Many people have heard, and almost as many have repeated, the Third Law asserted (in 1973) by the late author and visionary Sir Arthur C. Clarke:
Any sufficiently advanced technology is indistinguishable from magic.
Fewer, I’m certain, have considered (or lived up to) the simple, logical transformation propounded by Barry Gehm in 1991:
Any technology distinguishable from magic is insufficiently advanced.
I freely admit that I have quoted Gehm, more than once, as setting a bar that must be cleared to fulfill the promise of “cloud services”: a standard of abstracted, robust capability, defined by an interface or an experience rather than a mechanism. I have not changed my mind about that goal, but recent incidents make me wonder if we are being honest about the gap—which will, surely, always exist—between magical aspirations, and the real-world imperfections of increasingly complex and dangerously opaque machines.
We may have to do more than admit that the gap exists: we may have to expose it, even call attention to it, if we are not to commit a dangerous kind of malpractice.
Are talented people striving to give others an experience of magic? Absolutely, and so they should. As I discovered while verifying my quotations above, my good friend Alistair Croll has admonished startup founders that “if your product doesn’t seem like magic to your target market, something’s got to change. A magical product means the difference between a long, costly sales cycle and a market that just pulls what you make out of your hands.”
Alistair is of course correct, and his advice aligns with the dictum of the late Steve Jobs that “people don't know what they want until you show it to them.” The corollary, it seems clear, is what happens when you do show people something they recognize as what they always wanted, but lacked any frame of reference in which they could visualize it. In the best of these situations, you create a new product category; people may even invent a new word to describe what suddenly fills a conceptual hole that had never before been noticed.
For example, people were downloading all manner of content from the Internet for years before Apple introduced the iPod in 2001, but in 2004 the word “podcasting” became the term that described such sharing. A coinage like “podcast” signals that the curves of experience versus implementation have crossed, taking us into a domain where something is easier to do than to describe: this may be a useful signpost of approaching Third Law attainment, and this is a good destination for us to seek. When we reach for the stars, we don’t come up with a handful of mud.
Even so, I feel as if recent events—ranging from personal incidents to globally noted disasters—require a rule of engagement that admits we’re not there yet. What I’m proposing is not so much a law of how things work, as it is a code of conduct like the Hippocratic Oath. It seems to me that we need a statement that I’m perfectly willing to see called Coffee’s Caution:
Since you know that what you’re selling isn’t magic, admit it—and give your customers what they need to deal with the difference.
Yes, there are examples: I’m not indulging in fantasy, but I’m not done here yet.
What started me thinking along these lines was a morning when I unplugged my phone from a hotel-room electrical outlet, to find that it had not charged its battery during the night. As it turned out, the electrical outlet that I used was equipped with a ground-fault interrupter device, which apparently had been triggered by some cause unknown. I pushed the outlet’s reset button, restored normal function, and collected some precious Joules before I had to go out and face the day.
Was this all my fault? I don’t know: I would think that I’d have noticed if the phone did not make its usual “OK, I’m charging” noise when I plugged it in, and perhaps the outlet only went into its fail-safe mode later on. What’s certain is that the outlet itself gave only a subtle signal of anything out of the ordinary: a pushbutton protruding by, at most, an extra millimeter from its usual position. That’s easily overlooked, and perhaps effectively invisible to anyone who has never had this happen before and would not have had reason to look for it.
This is a sub-microscopic annoyance compared to the past few months’ pair of fatal airliner crashes, each killing an entire complement of many dozens of passengers and crew. Even so, these failures have noteworthy parallels, despite their being on totally different scales. In each case, a device (perversely, one that was intended to provide added safety) silently failed; a user interface did not clearly signal the failure; a person relying on the normal behavior of the device did not get what was expected, and only previous experience with that failure mode could lead to prompt recovery from the situation.
It appears, as of the time that this is written, that only some of the buyers of the aircraft type in question chose to purchase two optional additions to its flight controls. The first was a warning light that signaled when two different (and meant to be redundant) sensors disagreed. The second was a panel instrument that showed the actual readings from those sensors, rather than letting a single faulty sensor trigger a complex and dangerous automatic behavior without clearly telling the pilot what was happening.
Whether such things should be extra-cost options is surely among the questions that will be raised – but my broader point, here, is to note how many of the things we use today lack even the opportunity to ask for such insight into what they are doing.
When a thing-that-fails-to-be-magical is not working as it should, what is the user to do? On another recent day, for example, I found that a wearable fitness tracker was not performing its usual sync-up with phone and cloud to update its overnight measurements. Error code displayed on the device? None. Diagnostic message in the corresponding smartphone app? None. Text or email from associated cloud service? None. Resolution strategy? Restart the device, “forget” and then re-pair the device with the phone, remove and reinstall the app, perform a ritual burning of incense and place the device in the smoke cloud… No, I did not do the last of those, but the whole process felt about that primitive.
Airline pilots can, and should, be exposed to a wide range of realistic simulations of foreseeable failures, so that the proper response is essentially a matter of habit – even when a real-life version of a failure is exceedingly rare. Most of us are neither responsible for other people’s lives, nor under pressure to make immediate decisions. Most of us have the luxury of asking a device, “Where does it hurt?”
Unfortunately, too many devices—and far too many multi-layer systems—are not equipped to tell us what is not working as per specification. Their designers either do not wish to take the time, or do not think we will accept the costs, required to make devices admit that they’re not magic. It’s up to us to demand that honesty, and to face the fact that we can either pay up front for the tools to solve a problem – or pay later, perhaps much more, for the consequences of not knowing what’s gone wrong.