Security is a design choice. It is a choice that needs to be made from the get-go, one that will always cost a significant amount of money and time and one that can work against the intended goal of a project.

I was approached my freshman year at college to work on a Formula SAE car. This is a little race car powered by motorcycle engine, designed and built by college students. My school was renowned for doing all sorts of crazydumb stuff - whereas UW welded theirs up out of chrome-moly, WWU had billet aluminum uprights holding four large carbon fiber tubes for rigidity and weight savings. And rather than going with standard disc brakes, they decided they were going to use grade 8 Allen screws to hold discs to the insides of the rims.

I looked at these discs and said "...I don't have a real good feeling about that. It looks a lot like an expensive way to shear a dozen allen screws the first time you tap the brakes." No, no, the grad student running the program said; they'd run the calcs and they had a factor of 100 more shear than they needed. I nodded uneasily and affirmed that I wouldn't be driving the thing so, okay, I guess their math is better than my gut. After all, they were vehicle design engineers, and I was just a guy who built cars.

It's worth noting that, minus the engineering-speak, they were basically replacing Ussain Bolt's starting blocks with number 2 Ticonderoga wood pencils stuck into the ground and then saying "not only is Usain Bolt not going to kick these things in half the first time he launches, we could replace Usain with a Clydesdale horse and we'd still be fine because math."

Sure'nuff, the first time they tapped the brakes on that thing the discs stripped right the fuck out of the rims and sheared two dozen teensy little Allen screws as if they weren't hardly there. Didn't really slow the car an iota. Fortunately the thing was going only about 50mph, in a straight line, in a parking lot.

I bring this up because piker experimenter engineers can make mistakes, and they're awful mistakes, and you end up without brakes. It's obvious to anyone who watches closely that the more innovative you are, the more likely you are to encounter problems no one has ever faced before. Even in a mature industry there are still ample opportunities for terrible outcomes - General Motors has been in business for over a hundred years and by their own estimates, has sold more than half a billion cars. But they still managed to kill 153 people with a poorly designed ignition switch.

And that's a mature industry governed by physics. Computing and information technology? They're still celebrating "go fast and break things." The first Model T rolled out in 1908. Seat belts weren't even offered until 1949 and weren't mandatory until 1968. At the rate we're going, I'll be able to Google anyone's DNA and retina scan before the iPhone 12 is out.

Meanwhile, 30 flipping years ago: