[vox] The real problem with Toyota acceleration!

[Bill Broadley]

Norm Matloff wrote:
> When cars with ABS first came out, I refused to buy them, simply because
> software glitches like this would be too easy to pop up. 

True.  Then again it doesn't seem terribly hard for ABS to be statistically
safer than the chance of an ABS software error making things worse.   Alas
people don't tend to drive the same speed with ABS, it raises the performance
envelope and people adjust the speed they are comfortable with.

Maybe they should just mount a dagger in the middle of the steering well.  Or
maybe add disconcerting force feedback/noises to make things feel as bad as
they are.  Seems like between ABS, sensors, and differentials that the car
would tell when you are driving unusually close to the limit and that you
might not be aware of it.   Then again would people respect a blinking it's
slippery light on the dashboard if the car feels solid at posted speed limits
in heavy snow/rain?

> I buy them
> now, but only because it's almost impossible to find a car without them.

Seems a bit paranoid, even if they help in only 0.05 % of accidents that's
many times more than the experienced ABS software problems.

> We've not heard the last of such problems.

Previous cases of unintended acceleration seemed a bit silly, basically
american car drivers used to wide brake pedals ended up forgetting and hitting
the gas.  Surprising how marketing (to american sedan drivers) had the
unintended consequence of almost sinking the company despite no engineering
problems.

> On Tue, Feb 23, 2010 at 10:22:04AM -0800, Brian Lavender wrote:
>> I couldn't help entertaining myself with the following. What if Toyota
>> was using an 8-bit micro controller, and they were using a signed value
>> to track the position of the accelerator? You press more and it adds to
>> a vector. The problem strangely started occuring when Toyota started
>> taking out the throttle cable on their vehicles. I own two Toyotas. One
>> with a throttle cable, the other without (uses a sensor to detect
>> throttle position).

Heh, seems rather unlikely.  I am aware of various problems where the common
sense solution of decreasing the timestep to increase accuracy instead
resulted in the error accumulating faster.  I believe a stock market (outside
the USA) had that happen.  Related problems happened to patriot an anti-missle
missle system and a rocket launch system.

In this particular case (toyota) though it sounds like it might be:
http://en.wikipedia.org/wiki/Whisker_(metallurgy)