Balancing Bottom-up and Top-down

A question I frequently encounter whenever I start a new information architecture project: What’s the right balance between defining an architecture from the top-down and leaving room for bottom-up structures to emerge?

The answer is different for each project. Some require a great deal of structure upfront, while others need just enough for bottom-up structures to emerge. The answer hinges on the type of environment we’re structuring, the team’s skills, the organization’s needs, and myriad other factors.

But what if this isn’t a dichotomy? What if we could design top-down structures broad enough (or perhaps deep enough) to allow for bottom-up organization to emerge organically? What architecture would lead to an environment that could be a receptacle for serendipity, happy accidents, improvisation — for humanity — while also accomplishing its business outcomes?

Suggested Searches in Apple Photos

One of the most amazing features enabled by machine learning algorithms is the ability to issue text searches on photos and images. Google Photos’s search abilities are one of its headline features. And Apple, too, has been working to improve search in the Photos app that comes with iPhones and other iOS devices.

As an iOS user, I’ve been watching Photos’s search functionality improve over the last couple of years. Although it’s a bit slower than I’d prefer, it’s still very useful. I can search by dates and common terms (e.g. “Halloween”) and often find what I’m looking for. However, sometimes the search yields no results at all — even when the term I’m searching for is a common word.

Recently I noticed a change in Photos’s search results UI that makes its operation more transparent:

Searching Apple Photos on the iPhone

What’s going on here? I’ve typed the word gorilla into the search box and Photos finds no results. (Yes, I do have photos of gorillas in my collection.) Rather than leave me with nothing, Photos offers to broaden the scope of the search. I’m offered two alternate searches:

gorilla → Mammal
gorilla → Elephant

There’s clearly some term mapping happening behind the scenes. That’s not unusual for search systems. What’s intriguing is how Apple has represented the mapping of terms, with the arrow pointing from my original search term to the suggested alternatives. Neither alternative is very useful to me in this particular case, but I understand why the system is suggesting these terms. (“Mammal” is a broader category of which gorillas are a member, and “elephant” is a sibling in that group.) That said, I appreciate the ability to change the scope of the search with one tap and the compact clarity of this UI.

TAOI: Disneyland App

The architecture of information:

Digital experiences are changing our understanding of physical environments. Google Maps gives you the ability to walk around a new city as though you’d known it for a long time. And should you develop a sudden hankering for ice cream, Yelp allows you to locate the nearest gelateria. The most noticeable change comes from layering information on the environment. For example, when trying to decide between two neighboring restaurants you’re no longer constrained to judging them solely by their appearance; you can also peruse their reviews in Yelp. Restaurant A has four-and-a-half stars, whereas restaurant B has three — A it is!

The number of stars is information about the place. You won’t find it in the physical place itself, but in its representation in an information environment which you access through your magical pocket-sized slab of glass. We’ve grown used to these augmented interactions with physical space, and mostly take them for granted. But recently I had one such interaction with an app I hadn’t used before, and which stood out to me for 1) its clarity of purpose and 2) the degree to which that purpose changed the experience of the place. I’m referring to the Disneyland app.

My family and I visited Disneyland a few weeks ago. We hadn’t been in five years, and the Disneyland app was one of the novelties since our last visit. The app presents a map of the Disney theme parks. As such it mostly replaced the parks’ old (and sometimes beautiful) paper-based maps. Thanks to the phone’s sensors, the Disneyland app makes it easy to figure out where you are, where to go next, and how to get there. But the app adds an additional key piece of information to the experience that can’t be had with paper-based maps: attraction wait times. Over every representation of an attraction in the park, you see a little callout that indicates how long you’ll have to wait in line to experience that ride or show:

Disneyland app

This piece of information is always available at all levels of zoom in the map. It’s the definitive element of the experience: in these maps, attraction wait times have the highest visual priority. As a result, wait times become the defining factor in sequencing the exploration of the park. The apps preferred answer to the question “What should we do next?” is always “Whatever is closest that has the shortest lines.”

This is an interesting choice that recalls the park’s old ticket levels. A long time ago, each Disneyland attraction required a separate ticket. Not all attractions used the same tickets; there were several levels ranging from A to E. “E-tickets,” such as the Haunted Mansion, were the most popular and desirable. These were considered the park’s premium attractions; their tickets were worth more than the others. This economic scheme influenced how visitors experienced the park. Ticket “coupon books” only included a limited number of E-tickets as compared to the lower denominations. Guests could buy more tickets inside the park, but having a limited number of the various level tickets affected choices. (I remember visiting Walt Disney World when it had a similar scheme, and hearing things like, “let’s visit this ride next, we have to use up our C-tickets.”)

The Disneyland app creates a similar economy by making attraction wait times the key informational element of the experience. When you’re trying to decide between two rides, knowing you’ll have to wait 65 minutes in line in one versus 15 minutes in another could be the key factor in your choice. (It was for my wife and me. Children get very cranky after waiting in long lines all day!) Our choosing to go on the ride with the lower wait times would contribute to slightly increasing that ride’s wait times and lowering the wait times for the more popular rides. I don’t have data, but my expectation is that this would help even out wait times throughout the park.

That is, of course, if all other things are equal — which they aren’t. The Haunted Mansion is a much more elaborate and compelling experience than Dumbo the Flying Elephant. Also, some rides have higher throughput than others. So the choice of riding one rather than the other doesn’t come down solely to which has the shortest waits.

That said, for someone like myself, who knows Disneyland very well, having this extra bit of information made the experience of visiting the park much better. In our two days at Disneyland, my family and I experienced more of the park than we’d ever been able to before. We also had more fun, since we spent a lower percentage of our time there in queues. But I wonder about the effect on folks who are less familiar with the parks. Will the emphasis on wait times drive them to prioritize less popular attractions over the park’s highlights? Adding feedback mechanisms to a system influences the way the system works. In what unexpected ways does this app change the experience of visiting Disneyland?

Neal Stephenson on Social Media

Speaking in an interview with Tyler Cowen, Neal Stephenson offers an excellent analysis of how social media has hurt civic discourse:

COWEN: You saw some of the downsides of social media earlier than most people did in Seveneves. It’s also in your new book, Fall. What’s the worst-case scenario for how social media evolved? And what’s the institutional failure? Why do many people think they’re screwing things up?

STEPHENSON: I think we’re actually living through the worst-case scenario right now, so look about you, and that’s what we’ve got. Our civil institutions were founded upon an assumption that people would be able to agree on what reality is, agree on facts, and that they would then make rational, good-faith decisions based on that. They might disagree as to how to interpret those facts or what their political philosophy was, but it was all founded on a shared understanding of reality.

And that’s now been dissolved out from under us, and we don’t have a mechanism to address that problem.

Mr. Stephenson’s observation corresponds to my experience of social media (especially Twitter): It’s not that folks are talking past each other, it’s that they’re not even interacting with people who don’t share their mental models. The mere hint of the possibility of an alternate take can lead to ostracism — or worse. Amplified through continuous validation and a complete lack of pushback, opinions replace facts as the basis for worldviews. To talk of filter bubbles is misleading: these aren’t tenuous membranes; they’re thick, hardened shells.

The interview continues:

COWEN: But what’s the fundamental problem there? Is it that decentralized communications media intrinsically fail because there are too many voices? Is there something about the particular structure of social media now?

STEPHENSON: The problem seems to be the fact that it’s algorithmically driven, and that there are not humans in the loop making decisions, making editorial, sort of curatorial decisions about what is going to be disseminated on those networks.

As such, it’s very easy for people who are acting in bad faith to game that system and produce whatever kind of depiction of reality best suits them. Sometimes that may be something that drives people in a particular direction politically, but there’s also just a completely nihilistic, let-it-all-burn kind of approach that some of these actors are taking, which is just to destroy people’s faith in any kind of information and create a kind of gridlock in which nobody can agree on anything.

In other words, it’s a structural problem. As such, it’s also systemic. Unmentioned in the interview is the driving force behind these algorithmic constructs: business models based on monetizing users’ attention. Incentivizing engagement leads to systems that produce fragmentation and conflict.

Neal Stephenson on Depictions of Reality (Ep. 71)

Are Your Navigation Structures Working?

One of the most important elements of a digital product is its navigation structures. By this, I mean the sets of links that allow users to move from one part of the environment to another. This includes global navigation bars, hamburger menus, contextual navigation blocks, etc. At its most basic, a well-designed navigation structure will help your users find the stuff they’re looking without fussing about. They’ll look at the choices before them, pick one, and find themselves faced with the information they’re looking for.

Achieving this best-case scenario is easier said than done; there are many ways to mess up navigation structures. I’ve experienced broken structures in major websites that leave me scratching my head. Often, the problem is that the structure reflects the organization’s conceptual model of its business without considering that the user may bring to the interaction a different mental model. Labels that may be clear to people in the organization can be ambiguous to outsiders.

Fortunately, it’s relatively easy to gauge​ the effectiveness of navigation structures. For example, studying site log files can reveal patterns of use that reveal navigational gaps. Observing users interacting with the environment (whether in production or using prototypes) is also a very effective way of revealing issues with navigation. Interviewing users is also valuable​ since it can help you understand​ how they see the domain. When examining the outcomes of such studies, you should ask questions such as:

  • Are we offering users the choices they expect?
  • Are choices labeled clearly? (By this I mean that they’re understandable to users.)
  • Are choices clearly distinct from one another? Or are some possibly ambiguous?
  • Are we offering users too much or too little choice?
  • Are choices presented at the right level of granularity for this part of the environment?
  • Is the set of choices helping bridge the user’s mental model of the business domain?

This last question is particularly important to me. Users bring to the interaction expectations of how the domain is organized. These expectations may or may not match reality. Situations in which users come to the experience with a clear, accurate understanding of the domain are relatively easy to deal with. But in some cases, users’ understanding may be off the mark. What then?

Navigation plays an important role not just in helping users move around, but also in educating them on the options available to them. Options in don’t exist in a void; people read them as sets. Users will understand the domain not through individual labels, but through the groups of choices they represent.

Defining these choices, the distinctions​ between them, and the labels that represent them​ is a design act that requires understanding the people who will be using them. In particular, understanding users’ mental models of the domain — and the degree to which those mental models differ from the organization’s conceptual models — is essential for designing navigation structures that work. This calls for research: understanding how folks think about your business’ domain, what they expect from you, and how they talk about those choices.

What I Unlearned From Architecture

I got an interesting question via Twitter:

“What were some of the mindsets, habits of thinking you had to unlearn transitioning from [architecture] to [information architecture]?”

The answer that comes immediately to mind is: “not that many!” I consider architecture a perfect training ground for information architecture. There are many more architectural mindsets that apply to information architecture than mindsets that require unlearning. That said, as I’ve thought about it I’ve realized there is, in fact,​ a mindset I picked up from architecture that I’ve had to unlearn over time: the idea of the architect as the central element of the design process.

Architecture is rife with what are referred to as starchitects — practitioners whose work and style is known around the world, and whose fame influences the work. Clients seek out Frank Gehry because they want a Frank Gehry building. Gehry’s office is bound to produce buildings in the Gehry style regardless of what contextual conditions call for.

When I was a student, most of the works we looked at were produced by starchitects. The implication was that that’s what we ought to aspire to. The first few years of my career, I labored under the delusion that I was at the center of the work. Over time, I came to realize that effective designers (in any field!) primarily serve not themselves or their architectural ideologies, but the work. I came to suspect the idea of having a “house style” — something I longed for at first.

To put it bluntly, I left architecture school with an inflated ego. The main mindset I had to unlearn as I transitioned to information architecture was the centrality of my own ideas, desires, and “style” in the design process. Instead, the core of what I aspire to now is form-context fit. This calls for understanding through collaboration; it calls for research and open-mindedness. Experience is primarily in service to the process, not the other way around. Getting my ego out of the way — embracing beginner’s mind — took many years of practice.

TAOI: The “Agony” of Browsing Amazon

The architecture of information:

As I’ve noted before, as software eats more of the world we’re likely to see more stories about how poor information architecture is hurting businesses. Along these lines, a recent opinion piece in Bloomberg claims browsing for products in Amazon is “agonizing”:

Try clicking through Amazon’s beauty products section to the pages for foundation, one of the (ahem) foundations of any makeup line. It’s a notoriously tricky item to buy online or in stores because people must match the shade to their skin tone. I found hunting for foundation on Amazon comically impossible.

There are 200 pages of products grouped seemingly with little reason. It’s not possible to narrow the product listings by liquid or powder foundation — equivalent to not giving people the choice between boxers or briefs.

Today is the second of two “Amazon Prime Days,” when the retailer presents many products at a deep discount. I’m in search of a new pair of sunglasses, so I’ve perused the site’s offers a couple of times since yesterday. The findings in the Bloomberg piece correspond with my experience: Amazon is relatively easy to use if you know what you’re looking for, less so if you’re browsing.

As the piece notes, it’s hard to state precisely how much (if any) this is hurting Amazon’s business. Speaking from my own experience, I find myself using other information environments as my first resort for browsing for products more often than not. (It’s not just bad IA; lately I also find Amazon’s UI slow to load and clumsy to operate.)

Amazon Shopping Is Easy. Browsing Is Agonizing.

Learn a Second Language

Do you want to become a better information architect? Learn a second language.

IA is focused on establishing distinctions. You do this with words. As a result, mastery of language is important for information architects. You master language by reading and writing — especially reading things that are outside your comfort zone. (One of the under-appreciated wonders of reading using tablets and e-readers is that they allow you to look up the definition words on the spot.) The broader your vocabulary, the more nuanced the distinctions you’re able to draw. (That said, you should avoid obscure terms when designing something for a mass audience. Not everybody will have as broad a vocabulary as you.)

But even having a broad vocabulary in one language may not be enough. Language is so foundational to how we experience reality that we can easily take it for granted. It’s the ground on which we stand. If we only know the one ground, we risk assuming everyone is standing on it. That isn’t the case.

Learning a new language forces you to realize that languages are constructs. Yes, they all have certain things in common. All languages have words for numbers, for example. But things like categorization schemes can vary significantly. Some languages have category terms that don’t exist in other languages. Some have more categories for a particular domain, others less. This video makes the point:

You can learn about these things intellectually. But you only grok the differences deeply when you must communicate with people who speak a different language. You start questioning things you’ve taken for granted most of your life, such as figures of speech and metaphors. You become aware of the historical contingencies of languages. None of the major ones have emerged fully formed; they’ve changed and influenced each other over time. And you, too, have the power to influence how they change.

Wittgenstein said that “the limits of my language are the limits of my world.” You must know the limits. This requires you to transcend them. Learning a second language — and putting yourself in a position to rely on it — pushes you beyond the limits of your mother tongue. A second language throws contrast, making the edges between distinctions visible. It’s an important skill for people who aspire to design worlds through words.

Information Metaphors

The ways we deal with information since the advent of the web are new. Although people have dealt with information in the past — through spoken language, print media, in the environment, etc. — the web changed how we produce and use information. We don’t yet have precise language to describe the effects of this change upon us as individuals and societies.

Language reveals how we think about things. Given the newness of the experience, I’m curious about the metaphors we use to talk about how we use of information online. I’ve noticed three come up often:

  • information as resource,
  • information as sustenance, and
  • information as an environment.

Let’s look at them in more detail.

Information as Resource

Under this metaphor, we see information as something to be bought, sold, mined, traded, shared, etc. We can own information, gain access to it, stream it. We must protect our information lest it fall into the wrong hands.


“A new commodity spawns a lucrative, fast-growing industry, prompting antitrust regulators to step in to restrain those who control its flow. A century ago, the resource in question was oil. Now similar concerns are being raised by the giants that deal in data, the oil of the digital era.” — The Economist

“Think twice about sharing your social security number with anyone, unless it’s your bank, a credit bureau, a company that wants to do a background check on you or some other entity that has to report to the IRS. If someone gets their hands on it and has information such your birth date and address they can steal your identity and take out credit cards and pile up other debt in your name.” — Christina DesMarais, TIME

Information as Sustenance

This metaphor posits that information is like food and drink; it changes us as we consume it. Information enters you and transforms you. You are what you eat; you are what you read online. As with food, you have the ability to say “no” to information, to change your consumption patterns. You could go on an “information diet” if you wished.


“We monitor what we eat and drink, optimizing our diet for health and performance, not just enjoyment–and yet we can be heedless about what we read, watch, and listen to. Our information diet is often the result of accident or happenstance rather than thoughtful planning. Even when we do choose deliberately, the intent behind much of our media consumption is simply to soothe or distract ourselves, not to nourish or enrich. It’s like having french fries for every meal.” — Ed Batista

“We define digital nutrition as two distinct but complementary behaviors. The first is the healthful consumption of digital assets, or any positive, purposeful content designed to alleviate emotional distress or maximize human potential, health, and happiness. The second behavior is smarter decision-making, aided by greater transparency around the composition and behavioral consequences of specific types of digital content.” —
Michael Phillips Moskowitz

Information as Environment

Another metaphor is that of information as something you inhabit; an environment. Under this metaphor, information defines the boundaries of spaces where we interact. We’ve been using this type of language from very early in the online revolution; we’ve been talking of “chat rooms” and “home pages” for a long time.


“When all discussion takes place under the eye of software, in a for-profit medium working to shape the participants’ behavior, it may not be possible to create the consensus and shared sense of reality that is a prerequisite for self-government. If that is true, then the move away from ambient privacy will be an irreversible change, because it will remove our ability to function as a democracy.” — Maciej Cegłowski

“Dark forests like newsletters and podcasts are growing areas of activity. As are other dark forests, like Slack channels, private Instagrams, invite-only message boards, text groups, Snapchat, WeChat, and on and on. This is where Facebook is pivoting with Groups (and trying to redefine what the word ‘privacy’ means in the process).

These are all spaces where depressurized conversation is possible because of their non-indexed, non-optimized, and non-gamified environments. The cultures of those spaces have more in common with the physical world than the internet.” — Yancey Strickler

While all three metaphors are valid, you won’t be surprised to learn I favor the “environment” metaphor — as evidenced by the title of my book.

The “resource” metaphor brings with it the language of ownership and trade. The “sustenance” metaphor reduces our agency to which types of information we choose to let in. (After all, most of us don’t produce our own food.) While both are valid, they miss an important angle: the fact that our interactions with each other and our social institutions are increasingly mediated through information. The language of inhabitation nudges us to consider the pervasive influence of information on our actions and empowers us to reconfigure our information structures to affect outcomes. It gives us agency with regards to information while acknowledging the degree to which it influences our decisions.

Have you found other information metaphors? Please let me know.