Hacker News new | ask | show | jobs
by pjc50 4464 days ago
GM is currently recalling 1.6M cars which can accidentally turn off while driving: http://www.thewire.com/business/2014/03/gm-recall-mary-barra...

(this has apparently resulted in a number of deaths, as the power steering stops working)

4 comments

Not only were they aware of the problem, they did the calculus and decided to release a brand new model knowing this issue effected.

http://consumerist.com/2014/03/25/gm-knew-chevy-cobalt-ignit...

"Take the nubmer of vehicles in the field, A, multiply by the probabilty rate of failure, B, multiply by the average ouf-of-court settlement, C. A x B x C = X. If X is less than the cost of a recall, we don't do one."
Cost-benefit analysis, psychopath style.
In all honesty, how would you recommend doing cost benefit analysis for a car manufacturer? There will always be safety risks to the drivers, after all.
Yes, perhaps the law should be more strict , but in any case there is a value -- there is a number which sometime you'll have to assign even implicitly to a customer's life.

But consider the possibility that the law had a loophole (or the law is simply inadequate in light of common expectations from the population) which brought the maximum litigation from crashes to a very low number. Although this simplistic view would suggest a very low cost, that would be unrealistic -- the actual cost should include heavy brand damage from consumer distrust in the products and even brand damage from raw ideological/moral basis. This is way promoting values inside companies makes sense -- you can't neglect humans have personal values and sometimes make non-economic choices to stay aligned with those -- so you have to adjust your "psychopath" economics towards consumer irrationality. In the end I suspect the optimal choice is much closer to Tesla's reaction than GM's.

Are you serious? A non-psychopath might start by assigning a nonzero value to human life.
I see two issues with your comment. First, the quoted formula doesn't assume a zero value for human life. Second, even if you assign an explicit finite value to human life, a lot of people will call that psychopathic.
A similar equation is fine, but 'average settlement' is probably too low. Add 10 million to C before multiplying and you have a better balance, with a margin of error on the side of safety.
It's from Fight Club :)
Corporate American style. You want to guess the number of times an airplane design defect that threatens the life of passengers is allowed to be corrected "over time" -- basically because fixing them ASAP would be too expensive?
That's really comparing apples to oranges.

At the airplane industry it's the government that does this calculation, with a publicly known value of C. Also, airplanes don't have the kind of problem that cause an accident by itself, every issue can be corrected on several levels, and normally when the proper fix isn't instantly applied some other action is done to attenuate the problem.

On any advanced country, around here it calculates nothing, even yelling "Are you insane? Do you know how many hospitals we could build with that kind of money?" won't make they think about numbers.

Which company do you work for?
A major one.
Isn't it true that every car manufacturer will be aware of some level of risk and will have to make decisions based on that? There's always more you can put in a car to make it safer, but at some point you stop and release the car.

Note that I'm not saying this particular case is justified. I'm just saying that it's a continuum, and I'm not sure how we should expect car manufacturers to draw the line on the continuum.

Certainly. Knowing that a car may lose control if it runs over the alternator that fell off another car at highway speeds (as shown in the Tesla video) is one thing. Knowing the car can randomly shut off including airbags at highway speeds is another.
Don't know why this is astonishing; hell, Fight Club devoted a full 2 or 3 minutes explaining this. Big companies are evil, etc.
And just to give a sense of the orders of magnitude:

    1,600,000
       29,000
While I agree with the general point this comparison is not fair - if you sell more cars, more cars may be affected.
If you sell more cars, you should be able to invest more in engineering reliability. The cost should be spread out over many more vehicles. This all goes out the window depending on margins, priorities, and presumably other factors, but for the sake of argument: I'd say both numbers on their own aren't fair.
Or you could just report recalls as a percentage of production, and have a meaningful metric to being with...
And they were aware of the problem.