sway opinion, that is not how Gladwell writes. Both Toyota and GM acted maliciously in covering up flaws in the products they sold. For Gladwell to spin the folksy, aww shucks it is just the floor mats, you still have breaks is just as disingenuous as GM telling teenagers not to hang so much shit from their key chain. http://www.safetyresearch.net/blog/articles/toyota-unintended-acceleration-and-big-bowl-%E2%80%9Cspaghetti%E2%80%9D-code http://www.safetyresearch.net/blog/articles/toyota-unintende... > Jean Bookout and her friend and passenger Barbara Schwarz were exiting Interstate Highway 69 in Oklahoma, when she lost throttle control of her 2005 Camry. When the service brakes would not stop her speeding sedan, she threw the parking brake, leaving a 150-foot skid mark from right rear tire, and a 25-foot skid mark from the left. The Camry, however, continued speeding down the ramp and across the road at the bottom, crashing into an embankment. Schwarz died of her injuries; Bookout spent five months recovering from head and back injuries.
  - ArkyBeagle 12y ago
    I am also a (software) engineer. I work on stuff that's much like the throttle controller on the Toyota without being directly automotive. I don't think it's an opinion piece. It's just not journalism. People find Gladwell infuriating because he somewhat defies categorization and tries to take a very neutral, nonjudgemental tone. I said "essay" because that's as close as I could get. I don't buy "acted maliciously". I don't agree that it was covered up. Malice implies intent; there was no intent. Indeed, that's the problem. There really isn't, to my knowledge, a fixed protocol for these things outside of whatever management or PR protocol you follow and then possibly the courts. Toyota as an entity believed that it was acting responsibly. From the article: "The engineers were right. A series of exhaustive investigations by federal regulators, with help from NASA engineers, established that the perception of an electronic failure was almost certainly illusory. The problem was caused either by the fact that some people put in poorly fitted, nonstandard floor mats or by the fact that drivers were pressing the accelerator thinking that it was the brake." For one, Gladwell isn't the one saying it was the floor mats. The team of Federal regulators, with help from NASA engineers said that. Toyota took this seriously and launched an investigation. Prior to Micheal Barr's analysis, I'd say it was formally undecided whether there was a software problem. Even then, the defects found were not a smoking gun. This is new territory. The Barr Group is one of the first of its kind.
- SilasX 12y ago
  >It's an essay. An essay is whatever you want it to be. I really can't say I like the trend of exploiting ambiguity to obscure the line between fact and fiction.
- davidgerard 12y ago
  "Gladwell" is a synonym for "ignorant glibness". He doesn't understand anything he writes, and keeps right on going just as if he does. (Continued execution past a stack overflow.) Everything he writes should be assumed to be on the way to "not even wrong".

lambdaelite 12y ago

Is there a public version of their findings? Toyota hired Exponent to do an independent analysis of their electronic throttle and released that report to the public. It'd be interesting to compare Exponent's findings with the Barr Group. Exponent didn't find any plausible failure modes for the Toyota electronic throttle systems—their study looked very comprehensive, including a static code analysis and hardware-in-the-loop testing on the software side. They also did extensive hardware failure testing. It sounded like the Barr Group is claiming that the software was failing their quality tests, but it didn't sound like they were proposing a specific hardware or software failure that would lead to an uncommanded acceleration.

ArkyBeagle 12y ago
Right. No smoking gun. There were multiple scenarios described in which such a failure was plausible. The story is nowhere as cut and dried as in the Therac or the Arianne cases. Here are his Embedded Systems Conference slides: http://www.barrgroup.com/killer-apps/ http://www.barrgroup.com/killer-apps/
- Aloha 12y ago
  Here is a copy of his testimony and slides he did. http://www.safetyresearch.net/Library/Bookout_v_Toyota_Barr_REDACTED.pdf http://www.safetyresearch.net/Library/Bookout_v_Toyota_Barr_... http://www.safetyresearch.net/Library/BarrSlides_FINAL_SCRUBBED.pdf http://www.safetyresearch.net/Library/BarrSlides_FINAL_SCRUB...

pnathan 12y ago

Somewhere in my computers I have a PDF of Michael Barr's testimony. I read through it (< 300 pages) one evening with a friend. Both of us had worked in an embedded device company that dealt in life and death type devices, such as cars, traffic controls, etc. So we are familiar with the tools and rigor needed to design and develop safety critical systems. It is a matter of great professional interest to us to observe what transpires in legal cases relating to software quality. Some of the key overall findings: The management was at fault, by picking people who were not experienced in developing real-time devices to lead software development (many software faults were those of inexperienced developers, and of developers inexperienced in real-time control). The lead developers were at fault, following extremely poor coding styles. The end code, of course, was badly done, as you might expect from management who didn't understand and leads inexperienced in the field. Toyota, bluntly, messed up in breadth and depth when it came to software quality control. If you can find an unsafe coding practice listed on the web, Barr found it and detailed the precise ways it was unsafe. Huge interrupt service routines, complex logic in the wrong places, huge cyclomatic complexity. It is understood in the world of safety and real-time software that certain designs of code are to be encouraged and others discouraged, based upon painful experience. It is not a handwavy "different strokes for different folks" field. The information from the Barr transcript can be used as a comprehensive manual of things to not do.