Prosaic Times: Let’s not have software that looks like Tyson’s Corner

AI may make it a little easier to make technology beautiful

Let’s make technology beautiful — or at least less ugly.

Le Corbusier said, “A house is a machine for living in.” He also wanted to dynamite the center of Paris and turn it into something that looks like Tyson’s Corner, Virginia. I’ve spent some memorable days (and nights) in conference rooms in Tyson’s Corner, but I hope nobody believes it should be an architectural aspiration.

Would you rather live in Frank Lloyd Wright’s Robie House or Corbusier’s Villa Stein? (Even if you’ve never heard of either, I promise you it’s the Robie house.) Also Corbusier built lousy machines. Flat (rather than sloped) roofs retain rainwater and eventually leak.

Which would you rather have in your neighborhood — one of Chicago’s art deco skyscrapers (photo taken last spring) or basically anything penned by Frank Gehry?

Far better to remember that Pininfarina’s Cisitalia is part of the MoMA’s permanent collection and try to create beautiful machines.

So how do you create technology that’s more Art Deco and less Frank Gehry?

1. Read Edward Tufte’s “Visual Display of Quantitative Information.” Nobody who can’t or doesn’t quote from this book should be allowed anywhere near a UX.

2. Think personas as much as journeys. Yes journeys matter. But users will employ the systems we build in new and wonderful ways we would not have contemplated. In contrast Gehry’s buildings are about Gehry, not the people who live or work there.

3. Focus on structure. As Tufte points out, humans can process massive amounts of information — if it fits into a logical structure. Think of applications you use. How many help you find functionality easily through logical groupings? How many have random assemblages of stuff?

4. Don’t fixate on the least sophisticated users — this often leads to “let’s just put the most often used features up front,” which creates a confusing jumble.

5. Manage risk, rather than eliminate it. Everybody will think of every bad thing that can happen. That’s good: threat modeling is important. But you also have to take likelihood and impact of “user friction” into account.

6. Focus on the small things. Mies van der Rohe sweated the details much harder than the Mies-Lings. Even more so, the details in Art Deco enrich the soul. Would you rather walk through the halls of Rockefeller Center or Lincoln Center? (Amy and I saw “The Amazing Adventures of Kavalier and Klay” at Lincoln Center recently — the production was much more inspiring than the venue!)

7. Ask the cats about the cat food after you serve it. It never ceases to amaze me how little time architects spend talking to users after systems go live.

We can all recite the reasons for software that deadens the soul with its ugliness. Too little insight into what users want. Too little design expertise, especially in corporate technology environments. Too many requirements and too much information to convey. Too little time and not enough funding. Ship the release, tick some items off the backlog and move onto the next sprint. The result: systems that users can’t or won’t get value from.

We may, with enough thought and care in how we apply AI, see an era where we can transcend long-assumed constraints on enterprise technology.

• Yes, business users are pretty terrible at communicating the experiences they want. We can now track and interrogate data about user experience and feedback more economically than ever before. Will your organization invest in capturing this insight so they have a better chance of delighting their users?

• Will shops that can’t employ a stable of designers use AI to mock up more attractive interfaces? No, the early experiences of desktop publishing weren’t pretty, but we can now use prompting to enforce design guidelines.

• We’ve all asked ourselves: do I put this extra widget on the home screen or, in the interest of a cleaner design, do I require an extra click to get to it. Engineers can now build composable UIs that expose the right screen element to the right user at the right time. Will your engineers master this skill set — and will they leverage GenAI-enabled chat functionality so that users can just ask for what they want?

• Implemented appropriately, AI-enabled software engineering can double throughput. Will your organization reinvest some of this windfall in more beautiful user experiences?

As in so many areas recently, CIOs, CTOs and other enterprise technology leaders have a choice in front of them here — will they make technology beautiful or not?