Interview with Bill Gaver

Together with Tony Dunne and Elena Pacenti, Bill Gaver developed the cultural probes. He also introduced the idea of Homo ludens in interface design. He is a design professor at Goldsmith College, where he heads the Interaction Research Studio, and is also involved in the Equator project.

Your last projects particularly address the domestic use context. What do you think is special in a domestic use context with regard to the designer’s work and methods? What implications do you draw for the designer’s activity?

What I like about designing for the home is that it contrasts so nicely with designing for the workplace. As an interaction designer dealing with computational technologies, it is important to me to get beyond traditional concerns with problem-solving, efficiency, utility and even usability to explore other values and the approaches that are appropriate for dealing with them. The home is a great setting for working these issues through.

Of course the home presents special challenges for design. People tend to be pretty protective of their homes, so you need to be sensitive about the things you seek to introduce and also with how you get access to the home, for instance in initial studies or later field trials with prototypes.
Still, those sensitivities are themselves important topics for design.
And so far we've been very lucky with the people who we've worked with, they've all been very generous about allowing us into their homes, telling us about themselves, trying out the things we've made, and telling us what they like - and dislike - about them. I like to think this is because we've developed appropriate approaches and designs, but perhaps we've just been lucky.

And it's worth noting that the home is not the opposite of the workplace.
People do a lot of work at home, both in terms of their jobs and in terms of the effort needed to build and maintain a running household. Equally, people play and explore and indulge their curiosity while at work. But the home is a place where people have relative freedom to decide for themselves what they want to do, how they want things to look, what sorts of values they want to pursue. So it's been a great place for us to explore designs that are open-ended and that encourage curiosity and exploration. If I can quote Charles H. Parkhurst out of context: "Home is heaven for beginners." It's great to think about what sorts of things one can design for a heaven.

What kind of strategies do you have to develop ambiguous artefacts? According to which criteria do you choose a particular design for prototyping? Which properties do you consider important for ambiguity?

First, it's probably important to say that 'ambiguity' is one point in a shifting constellation of attributes that characterise our designs.
Depending on the particular instance, its sometimes more precise to say that our designs are open-ended, or multilayered, that they provide resources for a space of interpretations, or that they purposely try to create ambivalence. So 'ambiguity' should be read in a fairly broad way, not to suggest that we're providing some set number of interpretations to choose among.

Having said that, there's a few tactics we've used to create ambiguity.
One is simply to create a situation, or provide resources, without saying how people are to properly approach it. So the Drift Table, for instance, simply scrolls aerial photography depending on where weights are, but nothing about its design suggests what it's to be used for. In fact, we even purposely blocked some obvious interpretations - e.g. that its for going to particular places - by making it very slow and offering little in the way of navigation aids. Other tactics have included producing exaggerated outputs, such as the horoscopes used by our Home Health Monitor, which invite people to question the system's authority and instead value their own interpretations. Our latest designs tend to create situations or provide resources that can be interpreted from a multitude of angles. For instance, the Plane Tracker, which imagines journeys based on information picked up from passing aircraft, can be interesting from many points of view - for instance, those of people living on flight paths, or plane spotters, or people who love to travel, or people curious about geography, or environmentalists, or....

As for how we choose what to prototype, that's a complicated question.
It's not a very rational process: we generate lots of design ideas, and a few seem to stick as interesting enough to build. What makes them worthwhile is a set of considerations including the ways they speak to research issues in the communities we address, and whether the technologies they use present an interesting but achievable challenge. But paramount is whether the situation they create seems to us to be rich enough to engage us and others for a good amount of time. Some ideas are great, but don't really need to be prototyped - the proposal is enough.
Other things really need to be built to understand how they will play out in people's lives.

Its really important that it is the making that guides our research. All the methodological and theory work we do is pulled along by the things we make, and that fact is absolutely central to our work.

How do the cultural probes you collect inspire the design process?

What the Probes don't do is lead in any straight-forward way to specific design moves. We rarely if ever have worked from a particular return or set of returns to a design in a way that could be described neatly. When you get Probe materials back, you're still left with the problem of what to do next - they don't really help with that at all, and that can be frustrated to people first starting to work with them.

But the Probes are great at creating a context in which to design.
Returns tend to provide us with a really rich and detailed view into people's lives. At the same time, it's a picture that is fragmented, and often impossible to interpret with any confidence. That is crucial to the approach. Taken the right way, that uncertainty becomes a license to imagine and over-interpret, to tell stories that can lead to designs.
More definite sources of information seem to close down possibilities, to provide rigid parameters for design. When the work, Probes create a productive balance between confronting us with lots of things we didn't know about or expect, while leaving huge amounts of room for us to move.

What role does irritation play in your work? is it a part of the plan to irritate the users by confronting them with new (and alien) functions of everyday things? or do you consider the familiar part the more important, to make the enhanced artefacts fit well into the domestic use context?

Good lord no! We don't want to irritate people! At the same time, we're not interested in building on warmly familiar objects either - we don't look at dust-covered picture frames and think they'd be better if only we added SMS or whatever.

Instead, we try to create new artefacts that are well suited to the world they are meant to inhabit. We genuinely want people to find the things we do valuable and meaningful. Thus the things we make often draw on a critical awareness of assumptions and trends, but they are not themselves 'critical'. On the contrary, our designs are meant to offer positive alternatives rather than simply highlight what's wrong with prevailing views.

Considering the history tablecloth, you mention the intriguing effect that some random behaviour, caused by unforeseen difficulties with the particular location of the table, had on your test subjects. How did this experience influence your further projects?

Briefly, what we found is that whereas we perceived 'random' behaviour as a problem, the participants who tried the table saw it as an interesting facet of how the table operated - making it something to be influenced rather than controlled. We aren't likely to draw on that in a linear way by, say, introducing randomness to our designs. But it has added to our understanding that people will make sense of systems even if they appear 'noisy' for one reason or another.

That realisation (which wasn't just from the History Tablecloth by any means) underlies a lot of the systems we build. The Home Health Horoscope, for instance, is based on the idea that people will use their own interpretations of their situation to correct or go beyond the inevitably limited interpretations of a sensor-based system. We think this is a key insight for Ubiquitous Computing.

Our two latest designs, the Plane Tracker and Local Barometers, both rely on events in the world (passing aircraft and wind speed) that we don't control, and on 'readymade' information (GoogleEarth and local
advertisements) which we also don't control. This means they can be unpredictable. Changing winds mean aircraft take new routes.
Advertisements depend on whatever people have to sell at the moment. But that noise, that randomness, is okay because people can make sense of it.
Because people are good at dealing with noisy, difficult to predict situations, these systems can be far richer than if we had programmed all their actions and content ourselves.

One of your projects, the home health horoscope, is a system where sensors in the home capture activity patterns and draw assumptions about the well-being of the inhabitants. Based on this, it prints out a horoscope every day in the morning.The system interprets the data, and the ambiguous form enables the user to apply his own understanding. Here, the purpose of the system is more obvious as i.e. for the drift table. Also the interaction with the system is less direct. Where do you see the advantages or disadvantages of both approaches?

I'm not so sure the Hhhoroscope is clearer in terms of its purpose than the drift table. It simply prints out a statement once a day. What are you to do with it? Believe it? Ignore it? Argue with it? Base your life on it? It's not clear - people have to find their own orientations to it, and a number of different ones appeared over time in the family who lived with it.

The Hhhoroscope does have a far less direct interaction than the drift table, however. You can't really play with it in the same way. Our volunteers did fool around with the sensors a bit, but the reaction is so slow (once a day) that I think they lost interest in doing that. Instead they lived with it as another voice in the home. That's a strange sort of relationship; it means the interaction is much more about the idea of the system than any physical interaction.

In fact, most of our recent pieces aren't really built for interaction in the typical sense, not even to the degree the Drift Table was. Instead they simply do their thing, and people can engage with them or not. We think of the systems as offering resources to people that might allow new perspectives or interpretations to form. So the interactions are not physical ones, they're about interpretations or orientations, about how people understand the systems and how they understand themselves and the world based on the systems. Of course, that sort of interpretative appropriation happens with all systems, but the nice thing about building the systems we do is that it allows us to really focus on this sort of interaction as a primary interest.

You also report some scepticism of the home’s inhabitants towards the sensors you installed in their home – that they suspected them to spy the family (and even took one apart to see what’s inside). You mentioned that you tried not to influence the test subjects by revealing too much of the system’s behaviour. Could you comment on their reaction?

Well, it's not surprising that people are suspicious of sensors in their homes, is it? One of the points of the home health horoscope is to turn a skeptical eye towards the culture of surveillance we live in. There's lots of talk about using sensors to control houses, or look after vulnerable people, or even to allow computers to be emotionally sensitive, but a lot of us feel that the potential invasions of our privacy makes these ideas suspect. One of the ideas of building a system that explicitly relies on people's interpretations is to relax demands that sensing be increasingly precise and comprehensive - perhaps a few, indicative ones will serve as well.

But in the case of the Hhhoroscope trial, as you say, we didn't tell the volunteers much about the sensors. Our reasoning was that we didn't want to destroy the novelty of the system, or to encourage the volunteers to develop all sorts of preconceptions about what was, after all, a pretty simple system. We wanted them to discover it for themselves. This turned out to be a huge mistake, obvious in hindsight. Because we didn't tell them what all the sensors were, they were naturally very curious - and sometimes suspicious - about all these boxes in their home. And they couldn't tell much just by looking at them (or even taking them apart).
So for a long time, that curiosity overwhelmed their experience of the system, and coloured the way they interpreted the horoscopes.

We're preparing to do a field trial of a new version of the system -- no longer using horoscopes, I might add -- and this time we're telling the volunteers a lot more about them. In fact, we've done a real U-turn:
instead of trying to set things up so we simply give people the system with no explanation, we're borrowing an idea from Kirsten Boehner, a colleague of ours, and trying to bring them in as participants in the evaluation of the system. We're going to be much clearer about what's going on, and ask them to help us tune the system in situ to work as well as it can. We'll see what happens...

You have a lot of experience with long-term field testing, where you observe the usage of one of your objects in a household over several months, like for the home health horoscope. Here, you have to rely on a few people as tests subjects. What is your strategy to make sure that the people you choose are appropriate? What kind of people are you looking for?

Yes, we tend to try our systems with very few households, but over very long periods of time. The reason for that is that pragmatically we can only spend so much time and energy on testing, and we believe it is really necessary for people to live with the devices for a long time to have any idea of the relationships they will form. With experience, we've begun to see similarities in what my colleague John Bowers calls a 'trajectory of appreciation' over different trials: first people love the novelty of the systems, then they become disillusioned when the thing turns out to be different than what they imagined it was or could be. Then they either get over that and start to engage with the system as it is, or they lose interest completely. That initial process takes a week or two, so we have to test our designs longer than that. And if people do maintain their interest -- and we've been lucky with most trials -- then it takes weeks longer for people to develop stable ways of orienting to the system, and even after months new things keep happening. Given the choice, I'd far rather do a single long term trial than many shorter ones.

As for choosing volunteers, it's important to think about the aim of our field trials. We're not in the business of making commercial products, so these are not marketing trials where we try to assess the potential customer base for our designs. Instead, they're explorations of whether the things we build embody values or encourage activities that people might find motivating. So we're looking for a sort of existence proof with our trials: confirmation that at least some people will enjoy our designs some of the time, and insight into how they enjoy them and how the prototypes fit in their lives over time. That feeds back into the claims we make about the trials -- we're not trying to say that 73% of the population likes them and that they're particularly popular with 28 - 37 year olds, for instance. Instead, we try to tell detailed stories about the particular volunteers who tried them, what their reactions were, and allow our audiences to speculate about generalisability on that basis.

So how do we pick our volunteers? We're pretty open. On the one hand, we definitely look for people who we think might like the prototypes, who seem generally open-minded, and who are willing to talk about what happens. But surprise is extremely important to us as well: it's the reason to work with volunteers either in field tests or initial user studies. So we find people we don't know, and who aren't too similar to us. We've used friends of friends, though I tend to avoid that as I think you're likely to get people to similar to yourself. Lately, however, we've had some luck asking people who've tried our prototypes if they have friends who'd like to work with us. Best, in some ways, is to advertise in newspapers or magazines because you get a much broader and more surprising set of volunteers. On the whole, we've been very lucky -- all our volunteers have been extremely generous with their time and interest, and we've met some really fantastic people. Doing field trials with people you've never met is a huge amount of work, but it's also hugely enjoyable.