/ blog

Designing for the limits of attention

Our digital ecosystem is bursting with features, interaction patterns, and graphics, but are they adding much to our lives other than noise? Here I explore the limitations of interface design and the need for a new approach that considers human limitations as it keeps scaling.


When it comes to products with an established system, new features can be added by expanding on existing UI patterns or introducing new ones, such as a new button or tab. The sparkle/magic button that landed everywhere with new AI features is a good example. "Supercharging interfaces" as it's been called, is an unopinionated way to integrate new possibilities for the user. This creates a fair bit of noise over the long term and restricts the potential for a paradigm shift. It requires a great deal of effort to maintain and expand a system in this manner over time.

However, everything eventually progresses beyond just adding more of the same. A classic example of this idea is the invention of the car, which was not simply an attempt to create faster horses. That's where abstraction comes into play.


In contrast, some interaction patterns are so complex or cluttered that there is greater value in reducing fidelity to promote a simpler experience. The loss in fidelity of the control is offset by the ease and result: you get 90% or more of the desired outcome while reducing. Imagine a UI that seamlessly anticipates your needs - That's the vision that a lot of AI products are selling. It's the latest spin on the "less is more" motto.

Cameras are a great way to illustrate this. With the arrival of autofocus and machine learning, most users have agreed to trade an easy snap for less control. The big circle shutter button on our phone cameras is a lot easier than dialing everything manually on a DSLR.


At some point, everything evolves. Internet browsers seem to be ripe for a paradigm shift like cameras were 10 years ago. Too many widgets and ambitious concepts have left me unsatisfied with the state of things (Arc browser is one of the latest examples). We all have too many tabs open and some unhealthy digital habits around our browsers. Minimalism and good hygiene won't solve the problem at scale. So many facets of our lives depend on it. There are complex technical implications I'm clueless about but from an interaction perspective, the model is showing its limit with the arrival of spatial computing and AI. Do we want to project a good old rectangular window frame in augmented reality? Are we ready for some AI chatbot popups everywhere? I hope not.

Design systems

On a web design level, what is inside of each tab, we are also reaching a point of saturation. Some have speculated about a unified design system (like Brad Frost here, without offering much perspective other than an elegant problem statement) as a response to the challenges of scaling design systems and the sheer cacophony of the www. The web has been a reflection of our globalized economy where liberal capitalism patterns have echoed. I doubt that unification or even a standardized system would ever shape up, should we even agree on how it would look like.

Although liberalism often incentivizes commoditization, maybe I’ll be proven wrong. I agree with Brad on the potential for great global public utility... and great power to the maintainers. At the moment everyone is doing their own thing. Some larger organizations with more resources are incrementally improving standards along with Open Source projects, coming and fading.

Human limitations

Design systems provide a great example of structures built upon existing structures, and how they test the cognitive abilities of organizations to maintain at scale. Everything is built on top of something else.

It's fascinating to consider how technology advances alongside humans. In essence, digital tools allow us to handle more information than our brains can process and manage. The digital ecosystem seems to be reaching a watershed moment where its complexity is now offset by the value it provides.

Nowadays, decisions are increasingly influenced by human limitations rather than resource constraints: language and cognitive capacity. From the minimalists of the indie web purists to the venture-funded technocrats, I perceive a collective longing for ways to address these human challenges in the next generation of interfaces.

The list

The Vanilla web is my web-design-centric checklist aiming at addressing current challenges while considering the latest developments in tech. I agree with Steph, there is nothing boring about vanilla. It's a very high standard. Listed below are a couple of items I'd suggest to keep in mind when designing (I don't call them "principles", it sounds pretentious).

  • Design for attention: Create contexts, not screens, pages, or stories. Content > Context > UX. Narrow the focus, and dial down the layout. Lower fidelity. Reduce noise. 3 variables per context is most people's limits.
  • Bake in semantics: How it looks is how it is structured. Design, engineering, and content converge here, not with design engineers. Less tabs, labels, accordeons, filters... just search, or better, nothing.
  • Innovate on familiar ground: Aim for 80% existing pattern, 20% new. Acknowledge that less is more, but more is needed. Scale means debt for the maintainer, and should map to user value.
  • Make it kind on the eyes: You need a mighty good reason to animate something. If you do it should feel snappy, not bouncy. Mechanical and smooth. Fewer sticky things.
  • Incentivize autonomy: Don't make things too easy, ask for permission. By default, nothing should be automated or personalized. Favor curation. Bring RSS back!
  • Be device agnostic: One column is all that is needed. Streamline responsive behavior. There are too many screen sizes for a coherent breakpoint logic. Most things should collapse. The ultimate goal: Mobile = everything else.
  • Keep it real: The user is tired and not thrilled about the design. Drop SEO and growth marketing "best practices". Nobody likes to fill out forms. The fold doesn't matter and people read when they care...
  • Learn from mistakes: All tools are built for a purpose. Make it obvious, and stick to it. Resist designing dashboards, feeds, comment boxes… ditch all those patterns that have made the worst of today’s internet.
  • Make it personal: Data, UX insights, trends, bureaucracy, fatigue, reviews... the process is grueling. So much stuff out there has no soul. Users can feel that. Good tools have an extra something.

Caveat: predicting the future is a fool's game.
Reframing problems is key to progress.
If Brad Frost can get away with it, maybe I can too.

So much work.
So little progress.
That's how it has always been.
And will ever be.

← Index / Published on 2024-06-26