2014 Keynote by Sumana Harihareswara
Thank you so much for having me.
This community is super thoughtful and I've avidly read your posts for years; you set a high bar for insight and achievement in thinking about information, skillsharing, and service.
I work for the Wikimedia Foundation in the Engineering Community Team, and I help people integrate with our API for Wikipedia, Wikidata, Wikivoyage, Wiktionary, and all our other sites, and it's really exciting to me that we're finally moving towards a service-oriented architecture and away from the SCM, Spaghetti Code Model. But I won't really be talking too much about that today. I'm happy to talk about it and I'm here for the whole conference so please do come up to me if you want to talk about it.
For a few years I've helped run internship programs that bring new people into the community of practice that is open source, and I've done my share of mentorship, and I also want to talk with people about that this week, but not right now.
And last year I took a sabbatical and I went to Hacker School, in New York City, to improve my programming skills, and I found a tremendously educational and transforming experience. And I came back to Wikimedia and decided to stop being a manager and dive into tech writing and coding instead. But that's ALSO something to talk with me about later.
And we're hiring.
But today I am following in the steps of Jeremy Prevost and Bess Sadler and Mark Matienzo, among others, in their previous Code4Lib presentations and writings about emotion and design.
Now, you in this room are the experts on your pieces of software that you work on. And most of you already know if the user experience of that software isn't what you want it to be. So I want to give you some thoughts and some examples from other communities and products, about what good and bad user experience can do, and then talk about why we technologists usually don't put enough investment into user experience, and what we can do to fix that.
(And I'll be posting these talk notes, including links, to the Code4Lib wiki later today, as soon as Ryan Wick gets me a login, so you don't need to scramble to write down names and whatnot.)
So, we have a really big hurdle, for most services and products we make, and that is choosing to make the right things and make them usable.
I remember hearing this phrase in the 1990s, about the "last mile problem." Telecommunications companies laid down a bunch of fiber to connect one hub to another, but often the actual houses and schools and offices and shops didn't get hooked up for months or years, because it got finicky and expensive, or because companies weren't creative enough to do it well. And there are still people, in North America and in the larger world, dealing with this right now. We call this the "last mile problem," and I feel like usability is like the "last mile" problem. We have to be creative and disciplined enough to actually provide those services in a way people can use.
At least one of my friends said that's a bad metaphor, because that makes it sound like it's something to address at the end of the process. I do NOT mean we should address it AFTER we build other infrastructure, because of course we need to be thinking about design and usability right from the start, when we decide what to build. I mean that really usefully connecting with users is a hard problem of a different type than just building big pipes, and that if we don't do it we have a gap and aren't actually delivering.
Maybe another way of thinking about it is, when we're building services for people, we often have a lot more practice seeing from the computer's point of view than seeing from another person's point of view. In tech I think we understand how to build arteries better than we understand how to build capillaries.
Personally, I think capillaries are more interesting than arteries. Maybe it's just a personal temperament, but I like all the little surprising details of how people end up experiencing the ripple effects of big new systems, and how users actually interact with the user interface of a service, especially ones that we don't really think of as having a user interface. Like taxes, or healthcare, or hotels. But for all these big systems, there are little capillaries, where people exchange information or get healed or get whatever they need. And when those capillaries aren't working right, then those people just don't get what they need.
Over and over, in lots of different fields, we see that bad usability makes a HUGE difference, and that when choosing between two services, people will make very different choices depending on which service actually seems designed around the user's needs.
Earlier this month there was a conversation on MetaFilter about coffee machines, especially about those pod machines, Keurigs. And one person said, "This convenience thing is a bit overstated." And librarian Jessamyn West had a really interesting response, which I'd like to quote. She said:
"Not that I'm not more in your camp taste-wise than the Keurig camp but I see this as a great exercise for people generally in the "other people have different priorities in life and aren't just bad versions of you" direction. Not you [personally], just the general you. I know a lot of people for whom Keurigs solve a problem and they don't mind the downsides. I respect that. I also know a lot of people who wouldn't be caught dead with one and I respect that as well.
"I know it's tough to get your head around but the goal state for people with Keurigs is generally not to just have the best cup of coffee. It's to have a coffee solution that is easy to clean up after, or that turns itself off, or that has pre-measured sizes, or that has all the brands that people like, or that makes cocoa, or that offers holiday flavors, or that you can buy at the department store, or that is easy to clean, or that can be modified to accept change, or that you can put in a place with no running water, or that descales itself, or can be put in a place without a kitchen, or that has funky modern lines, or that can make ten cups of passable coffee in five flavors (caf and decaf) in ten minutes. These things do not solve problems for me, but they solve a problem for an awful lot of people which is why these things are so popular." http://www.metafilter.com/137379/Your-New-Coffee-Overlord#5456577
So, yes, usability makes a difference in what coffee people drink. And if you care about health, and education, and the working class, then this is actually scary, because the difference between user experience in two services is driving people to make bad choices. I want to give you a few examples.
- Payday lending and check cashing services have sprouted up in lots of working-class communities. And often it's just predatory or incredibly expensive, compared to traditional banking. And these services don't give people any way to save and earn interest instead of paying interest.
But Lisa J. Servon, an an urban policy professor at The New School, worked in a RiteCheck for four months and found that one reason people chose RiteCheck over a traditional bank was the user experience, the hospitality people got. She wrote in The Atlantic:
"The glue at RiteCheck is the customer/teller relationship. I interviewed 50 RiteCheck customers after my stint as a teller and, when I asked them why they brought their business to RiteCheck instead of the major well-known bank three blocks away, they often told me stories about the things the RiteCheck tellers did for them.....At RiteCheck, the tellers treated the customers as individuals and went the extra mile to assist them, perhaps in the same way that a neighborhood grocer might allow a trusted customer to run a monthly tab. On busy days, tellers regularly skipped lunch and coffee breaks in order to keep the wait times down. Ana Paula, our manager, often joined us at the window. The customer always came first and knew it."
Professor Servon believes that to attract these depositors, banks would need a better product -- fee and service structures that work for them.
- New York Public Library has been lending out ebooks for years, and New York City has a lot of people who have smartphones or dedicated ebook reading devices. I think we could expect that people who are already buying ebooks would be interested in borrowing them for free. But ebook borrowing rates at some of our libraries are just abysmal. For context, across the United States, as I understand it, for every 100 print books sold, somewhere from 30 to 80 ebooks get sold. (This is from some rough numbers that some friends and I put together and I checked with some Code4Lib folks yesterday.)
So given that New York Public Library lent out about 250,000 print books every week last year, you might hope that NYPL would be lending out, let's say, 150,000, or maybe even just a hundred thousand or fifty thousand ebooks per week. But the number is actually way lower than that: 19,000. So instead of a 3-to-2 or a 3-to-1 ratio it's more like a 13-to-1 ratio.
The reason I know about this is that New York Public Library has this division, NYPL Labs, which you might have heard of. And they have broken down the ebook borrowing experience and found that it currently takes eighteen separate steps for an NYPL user to borrow an ebook. So this spring they're starting a two-year project to take that down to three steps. That's their goal. And I would predict that this would raise ebook borrowing rates beyond this freaking 13:1 print-to-ebook ratio.
- At Wikimedia, our number one concern is gaining and retaining Wikipedia editors and contributors in general. If we want a world in which every single human being can freely share in the sum of all knowledge, then we need diverse perspectives EDITING and providing that knowledge.
And we've found that sometimes little things make a big difference. Like, for a subsection of an article, the "edit" link used to be all the way at the right side of the screen. So it wasn't obvious what it was for, and in fact some users thought it was the way you edited the previous section. When we moved that link to be right next to the subsection's headline, we got a much higher clickthrough rate, consistently -- the rate more than doubled. It went up by 117 percent. And that higher volume of clicks led to 8.6 percent more edits as well. https://meta.wikimedia.org/wiki/Research:Section_edit_modification#Overview_analysis_from_previous_test https://meta.wikimedia.org/wiki/Change_to_section_edit_links
And I don't know whether you've ever edited Wikipedia, but on Wikipedia, we see the bounce-off effect of hitting Edit and seeing wikitext, the markup language that evolved with MediaWiki. Generally when newcomers see that, they worry that something's broken. Because when else would you see that kind of jibberish? So we've invested in making a WYSIWYG editor, the VisualEditor, and you can try it right now on English Wikipedia, if you log in and turn on Beta Features in your preferences. Our hypothesis, which we're working to prove, is that if we make the user experience better, more people, and more different kinds of people, will edit.
One last note about Wikipedia and usability: you don't have to register to read Wikipedia. It's just there, no paywall, no regwall. And there are a lot of really great open educational resources out there, OERs, that have content that readers would really find useful -- except it's behind a registration wall. And that just massively reduces how many people are going to see it.
- OK, something a little different: let's look a little at health care. I think health care in the United States is kind of fractally broken -- like, it's broken on the macro and the micro levels, and you zoom in and you just keep finding new kinds of breakage. But you don't have to agree with me on that. Let's just look at the kind of traditional general practitioner acquisition, booking, and service. Usability is pretty poor; a lot of offices aren't open at night or on weekends, you have to call them during business hours to make an appointment, you have to book a whole appointment just to get a flu shot, and there's a long wait once you get there.
So in recent years we've seen some new startups that are usability hacks. Like ZocDoc, a website that lets you see doctors in your area who have open appointment slots this week, and lets you filter out the ones who don't take your insurance, and lets you book the appointment online. And MinuteClinics and other quick clinics at drugstores and big box stores are walk-in, low-cost ways to get quick diagnoses or shots.
But this convenience does come at a cost. Patients on ZocDoc are often just going to book with the next available doctor, and people who use MinuteClinics are skipping a longer doctor visit. So these patients maybe won't get long-term preventive care, and long-term relationships, and that might be something to consider. http://www.harihareswara.net/sumana/2007/01/16/0 One comparison I've heard in the library world is how ResearchGate and Academia.edu are hacks patrons use to get their needs fulfilled, which is great, except they're routing around the kind of deep and long-term research help your institutions could give them.
- OK, final example. Let's talk about privacy tools. Pretty Good Privacy, PGP - who's heard of it?
And who uses it on a regular basis?
PGP is basically the common standard for email encryption. But PGP is incredibly hard to wrap your head around, and encryption tech in general has terrrrrible usability, customer support, and localisation, and thus abysmal adoption rates. And as a result, journalists and dissidents and activists generally don't use it, even when the stakes are really high. Instead they're using Twitter and Facebook and unencrypted email, because those are just a lot easier to use. The Open Internet Tools Project's James Vasile said in his Open Source Bridge keynote last year that the privacy tech community would be better off spending the next year of our time on user experience design, localisation, and end user customer support, rather than writing new tools or features. https://www.youtube.com/watch?v=XiHsRPd_U-0
So these examples - banking, lending, Wikipedia, healthcare, encryption - show you how bad usability can really change what choices people make.
We in the open source community -- and code4lib is part of this -- put a lot of energy into lowering barriers to entry, like licensing and cost. And those are visible barriers that stop people from getting information. But bad usability is a barrier to entry too-- it's less visible but it's just as real.
Free access to education and free expression of political opinions and privacy from unwarranted surveillance are human rights (in the UN Declaration of Human Rights), and even if you don't want to go that far, in general they're social justice goals that we all like. We're here because we share a vision -- less waste, more empowerment, getting knowledge into people's hands. But actually getting these benefits to our users requires that our users be able to use our software. So, in getting rid of user experience obstacles, we are working to achieve social justice goals.
OK, since it's so clear why we ought to make our products usable, why do we have this bottleneck?
It's not just that the problem's hard. We've had other hard problems -- from mainframe to web, distributed computing, localisation -- that we've done better on.
Well, I can speak as a tech veteran.
This is where I start saying controversial things.
Let's look at what it takes to do user experience work. You have to look at your service from the point of view of someone who knows a lot less than you, and see where they're coming from. You have to imagine the reasons why they want what they want. Seeing that causation, seeing the connection between what someone's doing now and all the causation that went before it, is empathy. It's a little like reverse engineering; you're trying to unlock the DRM that's stopping them from getting what they need. Which is a really cool hack, actually.
We need to to exercise a disciplined empathy. It's an empathy that includes qualitative thinking, like interviews and watching people use stuff to see where the snags are, and quantitative thinking, like A/B testing and heatmaps.
But the tech industry is pretty crappy at empathy. And I'm speaking from my experience here - I know library tech is its own field - but in my experience of our industry, we just drop the ball on empathy and hospitality, a lot.
One reason is that our industry systematically undervalues the jobs and roles that require empathy, and has deeply gendered associations with hospitality and empathy -- they're not seen as masculine traits. And this isn't just a women versus men thing; it affects everyone. The tech industry values masculinity over femininity, meaning traits like hospitality are devalued. Those who perform or reinforce masculinity -- whether they're male or female, are privileged, and *everyone,* women and men, gets trapped in this cycle of (usually unknowingly) reinforcing all these things that say hospitality, empathy, and so on is stuff that gets you demoted down the respect ladder and the payscale ladder. And naturally, all this stuff gets smushed out of our software, because it's not just *not* incentivized, it's actually penalized.
And when the group making the software isn't very diverse, then the cycle just repeats, and gets even worse.
In general, marginalized people develop more empathy than the dominant group, because we *have to*. We have to be able to see from other people's point of view, the dominant point of view, as a matter of survival.
This is one reason diversity in a group is useful - it includes people with different perspectives, AND is likely to have more people with the ability to see from multiple perspectives. Including the users' perspectives.
And we need to be able to see from many different users' points of view, even when it's uncomfortable or shows us that we have failed.
This means treating customer support, those front-line desk and phone tasks, as a first-class responsibility and a source of important data. These are your bug reports. This is how you know if you aren't being as hospitable as you want to be. I heard some lore yesterday about a library that logs every time they have to say "no" to patrons, and then tries to fix it.
This means listening to designers and learning to speak their language, and it means that designers also learn how to integrate their work into the development workflow, which is happening. I recommend Crystal Beasley's blog post, "Code Talks and Designers Don't Speak the Language" http://skinnywhitegirl.com/blog/code-talks-and-designers-dont-speak-the-language/930/ , in which she says that in general, coders don't know what to expect from user experience designers, and developers don't have good judgment procedures to know whether proposed designs are correct. She says: "representing the work of a designer requires a shift in culture" which I think is totally true. Beasley also says, "The solutions to all these problems lie in communication and building a trusted relationship. It's a higher barrier for designers that takes time to overcome. I've found all of my team to be receptive when I've taken the time to explain the principles that guide my decisions." And I like that she mentions that -- I think in any negotiation, it's good to help people see the principles you're reasoning from, and the experiences you've had that led you to where you are now.
If you don't think empathy is one of your strong suits, you can change that. We all came into this community with some skills and without others; we're here at Code4Lib partly because this is a great community for learning and skill-sharing. And you can learn and exercise your empathy skills as well. If you have a sequential learning style where you prefer a structured approach, take a course in conflict resolution. If you want something more self-study, read novels and blogs written by people with lives very different from yours. And at your own institution, watch real users use your library and your library's digital services -- patrons and colleagues.
And the good news is that we have resources here! As Bess Sadler said, "at this intersection of technology and librarianship ... we need to not only bring technology skills and values to library services, we need to bring library skills and values to technology services." And I'd say that includes hospitality, and access.
U of Michigan's Library Information Technology group has a User Experience Department http://www.lib.umich.edu/library-information-technology/user-experience-department , and Suzanne Chapman there also has a really interesting blog at userslib.com .
University of Illinois's Library has a usability testing lab http://www.library.illinois.edu/sc/services/Usability_Testing/usability_testing.html -- two workstations and a room set up to conduct usability studies.
Matthew Reidsma at Grand Valley State University Libraries http://matthew.reidsrow.com/ has a "Work Notes" blog so you can see, for instance, how the GVSU Web team responded to data and anecdotes to make incremental improvements to their site. I also suggest you read his pieces "The Library with a Thousand Databases" http://matthew.reidsrow.com/articles/58 and "How We Do Usability Testing" http://matthew.reidsrow.com/articles/13 .
If you you may be interested in UserTesting.com -- it's a really quick and easy-to-use site that lets you commission a test, so testers perform tasks and record/narrate on video what they're doing. It's cheap - someone I know commissioned a test for $89 and got results in less than an hour. So if you want to start with usability testing and maybe just find glitches quickly, this is an option.
Aaron Schmidt, Amanda Etches, and Nate Hill's consultancy Influx http://weareinflux.com/ focuses on library user experience; they have a newsletter.
Within LITA, some folks have proposed a Usability Interest Group, so that's also something to look into.
So, you can see that other libraries are doing this. Sometimes we call it UI, or Human-Computer Interface, or user-centered design, or interaction design, and it intersects with product management, but it all goes to what I've been talking about. Several people on the Code4Lib list in October talked about what a huge difference a dedicated UX person or team makes. For example, Tom Cramer said, "We have been lucky to have a full time interaction designer within our library IT group for about 6 years. It makes a world of difference in the quality of our products." https://listserv.nd.edu/cgi-bin/wa?A2=ind1310&L=CODE4LIB&F=&S=&P=243880
So, we have some resources that I've just mentioned, and I know I missed a lot of people and guides and groups who can help out. So someone is starting, right now, an email thread on the code4lib list, to share usability resources. And if there's nothing at your institution right now, if you don't have what you need, then let's talk about how to hack your institution to get it. Let's talk about that this week, here in Raleigh, and on the list.
I want to thank all the people who have already talked about this with me and helped me work out the ideas in this talk:
- Coral Sheldon-Hess
- Mel Chua
- Andromeda Yelton
- Bess Sadler
- Emma Molls
- Leonard Richardson
- Jared Zimmerman
- Sky Croeser
and I want to thank all of you for the work you're going to do, hacking yourselves and your institutions to serve users better.
Better user experience is the best force multiplier we have at our command, so it's vital that we make it a first-class priority, throughout the development process. And with disciplined empathy we can do that - here at the intersection of libraries and tech, we can figure out how to scale hospitality, fix the new last mile problem, and actually achieve the social justice goals that so many of us got into this for.
- Sumana Harihareswara, Mar 25, 2014