Driverless cars

You’ve heard the news about the latest technology, driverless cars and trucks*.  Can it happen?  It has.  Already four states have made it legal.   Accidents have happened, and promoters have added radar to the detection array. Google’s explanations are amusing, calling it a misunderstanding and a learning experience.  But there is yet another serious hurdle.  It is statistical. Statistics #1.  If suddenly we changed to driverless cars, there would be accidents, lots of them.  The software would require multiple tunings. This takes time and testing; even the process of updating takes time. Accidents would continue.  Scrutiny would increase, the barrage of them keeping this always in mind.

On the basis only of accidents and their visual statistics, the program might possibly be abandoned.  But if not... folks would not be driving their cars anymore.  Cars would become mere taxis.  The thrill would be gone.  Car sales would sag, with people replacing them only as needed.  Car advertising would decrease and involve transportation/reliability/functionality issues only.  Nowadays, most car ads are directed toward forms of independence, thrill, power, and sport.  When your car is doing all of it, the excitement is gone and with it, sales.  Every car would become an automatic-ish Subaru. Yuck. 

Possibly, with this exception in mind, Tesla, the luxury electric car, is testing driverless software.  Great, but who will by this fantastic and expensive auto, only to sit by will it drives along with them out of the loop.  Testimonials about how great will arrive soon, nonetheless.

But driverless transportation?  I'd love it.  At my age, I just want to get there, even while reading a book if possible. 

Now suppose we have a benevolent government seeing the magnificence of automatic driverless cars?  With hundreds of millions of cars on the road, the transition to such could not be over night.  Even this would take time.  In the interim, even more accidents would occur.  Rush hours would become impossible, particularly in big cities.  Public outcry would result when the driverless software would encounter operational decision conflicts, resulting in operational failure. The beneficent leaders would need to regulate strongly enforced by laws, levies, and even jail for those “switching off” the programs. Election outcomes would hang upon driverless mandates. 

Already there has been an accident or two from driverless vehicles. This was disappointing to the programmers at Google. Now we hear Uber** is contracting with Carnegie-Mellon to perfect the software for driverless taxis. In view of the above comments, they must have some business plan to make it a reality.  It will take more than 1,000,000 miles of faultless conveyance.  Is this enough?  Could Google have an inside track to regulatory allowances.

Other risks abound. Here's one. We may be assured that when such self-driving software for cars or trucks become widely available, it will become a formidable tool for terrorists. Just plug the destination into the on board computer, and let it go - perhaps to destroy a major thoroughfare or perhaps an important tunnel. Walk away. The same remains true when the public are invited to trials of the software on their personal vehicles. Buy a wreck for a couple hundred, add the software, load it with explosives, and set it in motion to pre-set location.  Boom.

More statistics.  We have just seen the battery breakdown of The Samsung Galaxy 7 battery after about 1.5 million or more units were shipped.  The fault is not with Samsung, I believe.  It is statistical #2.  They may have tested one or even ten thousand of the batteries, but these faults only occur at the one million level and more.  This is a problem with sampling at scale. It is what makes getting new drugs approved so expensive, and what turns new drugs into multi-billion dollar lawsuits so lucrative. The FDA knows this well. It is the scale of rarely occurring events that do not manifest until very large scale tests are conducted.  Of course, Samsung could not test a million of its batteries.  Too expensive. A random guess is that Samsung bought into a new-ish battery technology that passed all their tests, but failed the big one – mass deployment. 

But driverless cars seem to be the direction we are heading.  From a common sense perspective, the numbers of variations of driving and road conditions is too vast and not yet possible for extant software products to process.

Fully driverless car options may have a merit not yet possible now or in the next decade.  But remember, beneficence may be purchased nowadays, probably with a price tag most high.

* https://en.wikipedia.org/wiki/Google_self-driving_car

**http://www.npr.org/2016/09/17/494394898/a-look-at-ubers-ambitions-for-a-driverless-future