Say “Ok Library”

Smartphones didn’t capitalize on that whole “library without walls” promise like we expected. Instead, the library as a place became more tangible and responsive than ever. Our user-centric push to be at our patrons’ point of need — especially through mobile-first responsive web design — reinforces precisely what makes libraries hyperlocal. Our potential to interact through geolocation, beacons, cameras, and context is a pretty strange and newfangled opportunity — especially since an internet-ready device is predictably on almost every person — to iron out the kinks that complicate how patrons interface with the library.

I have a favorite hypothetical when I talk about designing around the internet of things. Pretend you happen to wander into the range of a beacon and you get this text from the library:

Hey, it’s pretty chilly this morning. We just made some coffee. If you have time, swing by and have a cup.

Cool, huh? This isn’t vaporware. Let’s walk through this. First, a user comes in range of a beacon or GPS coordinates, which triggers a notification. Second, an opening sentence based off the time of day and the weather is puzzled together: “Hey, it’s pretty chilly this morning.” Then, third, the app makes a time- and weather-relevant suggestion to invite you into the building.

This alone is pretty nifty. Were this suggestion actionable and you swiped to confirm and tapped through an order, then the library just compelled you on a journey with the promise that each touchpoint is pleasant. At this point, it’s on the library to deliver.

Even then, wouldn’t it be easier if all you had to do was talk back?

Good idea. I’ll be there in ten. I take my coffee black.

A Voice User Interface

I read Design for Voice Interfaces by Laura Klein and I am having fun thinking about the library application for voice user interfaces (VUI). Voice recognition’s been around but it hasn’t quite lived-up to the dream (yet) fanned through scifi. We’re getting there, though.

Some products in 1999 had around a 65 percent recognition rate, whereas today’s rates are closer to 92 percent. ( Klein, p. 27)

Good speech recognition has unique design challenges in that unlike other visual interfaces where the way we interact has precedence in physical objects — we push buttons, we turn pages and move things around in our hands, sometimes there’s even haptic feedback — there is no analog for VUI other than human conversation.

Skeuomorphic material design uses depth and animation to feel like shuffling paper around on a desk. For Siri and Cortana to succeed they need to mimic consciousnesses who can parse complex meaning in human speech. The more artificial they seem by clearly misunderstanding, then the poorer the user experience and the greater the failure.

We need to practice contextually aware design. If I say, ‘Make it warmer’ in my house, something should know if I mean the toast or the temperature. So, successful voice interfaces must understand context. (Thomas Hebner, quoted in Klein, p. 6)

So, yeah, context is key. In “Does the best library web design eliminate choice?” I wrote about how services that anticipate your needs are built on three (or so) pillars: context, behavior, and personal data.

Context can be inferred from neutral information such as the time of day, business hours, weather, events, holidays. If the user opts in, then information such as device orientation, location or location in relation to another, and motion from the accelerometer can build a vague story about that patron and make educated guesses about which content to present — like our cup-of-coffee example. Behavior can be deduced using analytics for broad site-wide usage as well as user behavior specific to an individual through the use of cookies and other utilities.

However, it can only be a user experience for one when personal data can expose real preferences — Michael loves science fiction, westerns, prefers Overdrive audiobooks to other vendors and formats — to automatically skip the hunt-and-peck and tailor a unique service.

Unlike forms and other visual interfaces, the input for voice can be loaded with meaning that must buffer through a ton of logic. Whether your VUI has to open an app, search the web, text a friend, place a hold, tell you about upcoming events (but not all 23 — that’d be too much), is determined by what it knows about you. That’s why Siri is a “personal assistant:” she knows your agenda and watches you sleep.

An uncanny valley of personalization can be overcome if the voice interface is useful enough.

For libraries, that killer app might be real close to home: our bread and butter — complex search.

Ok Library: I need a recent full-text article about design patterns that’s been peer reviewed.

Footnotes

• I used Aaron Schmidt’s redesigned library-guy graphic as the logo in the “Ok Library” mockup from his post, “National Library Symbol History & Implications.”
• The card in the same mockup, which reads “Stories of Strength,” is from a screenshot from the Alvin Sherman Library’s Lists site, where librarians make lists (which is what librarians like to do).

Originally published at LibUX.