Your assumptions are your windows on the world. Scrub them off every once in a while, or the light won’t come in.
Isaac Asimov
It all started while on vacation with my wife. We were driving a rental car to a remote location in Washington state. The car had a safety feature that beeped every time I started to cross out of the lane. What’s not to like? The car was keeping me safe! Artificial Intelligence for the win! A useful merging of human and machine!
I hated it.
I was driving on a curvy 2-lane highway. It was almost physically impossible not to clip the edge of a lane though those curves. Please note I was driving sensibly and safely, just touching the line for a brief second while going through tight turn.
Why didn’t I just turn it off? I tried…. Lord how I tried. While you’re driving isn’t exactly the best time to interact with a complex interactive menu system but I cautiously made the attempt, for the sake of my sanity, to turn it off. Pushing the MENU button displayed a cryptic “LSA options ”menu item. Further tapping revealed this meant “Lane Steering Assist”. A HA! I was close! But nothing could turn it off. Even during a rest stop, I ran through the entire menu tree and the best I could find was to turn the “Sensitivity” setting from High to Low.
Narrator: That didn’t help
It wasn’t until the third day, when it became my life’s mission to blog about this, that I took exhaustive photos of the dashboard as well as videos of the menu system. Finally, in a tiny corner of the dash, down low and out of sight, I discovered a button that turned the damn thing off.
Bliss, the car transformed from a video game arcade into a standard, relaxing, yet still safe, driving experience.
Step 1: Fix the immediate problem
There is so much going wrong here: the immature tech, the hostile menu abbreviations, the confusing menu hierarchy, and the piece-de-resistance, a hidden hardware on/off button with no equivalent in the menu system. Where to start?
Initially I listed off a range of specific improvements to the buttons and menu navigation. The goal was to make it easier to quickly understand the system and turn it off. While a helpful and satisfying start, it still felt like a bandaid. Why work so hard to build a safety feature only to have the best UX options be to turn off? A bigger question is how could such a simplistic product like this make it to production?
Step 2: Concept vs implementation
I discussed my experience with colleagues and shockingly got a big dose of ‘what’s your problem?” Most thought this was a great idea, with comments like “Safety is important! If you can’t drive in the center of the lane, you’re driving dangerously.” When I tried to explain how irritating the lane warning was the response was an emphatic “Then stay in your lane!” My pain and frustration were enthusiastically trumped by an assertive appeal to a higher principle.
My wife had a similar reaction when we first started driving the rental car. She was trying to relax in the passenger seat and the warning sound was irritating. She too asked “Why can’t you just stay in the lane?” She quickly appreciated the problem when it was her turn to drive, “I feel like I’m in a video game!” she cried, white knuckling the steering wheel. Her hyper-focus on the lane position actually made her drive less safely as she paid less attention to other tasks like monitoring the rearview mirror or looking further down the road.
A naive enforcement of sensible rules can actually make things worse. I’m not against the concept but the implementation. What made my colleagues, as well as the car company, see this problem so simplistically? There is something much deeper here than just fixing buttons.
Step 3: The actual problem
My colleagues likely pictured driving on a straight highway where a warning beep would save them from drifting into another car. That’s clearly a good thing. In a sense, they pictured the platonic ideal of how the function was supposed to work. This gave them a gut instinct that bolstered their confidence. So much confidence that when I complained, they discounted my concerns.
My rental car had a simple type of Artificial Intelligence, monitoring the road and detecting lane changes. We’re seeing this type of simple, low level detection in many products. I fear we’re headed to a world filled with products like my rental car. Well meaning but naively implemented.
My discussion with my colleagues likely reflects discussions happening in many product teams which start with a gut instinct that is overly strengthened by a simple user case. There is another type of “AI” at work here: Assertive Instinct.
Assertive Instincts are well meaning. They extrapolate an overly optimistic gut reaction and apply it too broadly. Most people working in technology are familiar with Gartner’s Hype Cycle. Assertive Instincts are the reason why we have a “trough of disillusionment”. A new technology is exciting so people naturally get overly optimistic. The trough comes crashing down when people realize the unexpected power of edge cases. My rental car experience, like many of our product meetings, are just microcosms of the Gartner Hype Cycle.
Assertive Instincts aren’t good nor bad. What makes them so challenging is that it’s a deeply human way of problem solving for simple, daily use cases. It’s a shortcut that makes life practical. But when you’re trying to create a breakthrough product it can be deadly as it ignores reasonable concerns and keeps a product from fixing clear flaws.
Step 4: Break out of your instincts
Assertive Instincts are a two edged sword. We are all guilty of having them and a simple Tweet-like suggestion to “just stop it!” isn’t helpful. I’d like to share at a few things I use in my work.
Name your pain
Just giving the problem a name is an important start. There is cultural power in shared words. If we can all agree Assertive Instincts are a risk, we can be on the lookout and know when it needs to be addressed. Make introspection a part of your teams vocabulary and workflow.
Ask questions
The fundamental tension in UX design is that we don’t know what we don’t know. Assertive Instincts hide these gaps in our understanding. If the are especially assertive they can actually slow down better understanding. The only way forward is to question our assumptions, but this is a hard way to live! If you question everything, you accomplish nothing.
But when we are creating new AI products that react to users in complex ways, it’s dangerous to just follow our gut. New solutions almost always reveal new problems. We must challenge our assumptions. It’s more than a simple “Blind Yes” or a “Defiant No”. It’s a nuanced series of “How” questions that unlock deeper insights. For my rental car:
- How would this affect driving?
- How could it have false positives?
- How would road conditions affect it?
It can feel infinite as each time you answer one “How?” you discover another. It’s “Hows” all the way down. Of course, there is a fine line here. My goal isn’t to create an exhaustive cascade of questions but to at least try skewering a few of my initial instincts.
Build ugly prototypes
One effective path to ask questions is to build very fast and ugly prototypes to test out an assumption. This is where I spend most of my time as a designer. I work with engineers to build quick, fast, and very disposable prototypes to uncover what I don’t know. When I show these prototypes to others their simplistic layout is often judged harshly. I build them for me, to test my Assertive Instincts. We too often think we can only user test a final product. It’s important to test early ugly prototypes as they are such a fast way to learn.
For example, I’m currently working on a better, more mobile-first approach to editing text and I’ve probably gone through over 30 prototypes. Each starts with some Assertive Instinct I have. Spoiler alert: most of my instincts are a bit off. Some were outright wrong but most were just a bit flawed. I was on the right tract but didn’t really understand the problem fully. The act of failing taught me quite a bit and each successive version was better.
Big Finish
Your brilliant product idea is often based on an Assertive Instinct. It isn’t a solution to a user’s question but instead one that you’ve framed based on your gut. You don’t know what you don’t know. This is why user testing ugly prototypes is so helpful: it doesn’t test your product, it tests your question.
But there is no silver bullet, no simple formula for success. Combating your Assertive Instincts is something you must do even within tight time and budget constraints. I’m sure lane warning systems for cars will improve over time. It’s just a shame they have to waste so much time and money “prototyping and testing” these early solutions on unsuspecting consumers. Good UX design shortens that process, saving time, money and user frustration..
Functional products can be made with a chisel. Great products are made with sandpaper.