The Uncanny Prototype Valley
Over the course of my career so far, I've made lots of prototypes. They're helpful tools to demonstrate interaction patterns so clients can visualize how something will look and how it will work. They're not perfect, but that's kind of the point. They're supposed to be a proof of concept so that you can quickly show how something will work, get approval from clients, and start building.
But I've noticed something happens almost every single time I show a prototype to a client or stakeholder. They like the concept, they approve the direction, but rather than start building the real thing, they want to improve the prototype so that they can show it to people higher up the totem pole.
I understand why this happens, you're about to show your boss, so you want things to look as professional and polished as possible, but that's sort of missing the whole point of a prototype. It's supposed to be unpolished. It's supposed to just be a sketch.
So inevitably, the team will stop building and iterating on the actual thing so that they can spend time making the prototype look more polished. This can be a mobile app, a desktop-based website, whatever it is; inevitably stakeholders will want to make sure everything in the prototype is pixel-perfect, colors, design, layout... even down to making sure the placeholder names and emails and VIN's make sense. (purely hypothetical, obviously.)
Inevitably we spend a lot of time just fixing up this prototype that will get thrown away the minute after the big meeting with the executives, time that we could be spending advancing the design and making it better.
This has been so confusing to me that I had to understand the phenomenon, and I think I've found a near perfect parallel that has benefited from more documentation in recent years.
It's this concept in the world of artificial intelligence called the uncanny valley. Japanese roboticist Masahiro Mori first coined the term in 1970 and basically the idea is, if you make a robot, and you give it the face of some kind of appliance, you know — square head, sine-wave mouth, standard robot look — people will expect less of it, and it may actually over deliver on what humans expect. However, as the appearance of human likeness increases and becomes more uncanny, there is a portion of the graph that ceases to go up and to the right, but instead dips down into a horrifying valley, referred to as the uncanny valley.
Scientists realized that the more human an artificial entity looked, the more unsettling it was to their human counterparts... in layman's terms, the more real it looks, the more creepy it seems.
Scientists and roboticists still can't explain the phenomenon, but I've got a theory based on my experience with prototypes.
When a robot seems less realistic, you're able to suspend your disbelief, and you enter into a kind of non-verbal agreement where your expectations are low, and whatever the robot can do is impressive. Say it's a refrigerator, and you can keep your drinks cold, but it also tells you the weather. That's great. That sounds like an upgrade. You're not wondering why the refrigerator can't drive a car or walk up stairs.
But when the robot looks almost exactly like a human, you notice every tiny imperfection, because you're looking for it. When the designer of the robot made the robot with human features, they're entering into a non-verbal contract with you in which they say "don't bother suspending your disbelief, view this machine as a human," so maybe an eyelash is out of place, ears aren't quite level, accents aren't quite right... they all seem like glaring issues, because you're now thinking of this thing as "real" and can no longer see past the things that make it seem not so real.
This is a perfect illustration of what I've found prototyping the past few years.
When the prototype just looks like a sketch or a wireframe, clients and stakeholders are able to suspend their belief because the lack of design is a visual cue that this thing isn't finished. It's just a prototype, so no one expects it to work completely right in every scenario. Maybe a name is misspelled, maybe a button isn't the right color, but no one cares, because it looks unfinished. It gets the point across, which is the whole point. People can look past the lack of precision and see the idea, because that's clearly the only thing that's there. The function without the form.
But now they want to show the prototype around to executives to get funding, so they say "can we add color, and maybe tighten up the layout a little, and we just need to be able to click the submit button and see a confirmation." All well and good, but now we're starting to move along the x-axis up and to the right, dangerously close to the uncanny valley.
The next time the client sees it, it looks a little more finished, a little more designed, and yet a little more imperfect, even though it's actually more polished. This is the uncanny valley taking effect.
Now, that misspelled name, that jumpy transition, that mismatched VIN are all glaring problems that need immediate resolution before we can show this to the board. Oh, and by the way, I noticed that when I put in a real credit score, it doesn't adjust the monthly payment to reflect the correct APR, can we fix that too? (again, totally hypothetical)
When things start to look more polished, people start looking for imperfections, because they haven't been visually cued to suspend their disbelief. The very fact that there are details means they've been encouraged to look at the details. And what this means is that the creative team spends a lot of time doing throwaway work on a disposable proof of concept instead of iterating on the design.
Maybe one day, we'll live in a world where we figure out how to escape the uncanny valley... but for now, we'll smile and nod at the executives and say "Oh wow, I didn't see that, thanks for pointing that out. We'll get that fixed right away."