Edit: Blogger decided to post a blank post, so now I've added content for you to actually read. I don't expect any of you to really appreciate nothing on a page, except maybe Sam, who could probably find value in it as a break from studying.
--------
Found this online a bit ago:
I've been intrigued by the Toyota recall scandals lately. On one hand, when I see unsubstantiated claims of death and disaster and catastrophe because of automotive defects, I automatically assume that it boils down to stupid people and inconceivably bad driving (as the article suggests). This is particularly so because we're talking about Toyotas, which are traditionally about as well put-together as cars can be. On the other, it brings a very interesting question to the forefront: at what level are we willing to trust in technology, and specifically software, to perform perfectly in life-or-death situations?
I've been around software code most of my adult life, and it almost never performs totally up to spec, thanks to human error. Having said that, there are a lot of situations in which it had damn well better. On one end of the spectrum is Toyota, the utterly mundane -- at least until they race off at 15 million miles an hour because you're an idiot, and you have your foot on the wrong pedal, or you forget how neutral works or that you have a key that you can turn to make it, you know, stop. On the other end of the spectrum, to name one example, F-22 fighters pilots put their faith in highly complex machines that are so unstable in flight that they literally cannot be flown in anything approaching a straight line without enormous computer assistance. (A key point here is that if you cannot fly your advanced jet fighter aeroplane in a straight line, you are relatively more likely to die than in many other, more recreational pursuits. One of these pursuits may be that your car decides it wants to go faster for no better reason than it is feeling ornery that day.)
Luxury cars literally have over a million lines of code in them on average, and that one overlooked function in all those lines could kill you. That said, odds are you're more likely to get run over walking down the street by a car that works just fine. I can confidently state that I am unworried about software written by a bored technician off in a lab somewhere killing me directly or indirectly, but as time goes on, more and more everyday tasks and jobs and public functions will be handled by computers. They may not be as complex as fighter planes, but whether they're handled well or not may just depend on how hungover the coders / testers were the day they wrote the code that tells your car when the throttle is open. (This is not to mention lazy employees deciding they don't really need to test that one last edge case.)
Does that worry any of you? I'd be interested in your thoughts.
Even before cars became computerized, we entrusted our lives to the engineers who designed our cars. A badly designed gas tank or pedal assembly can kill you just as dead as a badly programmed computer. The only difference is the complexity of the systems. A badly designed brake pedal or fuel pump can be easily spotted by your local auto mechanic, or picked out by an editor at Car and Driver. A buggy piece of code is much more difficult to catch. As such, we have to put our trust in the hands of the designer to a far greater extent than we did in the past. Because of that, the manufacturer has an obligation to be that much more stringent in their product testing and quality control. When problems like this slip through the system, it calls that trust into question. If they aren't going to catch these things before they hit the streets, who will?
ReplyDeleteI guess when I go into surgery starting may, I shouldn't be too surprised when the laparascopic hands inside someone's abdomen suddenly decide to open wide instead of gripping something, therefore ripping your appendix and intestines into ribbons. I'll just think, "Man, sucks for this guy that some other dude was lazy with his coding."
ReplyDeleteMedical schools have kind of two big pushes within the past decade or two. One has been to start adding "professionalism" and "humanism" classes since in the past these things haven't been "taught". The other push has been to understand technology more and to take lab results with a huge piece of fucking salt. It's like a grain, but way bigger. With newer tests being developed everyday, it's not too much of a surprise that the enterprising young physician who doesn't have a solid grasp of what disease process you might have will take the shotgun approach to screening and diagnostic tests, likely placing you under unnecessary radiation and expenses. Besides, many of those tests that you hear about in popular media have very high fail rates to detect or diagnose disease (pap smears, mammogram, colonoscopy). They've been pushing us to go back and do more of the hands-on physical exam stuff.
I guess the point here is that in medicine, clinicians are getting scared of crazy technologies fucking up for them, so we're all learning to take a couple steps back.