Returning to scenes of youth is always complicated business, the stuff that makes high school reunions emotionally fraught. How have I changed, how haven’t I, and how do I express those things when I come home? I know Brad Frost was sweating these topics as he toiled over his commencement speech at the high school where he graduated 14 years ago.
Months before he gave the talk, he told me he was already nervous about it. Turns out he didn’t need to worry. In fact, the “what has/hasn’t changed” anxiety turned out to be central to his wonderful speech. I especially loved this message:
The things you will be doing in 14 years’ time will
no doubt be different than the things you’re doing
at this phase in your life. A recent study by the Department
of Labor showed that 65% of students going through
the education system today will work in jobs that haven’t
been invented yet. Think about that. That means that
the majority of today’s students — probably including
the majority of this graduating class — will end up
working in jobs that don’t presently exist. Technology
is advancing at a staggering rate, it’s disrupting
industries, it’s inventing new ones, and it’s constantly
changing the way we live and work.
When I was a kid, I didn’t say “Mom, Dad, I want to be a web designer when I grow up!” That wasn’t a thing. And yet that’s now how I spend most of my waking hours, and how I earn my living, and how I provide for my family.
Our daughter Nika is about to start her final year of high school, and she sometimes worries that she doesn’t yet have enough vision for what she’ll become in her life and career—that she’s behind. But she knows what she loves, and she has so many talents, so we try to reassure her that knowing her skills, values, and personality is far more important than knowing a vocation. Vocations shift far more quickly than the rest.
When I graduated from high school (30 years ago next year!), the web hadn’t been invented, mainstream email was years away, and phones were cabled to the wall. But even then, I had a passion for both storytelling and systems—and those have been the guiding threads of a career spanning many kinds of jobs, culminating (for now at least!) in work shaping experiences of and for connected devices.
The great thing about returning to scenes of youth is that sometimes—like Brad—you get to talk to the kids coming up behind you. You can share the advice you’d offer your younger self. Nice work, Brad.
Video analytics provider Mux overhauled their data visualizations and shared the process. The thinking and results make for a worthwhile read, but I was especially taken with the project’s framing questions:
When we wanted to revisit our charts, we looked at
them from both of these perspectives and asked ourselves:
Are we being truthful to our calculations? And, are
we presenting the data in a beautiful and sensible
manner?
In a bit of unexpected good news, Google announced last week that it will stop reading Gmail messages for the purpose of showings ads tailored to the content.
The New York Times reminds, however, that Google will still continue to read your messages for other purposes:
It will continue to scan Gmail to screen for potential
spam or phishing attacks as well as offering suggestions
for automated replies to email.
So the bots will continue to comb your email missives, but at least they’re seemingly doing this in the service of features that serve the user (instead of advertisers).
As always, the tricky trade-off of machine-learning services is that they only know what we feed them. But what are they eating? It’s us. How much are we willing to reveal, and in exchange for what services?
To put a finer point on it, a pattern library’s value
to an organization is tied directly to how much—and
how easily—it’s used. That’s why I usually start by
talking with clients about the intended audiences for
their pattern library. Doing so helps us better understand
who’ll access the pattern library, why they’ll do so,
and how we can design those points of contact.
While a basic pattern library might consist simply of a collection of components, an essential pattern library is presented in a way that makes it not just the “right” thing to use but the easiest. For what it’s worth, this is really hard to do well; as we’ve found in our own projects, there’s some seriously demanding UX/research work behind tailoring a design system to the people who use it.
I don’t think that a design system should remove design
from your organization. A lot of clients think like,
“Oh, if we had a design system, we don’t need to design
anymore, right? ‘Cause all the decisions would be made.”
And I think that I’ve never seen that actually happen.
Design systems should just help you design better.
So you’re still gonna have to go through the process
of design, but a design system should be a good tool
in your arsenal for you to be able to design better.
And eliminate some of the useless decisions that you
might have to make otherwise.
I love this. Design systems are at their best when they simply gather the best practices of the organization—the settled solutions. The most effective design systems are simply containers of institutional knowledge: “this is what good design looks like in our company.” Instead of building (and rebuilding and rebuilding) the same pattern over and over again, designers and developers Instead of designing a card pattern for the 15th time, it’s already done. Product teams can put their time and smarts into solving new problems. Put another way: the most exciting design systems are boring.
Dan also talks about the importance of making the system fit the workflow of the people who use it. In our projects together, we’ve found that deep user research is required to get it right:
I’ve worked with organizations that the design systems are for purely developers. That’s it. It’s not for designers. And then in other organizations, a design system has to work equally well for designers and developers. And so just those two examples, those two design systems have to be drastically different from each other ’cause they need to support different use cases and different people.
If that kind of user-centered research sounds like product work, that’s because it is:
It is a product and you have to treat it like a product. It grows like a product, it gets used like a product. And for people to think like, “Yeah we’ll just create it once and then it’ll sit there.” They’re not realizing the value of it and they may not actually get value from it if they’re not treating it like a thing that also needs to grow in time and adapt over time. … The reason that the design system is so hard to get off the ground is because it requires organizational change. It requires people to have different mindsets about how they’re going to work.
If smart machines and AI agents are meant to support human goals, how might we help them better understand human needs and behaviors? UC Berkeley professor and AI scientist Anca Dragan suggests we need a more human-centered approach in our algorithms:
Though we’re building these agents to help and support
humans, we haven’t been very good at telling these
agents how humans actually factor in. We make them
treat people like any other part of the world. For
instance, autonomous cars treat pedestrians, human-driven
vehicles, rolling balls, and plastic bags blowing down
the street as moving obstacles to be avoided. But people
are not just a regular part of the world. They are
people! And as people (unlike balls or plastic bags),
they act according to decisions that they make. AI
agents need to explicitly understand and account for
these decisions in order for them to actually do well. […]
How do we tell a robot what it should
strive to achieve? As researchers, we assume we’ll
just be able to write a suitable reward function for
a given problem. This leads to unexpected side effects,
though, as the agent gets better at optimizing for
the reward function, especially if the reward function
doesn’t fully account for the needs of the people the
robot is helping. What we really want is for these
agents to optimize for whatever is best for people.
To do this, we can’t have a single AI researcher designate
a reward function ahead of time and take that for granted.
Instead, the agent needs to work interactively with
people to figure out what the right reward function
is. […]
Supporting people is not an after-fix
for AI, it’s the goal.
To do this well, I believe this next generation should
be more diverse than the current one. I actually wonder
to what extent it was the lack of diversity in mindsets
and backgrounds that got us on a non-human-centered
track for AI in the first place.
In a wonderful interview with Backchannel, Google Cloud’s chief AI scientist Fei-Fei Li and Microsoft alum and philanthropist Melinda Gates press for more diversity in the artificial intelligence field. Among other projects, Li and Gates have launched AI4ALL, an educational nonprofit working to increase diversity and inclusion in artificial intelligence.
Fei-Fei Li: As an educator, as a woman, as
a woman of color, as a mother, I’m increasingly worried.
AI is about to make the biggest changes to humanity,
and we’re missing a whole generation of diverse technologists
and leaders.…
Melinda Gates: If we don’t get women
and people of color at the table — real technologists
doing the real work — we will bias systems. Trying
to reverse that a decade or two from now will be so
much more difficult, if not close to impossible. This
is the time to get women and diverse voices in so that
we build it properly, right?
This reminds me, too, of the call of anthropologist Genevieve Bell to look beyond even technologists to craft this emerging world of machine learning. “If we’re talking about a technology that is as potentially pervasive as this one and as potentially close to us as human beings,” Bell said, “I want more philosophers and psychologists and poets and artists and politicians and anthropologists and social scientists and critics of art.”
We need every perspective we can get. A hurdle: fancy new technologies often come with the damaging misperception that they’re accessible only to an elite few. I love how Gates simply dismisses that impression:
You can learn AI. And you can learn how to be part of the industry. Go find somebody who can explain things to you. If you’re at all interested, lean in and find somebody who can teach you."… I think sometimes when you hear a big technologist talking about AI, you think, “Oh, only he could do it.” No. Everybody can be part of it.
It’s true. Get after it. (If you’re a designer wondering how you fit in, I have some ideas about that: Design in the Era of the Algorithm.)
Machine learning is trying to one-up just-in-time inventory with what can only be called before-it’s-time inventory. The Economist reports that German online merchant Otto is using algorithms to predict what you’ll order a week before you order it, reducing surplus stock and speeding deliveries:
A deep-learning algorithm, which was originally designed
for particle-physics experiments at the CERN laboratory
in Geneva, does the heavy lifting. It analyses around
3bn past transactions and 200 variables (such as past
sales, searches on Otto’s site and weather information)
to predict what customers will buy a week before they
order.
The AI system has proved so reliable—it predicts with
90% accuracy what will be sold within 30 days—that
Otto allows it automatically to purchase around 200,000
items a month from third-party brands with no human
intervention. It would be impossible for a person to
scrutinise the variety of products, colours and sizes
that the machine orders. Online retailing is a natural
place for machine-learning technology, notes Nathan
Benaich, an investor in AI.
Overall, the surplus stock that Otto must hold has
declined by a fifth. The new AI system has reduced
product returns by more than 2m items a year. Customers
get their items sooner, which improves retention over
time, and the technology also benefits the environment,
because fewer packages get dispatched to begin with,
or sent back.
There’s much effort afoot to make the bots sound less… robotic. Amazon recently enhanced its Speech Synthesis Markup Language to give Alexa a more human range of expression. SSML now lets Alexa whisper, pause, bleep expletives, and vary the speed, volume, emphasis, and pitch of its speech.
This all comes on the heels of Amazon’s February release of so-called speechcons (like emoticons, get it?) meant to add some color to Alexa’s speech. These are phrases like “zoinks,” “yowza,” “read ’em and weep,” “oh brother,” and even “neener neener,” all pre-rendered with maximum inflection. (Still waiting on “whaboom” here.)
The effort is intended to make Alexa feel less transactional and, well, more human. Writing for Wired, however, Elizabeth Stinson considers whether human personality is really what we want from our bots—or whether it’s just unhelpful misdirection.
“If Alexa starts saying things like hmm and well, you’re
going to say things like that back to her,” says Alan
Black, a computer scientist at Carnegie Mellon who
helped pioneer the use of speech synthesis markup tags
in the 1990s. Humans tend to mimic conversational styles;
make a digital assistant too casual, and people will
reciprocate. “The cost of that is the assistant might
not recognize what the user’s saying,” Black says.
A voice assistant’s personality improving at the expense
of its function is a tradeoff that user interface designers
increasingly will wrestle with. "Do we want a
personality to talk to or do we want a utility to give
us information? I think in a lot of cases we want a
utility to give us information,” says John Jones, who
designs chatbots at the global design consultancy Fjord.
Just because Alexa can drop colloquialisms and pop
culture references doesn’t mean it should. Sometimes
you simply want efficiency. A digital assistant should
meet a direct command with a short reply, or perhaps
silence—not booyah! (Another speechcon Amazon added.)
Personality
and utility aren’t mutually exclusive, though. You’ve
probably heard the design maxim form should follow
function. Alexa has no physical form to speak of, but
its purpose should inform its persona. But the comprehension
skills of digital assistants remain too rudimentary
to bridge these two ideals. “If the speech is very
humanlike, it might lead users to think that all of
the other aspects of the technology are very good as
well,” says Michael McTear, coauthor of The Conversational
Interface. The wider the gap between how an assistant
sounds and what it can do, the greater the distance
between its abilities and what users expect from it.
When designing within the constraints of any system, the goal should be to channel user expectations and behavior to match the actual capabilities of the system. The risk of adding too much personality is that it will create an expectation/behavior mismatch. Zoinks!
Mike Loukides describes a fundamental weirdness in creating predictive algorithms: in order to make them flexible enough to deal with real-world data, you also have to make them imperfect.
Building a system that’s 100% accurate on training
data is a problem that’s well known to data scientists:
it’s called overfitting. It’s an easy and tempting
mistake to make, regardless of the technology you’re
using. Give me any set of points (stock market prices,
daily rainfall, whatever; I don’t care what they represent),
and I can find an equation that will pass through them
all. Does that equation say anything at all about the
next point you give me? Does it tell me how to invest
or what raingear to buy? No—all my equation has done
is "memorize" the sample data. Data only
has predictive value if the match between the predictor
and the data isn’t perfect. You’ll be much better off
getting out a ruler and eyeballing the straight line
that comes closest to fitting.
If a usable machine learning system can’t identify
the training data perfectly, what does that say about
its performance on real-world data? It’s also going
to be imperfect. How imperfect? That depends on the
application. 90–95% accuracy is achievable in many
applications, maybe even 99%, but never 100%. …
The right question to ask isn’t how to make an error-free system; it’s how much error you’re willing to tolerate, and how much you’re willing to pay to reduce errors to that level.
If errors are inevitable, then the job of design is to present the data in ways that set appropriate expectations. The more I ponder the future of UX in a machine-learning world, the more I’m convinced of this: large swaths of the UX discipline will revolve around presenting data in ways that anticipate the machines’ occasionally odd, strange, and just-plain-wrong pronouncements.
Dan Hon imagines what happens when all of a city’s systems are connected, incentivized, gamified, and bureaucratically weaponized:
I was late paying the water bill, so the parking meter refused service until I coughed up.
The meter said I had 30 seconds to pay the water bill until I had to move my car, and… I just froze. Then the meter attendant came. She said she was just doing her job as she booted my car, then looked down at her phone. Reminded me I hadn’t taken out my recycling.
This wasn’t turning out to be a good day.
She told me I was on my second strike: one more, and I’d lose streetlight privileges. I’d heard about that: a social shaming punishment. Streetlights would create a cone of darkness around just you.
The implicit argument is that when humans draw, they
make abstractions of the world. They sketch the generalized
concept of “pig,” not any particular animal. That is
to say, there is a connection between how our brains
store “pigness” and how we draw pigs. Learn how to
draw pigs and maybe you learn something about the human
ability to synthesize pigness. …
What can SketchRNN learn? Below is a network trained on firetrucks generating new fire trucks. Inside the model, there is a variable called “temperature,” which allows the researchers to crank the randomness of the output up or down. In the following images, bluer images have the temperature turned down, redder ones are “hotter.”
[…]
What [project leader Doug] Eck finds fascinating about sketches is that they contain so much with so little information. “You draw a smiley face and it’s just a few strokes,” he said, strokes that look nothing like the pixel-by-pixel photographic representation of a face. And yet any 3-year-old could tell you a face was a face, and if it was happy or sad. Eck sees it as a kind of compression, an encoding that SketchRNN decodes and then can re-encode at will.
In other words, sketches might teach AI portable, human-understandable symbols of abstract concepts—a shorthand description of the world. It strikes me that all creative pursuits, including design and language, traffic in similar symbols and shorthands. I’m impatient to find out how this particular branch of AI develops to understand (and create) the interfaces and interactions that designers make on an ongoing basis.
At the moment, this is the stuff of the research lab. But other flavors are starting to emerge in consumer products, too. Apple has been training iOS to anticipate strokes in sketches and handwriting in make writing with Apple Pencil seem buttery smooth. In Buzzfeed’s overview of iPad updates, John Paczkowski reports:
Meanwhile, the Apple Pencil’s latency — that slight lag you get
when drawing — has been reduced to the point where
it’s virtually imperceptible; Apple says it’s just
20 milliseconds. And since Apple is so intensely focused
on capturing the experience of putting pen to paper,
it’s doing additional work in the background to remove
the lag entirely with machine learning–based algorithms
designed to predict where a Pencil is headed next.
“We actually schedule the next frame for where we think
the Pencil’s going to be, so it draws it right when
you get there, instead of right after you have been
there,” Schiller says.
While Google and SketchRNN are chasing the lofty goal of understanding how humans communicate in symbols, Apple is meanwhile learning the commonplace but useful skill of learning how you write and draw. Machines may not yet be capable of their own creative works, but they’re already beginning to learn to understand and anticipate our own.
Brad Frost nailed down a pair of mighty useful metaphors for pattern libraries (“workshops”) and design-system style guides (“storefronts”). These are useful because they help delineate what kind of work happens where, which is a recurring source of confusion weâve seen in companies struggling to maintain jumbo design-system projects.
The workshop
Brad created the first version of Pattern Lab when we designed the Techcrunch website back in 2013. That pattern-library software was the first glimpse of his Atomic Design methodology, building design pattern “organisms” out of smaller “atoms” and “molecules.” Pattern Lab has since been open-sourced and remains our go-to tool for developing and sharing websites and full-blown design systems.
Pattern Lab is where all our work comes together. It’s a collaborative environment where information architecture gets stubbed out in the browser, where visual design comes together, where the code gets wrangled, and where content is edited. We share ongoing work inside Pattern Lab with stakeholders and clients. And it’s the final deliverable for web projects, complete with page templates and detailed pattern library. Our projects happen almost entirely inside Pattern Lab.
Our pattern libraries are always a wonderful mess, full of experiments and spare parts and tools.
Brad explains:
While Pattern Lab shares some qualities with style guides (for instance, it shows code snippets and you can add pattern documentation), the environment is really designed for teams to effectively build and work with UI components: the navigation across the top is small and unobtrusive, there are viewport resizing tools to stress-test UI components and pages, weâre able to organize components in a way that makes sense to us as creators (such as using the atomic design methodology), and we can design with dynamic data to ensure patterns are robust, resilient, and serve the needs of the organizationâs applications. Like my wifeâs jewelry workshop, the environment is designed for the design system team to be productive and creative.
This is not, however, an especially friendly place for people outside the working production team. Its organization around atoms, molecules, and organisms isn’t relevant to others; it contains building-block patterns that don’t have much useful meaning on their own; and it contains work-in-progress experiments that aren’t ready for prime time.
So when you’re sharing polished patterns and design systems with a group beyond the production team, you need something more refined. You need…
The storefront
If the pattern library is the workshop, then the design-system style guide is the storefront, as Brad explains:
A style guide is the storefront where all the ingredients
of the design system are put out on the shelves. The
style guide storefront is designed for a different
context than the design/dev environment workshop. Rather
than being a tool for only the design systems team
to make use of, the style guide communicates the design
system to the whole organization. That means the style
guide audience should be cross-disciplinary, since
a design system can help create a shared vocabulary between
all the people who are responsible for the success
of the products at the organization. The style guide
should provide information helpful for both makers
and users of the design system, and should be used
as a vehicle to continuously sell the value of the
design system to the organization.
This isn’t just a difference in presentation. There’s a difference in core content, too. In our recent projects building out gigantic enterprise design systems, we’ve found that the style guide always presents only a subset of the pattern library. We cherrypick the polished patterns that are ready to share, while excluding most of the experiments and building-block “atoms.”
Behind the scenes, we use Brad’s Style Guide Guide to import selected patterns and templates from Pattern Lab, and then display them in a polished website. (In practice, a simple Grunt task exports the HTML for all patterns, and then copies âem into the style guide directory. Style Guide Guide includes them automagically in pages when it builds its website.)
From there, we add lots of guidelines and documentation to help newcomers make sense of the UX, the visual design, and the underlying markup. The end result is a set of settled solutions for common problems, clearly understandable and ready for production.
Distinct places for distinct jobs
We build patterns in the workshop, and we display the best of them in the storefront, showcasing them in the best possible light.
Too often, though, we see organizations try to force everything into one place. We see workshop pattern libraries trying to do double duty as a canonical design-system reference. Or on the other side, we see pattern libraries set up as static storefront references that live outside a useful working development environment. When these resources set up outside the workflow of designers and developers, they donât get used, they get stale, they become irrelevant.
Following the workshop/storefront modelâand stitching the two together so that one feeds the otherâhas ensured that the design systems we create continue to be used, vital, dynamic.
Joy Buolamwini demonstrates “the coded gaze,” which recognizes her face only when she wears a white mask.
Joy Buolamwini on on a tear lately. The founder of the Algorithmic Justice League has received well-deserved press from the likes of the BBC and Guardian for her campaign to uncover inadvertent bias in machine-learning algorithms.
At Hackernoon, Buolamwini responds to criticism she received after demonstrating that facial recognition often breaks down for people of color. (Buolamwini, a woman of color, had to put on a white mask before one algorithm would even detect a face.) Some have told Buolamwini that it’s not the algorithm’s fault but rather that cameras are poor at discerning black faces: “Algorithms aren’t racist,” the argument goes. “Your skin is just too dark.”
Good lord. The problem is not with “photography.” If your eye can discern difference, the camera can, too. It’s true that camera technology has historically favored light skin. But that’s less a factor of underlying technology than of the skewed market forces and customer base that shaped early photography. In other words: it was a miserable design decision. For decades, for example, Kodak’s development process for color film was calibrated to photos called “Shirley cards” (named after the first model to pose for them). Shirley cards reflected a decidedly white concept of beauty. “In the early days, all of them were white and often tagged with the word ‘normal,’” NPR reported.
Now we’re carrying this original bias into the machine-learning era. Machine learning excels at determining what’s “normal” and trying to replicate itâor discard outliers. What the machines think is normal depends entirely on the data we feed their models. As the era of the algorithm begins to embrace the whole broad world, it’s urgent that we examine what “normal” really is and work to avoid propagating exclusionary notions of the past by encoding them into our models.
Instead of doing the hard work of creating truly inclusive algorithms, however, some suggest that Buolamwini should instead carry a lighting kit with her:
More than a few observers have recommended that instead
of pointing out failures, I should simply make sure
I use additional lighting. Silence is not the answer.
The suggestion to get more lights to increase illumination
in an already lit room is a stop gap solution. Suggesting
people with dark skin keep extra lights around to better
illuminate themselves misses the point.
Should we change ourselves to fit technology or make
technology that fits us?
Who has to take extra steps to make technology work?
Who are the default settings optimized for?
As always with emerging technologies, our challenge is making tech bend to our lives instead of the reverse. It’s profoundly unfair to make some lives bend more than others.
For designers, the arrival of the algorithm era introduces UX research challenges at an unprecedented scale. A big emerging job of design is to help identify where the prevailing definition of “normal” is flawed, and then move heaven and earth to make sure the data models embrace a new, more inclusive definition of normal. That is where we need to add more light.
Designer Anton Sten ponders the future role of digital designers in a world of more and more Alexas, Siris, and other non-visual interfaces. His conclusion is that much more of our work will be about designing for what goes wrong:
As technology offers us more and more options and possibilities,
our work as UX-designers will grow to include even
more edge-cases. As our acceptance of friction with
these services continues to decrease, our work will
increasingly need to include more ‘what if’ scenarios.
I agree, and I’m excited about this. Instead of etching buttons and controls for flows that we wholly control, I see our work evolving into the anticipation of scenarios that spin out of machine-generated content and interaction.
How will we handle the weird, the odd, the unexpected, and the wrong? These are exciting challenges, and they mean designing the experience and expectations around the interaction that the machines themselves create. Among other things, we have to help systems be smart enough to know when they’re not smart enough.
So happy and proud for my studiomate Lori Richmond, a marvelous illustrator who’s also become an impressive runner over the last several months. Only a year after taking up running, Lori was just featured in Runner’s World—for her inspired illustration project.
Here’s the concept: after every training run, Lori draws or paints a scene she saw. And she executes it in exactly the time it took her to finish that run.
Lori’s been posting all of these at her @loririchmonddraws Instagram account. And Runner’s World took notice, which is super-fun recognition for a new runner who’s already logged four half marathons. (Her goal is to run a half marathon in each of New York’s five boroughs.)
“I got kicked out of gym in the 6th grade because I was SO un-athletic,” Lori posted in our studio Slack channel. “The art kids will always come back for you!!”
The Style Guide Guide imports and displays design-pattern HTML from a separate pattern library. Anytime there’s a change in the underlying code of the patterns, the style guide picks it up—always up to date. From there, The Style Guide Guide mixes in your documentation, usage guidelines, and design principles. Because those are entered in Markdown, it’s easy for a whole team to contribute documentation and guidelines, with a very low technical barrier to entry
Brad, Dan, Ian, and I have been using The Style Guide Guide alongside Pattern Lab in our last few design-system projects. It’s proven to be a highly collaborative environment for creating and sharing a design system. (Stay tuned, Brad promises a blog post to walk you through the integration with Pattern Lab, very cool.)