“Use The Force, Harry.” -Gandalf

Bots, messaging, and the Interface Visibility Scale

People who’ve been around technology a while have a tendency to think of human-computer interfaces as phases in some kind of a Jobsian linear evolution, starting with encoded punch cards, evolving into command lines, then graphical interfaces, and eventually touch.¹

Each new interface paradigm has made human-computer interaction meaningfully more accessible to greater numbers of people. However, each shift has also contributed to the false expectation — now strongly held — that any new interface paradigm should function and feel entirely differently than its predecessors.

When the last two major interface epochs were triggered by the Mac (1984) and the iPhone (2007), how could the next wave of human-computer interaction possibly be predicted?

Well, the first step is to stop thinking of human computer interaction as a linear progression.

A better metaphor might be to think of interfaces as existing on a scale, ranging from visible to invisible.

So, what is an invisible interface?

I think we can all agree that it’s the job of any well-designed tool to provide something of value — say, an expanded capability, or useful experience — with minimal work, frustration, ugliness, and cognitive load (which is a nerdy way of saying friction related to usage).

So if we think of interfaces in terms of their visibility — not literally, but metaphorically — then we would characterize highly visible interfaces as being represented by low use and high cognitive load, necessitating a high degree of pre-requisite training. Similarly, an interface might make itself occasionally more visible by erroring out, taking too long to complete its task, or just being poorly designed in certain important ways. Examples of visible interfaces would include the punchcard, many command line interfaces, and quite a bit of very useful, but ultimately shoddy, pieces of software.

A visible interface might actually be confused for intuitive after 10,000 hours (see: Stockholm syndrome), but generally its hurdles to adoption are massive. To be clear, there’s nothing inherently wrong with highly visible interfaces. They’re often extremely useful and necessary— I still have a command line open at almost all times on my Mac — but we ought to take the time to understand what it is about them that makes them so visible.

Completely invisible interfaces, on the other hand, would be characterized by frictionless, low cognitive load usage with little to no (apparent) training necessary.² Invisibility doesn’t necessarily mean that you can’t physically see the interface (although some invisible interfaces may actually be invisible); instead, think of it as a measure of how fast and how much you can forget that the tool is there at all, even while you’re using it.

Examples of interfaces that approach invisibility include many forms of messaging, the Amazon Echo, the proximity-sensing / auto-locking doors on the Tesla Model S, and especially the ship computer in Star Trek (the voice interface, that is — not the LCARS GUI, which is a highly visible interface. Ahem!).

As we reduce a tool’s cog load, we tend to decrease its visibility. That is to say, the more users can forget they’re actually using a tool, the more it begins to feel like a natural extension of themselves. Which is why the closer we get to complete invisibility in our product, the better our experiences with those products tend to be.³

Golden Krishna sums it up quite well: the best interface is no interface.

Messaging isn’t an application, it’s an invisible interface

Conversation-driven product design is still nascent, but messaging-driven products are still represent massive growth and opportunity, expected to grow by another another billion users in the next two years alone.

For the next generation, Snapchat is the interface for communicating with friends visually, iMessage and Messenger is the interface for communicating with friends textually, and Slack is (or soon will be) the interface for communicating with colleagues about work. And that’s to say nothing of the nearly two billion users currently on WhatsApp, WeChat, and Line.

This didn’t happen by accident.

The only communication protocol humans share is the language — textual, and now increasingly visual — which is why the most invisible interfaces will usually be constructed in language, not UI.⁴

So, why bots?

As we move to increasingly invisible interfaces, I believe we’ll see a new class of messaging-centric platforms emerge alongside existing platforms in mobile, cloud, etc.

For consumers, the earliest and possibly most important example is the Messenger platform, while experiments like M hints at how Facebook is thinking about the future.⁵ For the workplace, the Slack platform will represent the hub of conversationally driven interactions.

As with every platform and interface paradigm, messaging has its own unique set of capabilities, limitations, and opportunities. That’s where bots come in. In the context of a conversation, bots are the primary mode for manifesting a machine interface.

Organizations will soon discover — yet again — that teams want to work the way they live, and we all live in messaging. Workflows will be retooled from the bottom-up to optimize around real-time, channel based, searchable, conversational interfaces.

Humans will always be the entities we desire talking to and collaborating with. But in the not too distant future, bots will be how things actually get done.


¹ VR will be another massive paradigm shift, but one deserving of a post all its own. I’ll get to that later!

² I say apparent because, in reality, there will always be cultural norms associated with the use of any human-computer interface, from language, to literacy, to mere passive exposure to technology. I’m not qualified to dive much deeper into it, but I also wanted to acknowledge the background cultural components of using digital tools.

³ This begs the question: is VR hyper-visible (because it’s the ultimate single-tasking medium, shutting out all other distraction), or completely invisible (because it’s totally immersive and feels real)?

⁴ And yes, email is definitely on the invisible end of the scale. Why do you think it’s been so hard to kill?

⁵ The fact Apple still hasn’t built a proper platform around iMessage strikes me as an unbelievably huge missed opportunity.

Thanks to Veronica Belmont, Amber Costley, Peter Rojas, Michael Galpert, Kristofer Joseph, and Brian LeRoux for reading drafts of this post.

Blatant self-promotion: want to see what I’m working on next? Drop your email here, and I’ll let you know! (0% chance of spam.)