I recently finished reading (actually listening to as it was an audiobook) Skunk Works: A Personal Memoir of My Years of Lockheed by Ben Rich.
tldr; I quite enjoyed the (audio)book. It’s a fascinating look at how a collective group of people can find innovative solutions at the intersection of science and technology, creating something that is almost more of an art form in the beginning and only later becomes a common, standardized form.
Suffice it to say, there are lots of interesting parallels in the book to creating software.
One of the little nuggets that stood out to me was how designers and engineers behind the original stealth plane, the Lockheed F-117 Nighthawk, had to subvert the “best practices” of creating an airplane in service of creating something truly stealthy.
The condensed version of the story goes something like this:
In the 1960’s, a soviet mathematician first conceived of the theory behind how to reduce radar reflection. But his theory was ahead of its time. The design necessary to meet his theory would result in an aircraft too aerodynamically unstable to fly by a human pilot.
Years later, however, an engineer at the Skunk Works realized that computer technology for flight was sufficiently advanced that, while a human pilot couldn’t keep such an aircraft afloat, a computer could.
So they set out to build a prototype they called “Have Blue”. The aircraft’s eventual design featured “faceted surfaces to deflect radar waves away from a radar receiver”. In many ways, it was the very antithesis of what people believed an airplane should be.
Even the Skunk Works director, Kelly Johnson, had a hard time coming around to the design of the plane. He believed the epitome of an airplane was one with smooth, blended edges making it aerodynamically stable and efficient. Is that not the very purpose of an airplane? Flying smoothly and efficiently through the air?
But this stealth aircraft had as its design principle: stealth over all else. Who cares if it’s aerodynamically inefficient? Who cares if it doesn’t look like an airplane? Short of “it must be able to remain airborne” everything else was on the table.
What they came up with was an airplane made of two-dimensional flat surfaces — like a bunch of little triangles put together. To many, it seemed ludicrous. Some balked at the idea of such a craft even being able to fly. From Wikipedia:
The resulting unusual design surprised and puzzled experienced pilots. A Royal Air Force (RAF) pilot who flew it as an exchange officer stated that when he first saw a photograph of the still-secret F-117, he "promptly giggled and thought [to himself] 'this clearly can't fly'".
But the prototype worked, and the results spoke for themselves: the airplane's appearance on radar was orders of magnitude less than any other aircraft in the world.
In fact, at one point in testing, their model went completely undetected by military personnel until a bird landed on it during the test.
Now-a-days, the iconic shape of the original stealth plane has changed because technology has caught up and made stealth possible for more traditionally-designed aircraft.
It’s a good reminder when you’re working on something to continually ask yourself about the purpose behind what you’re making. It’s very possible you might have to deviate from the “best practices” or “accepted conventions” in service of a goal that is different or beyond the tradition of any medium or form.
It’s also an intriguing example of how far a principle can take you. In their case, stealth above all else made people invent some intriguing and creative workarounds to the otherwise traditional constraints and pre-conceived notions of an airplane’s design.
In this way, best practices are kind of like a grid in design: useful to follow, but where it gets interesting is where you break out of the grid with purposeful intent.