[colug-432] Great moments in computer science

[Rick Hornsby]

Steve Roggenkamp wrote:
> One problem comparing Toyota to Boeing/Airbus is the different markets
> they serve.  Toyota and other car makers sell millions of vehicles to
> unsophisticated consumers.  Boeing and Airbus sell a few hundred
> aircraft a year to very sophisticated buyers who have very specific
> requirements.  When a car crashes, it might make local news if a fiew
> people die.  When an airliner crashes, it makes global news, especially
> if people die.
Yep, you make completely valid points.  There are major, huge 
differences in the two markets.

I'm not as much advocating for the same kind of oversight provided by 
the FAA and your local FSDO, but rather similar cooperation and 
oversight from the manufacturers as it relates especially to their software.

If there is a ginormous spike attached to the steering wheel pointing at 
the driver's face - that's a visible, physical, tangible danger.  If a 
car company puts bad springs, or swiss cheese for an airbag - those 
things can be discovered by disassembling the vehicle (iFixIt style, if 
you will) or more likely during routine maintenance.  The market will 
generally deal with those things swiftly.  If Toyota is hiding a giant 
spike (or dozens of them) pointed at the driver's head in their software 
(it's not a spike, it's a feature!) - and then cowering behind IP/DMCA 
laws - no one will ever discover the problems until something like a 
trial comes up - and even then Toyota was extremely uncooperative to the 
point of active deceit.  Toyota it seems settled to avoid more of their 
hackery (in the bad way) from coming out.

> Individuals may buy a car every few years from a different manufacturer,
> so there is not as much commitment from the customer to the manufacturer
> and vice versa.  Plus dealers come between the car buyer and the
> manufacturer.  Need I say more?  Car manufacturers can sweep problems
> under the rug as long as the numbers are not too big.  Look back to
> Ralph Nader's "Unsafe at Any Speed" from the early 1960s to see that's
> been the modis operandi for decades from car manufacturers.
We're in the midst of a massive paradigm shift where tens of thousands 
of lines of software - that as you correctly point out there is little 
incentive to fix or get right - are at least in part operating 4000lb 
machines with novice operators behind the wheel who have no training 
(none is available - because the software is awesome) and precious 
little time to override faults before hurling off Jeff's embankment.  
The shift is toward more software - including what amounts to full 
auto-pilot for these vehicles.

What happens when the collision avoidance mechanisms (radar, lidar, 
other RF signals) between two vehicles interacts in a way that results 
in someone getting killed?  No one is looking at the software and quite 
obviously from the article (and your insightful statements) no one will.

Before we unleash fully automated FBW cars onto our roads, is it 
possible to balance intellectual property rights - a person or company's 
investment in software - with open, audit-able, and standards-based 
conformance in these mission critical applications?