Designing Questions and Environments to Generate Data for Innovation and Feedback Loops
Thomas John McLeish (or TJ), it is a bit of a challenge to tie up all the strings of your background into a tidy description. You seem to hack in many different domains on how things are designed and we experience them. When you introduced me to your work, we talked about coming up with useful questions that help then generate the data for design improvements and iteration. As I am a huge fan of questions that lead to better questions, I was curious. I hope we can we talk about how you come up with useful questions and then how you design for the data to help answer those questions.
Data seems so sexy right now, especially “big data” and I wonder what that means to you, TJ.
The notion of data – as a new lens … We have a full scale map of the world. The idea of big data is interesting right now… is actually not all that big in terms of what data is out there. It is being manageable – having a workable version.
How do I want to see myself in the world and in relation to the other?
The technocrats seemed to be obsessed with the data in itself. Mystified with the data.
It isn’t about data driving design so much as questions driving data. We ask, “what is it that I want to know about the world, and what do I understand about my model of the world?”
What has changed is that we have a deeper ability to measure on a very granular level physical activities. We have a small sensors that have an electric signal running through them. And then, as a result, we have, say, a picture, that you can put on FB.
We are able to practically measure many more things that than we were able to it in the past AND have increasingly sophisticated ways of applying algorithms to it to understand it. So we can digest this information.
How might you describe your current work?
It is fundamentally ethnographies built around models and hypothesis of something that is augmented by traditional techniques — and new techniques made possible by distributed sensing, massive data storage, increasingly sophistical algorithms (machine learning or clever statistics).
I am not a social scientist. Rick Robinson is. I get to work with Rick on translation and instrumentation of the world.
Basically, we enable decision makers to innovate. To see the world in a new ways allows you to innovate. It has the potential to allow us to validate models of how the world works. It allows us to build models of behavior to inform design around the future of what it means to interact with the internet of things or digital presence or however you describe this thing that is coming out way. Storyscaping.
How did you get into your current work?
I have always been interested in the integration of physical technologies into our physical environment. How does this help us lead better lives and better understand the world we live in? I am always looking at the balance between science and art. One of my parents is a rocket scientist and the other is an artist.
Origins? I studied architecture to understand the built environment. How to make things. Architecture is one of those areas where you are designing beyond the scope of human comprehension. It is so complex. That allowed me to be able to think about the relationship between digital computation tools and the built environment that people inhabit. In practice, I worked with Helmut Jahn. I had the privilege of working on some of the most technologically advanced buildings (or so we thought). These are high tech buildings: Sony Center in Berlin, New Bangkok International Airport, Flughafen Kolu/Bonn (FKB), Deutsch Post. Skyscrapers, airports, giant buildings.
There is a gap between the desire and the application of technology design. I was trying, at the time, to use digital tools to improve the decision process around design. Renderings help with decisions and become marketing materials. I learned a lot. But eventually, I was frustrated with the relationship of what I conceived of digital technologies and what was getting built on. So I went back to grad school. I got to work with Bill Mitchell and Kent Larson on House_n, thinking about the home of the future and how it is built. Not as a singular instance, but as an idea. I worked on a collection of various interesting projects. Inventing things around the deep integration of technology and the built environment. Embedded distributed network: The Place Lab. One of which it is how easy it would be to integrate technology into our environment if all the players cooperated. The promise of being able to have deeper insights of human behavior in the context of the home… which leads to human behavior in the world. How we really behave in the world. Not how we think we do, but how we actually in practice behave.
There is a certain gap between who we think we are consciously and who we really are being subconsciously.
There is a promise for deep insight into human behavior for informing design. Working with Rick and John for three years in pursuit of getting closer to people, reinventing how we.. [this section isn’t coherent….should I cut it or do you want to craft a sentence of two to segue?
These are sensor based ethnologies.
I am not sure what you mean by “Sensor based ethnographies” TJ, can can you share more about how the data is creating an ethnology?
Data, information, knowledge, wisdom, each step up requires confidence and interpretation. (triangle)
Confidence means different things in different contexts. I don’t just mean confidence in terms of machine learning (a number that represents confidence and standards of what it means to be confident) and I also mean confidence in terms of social science – the belief in.
And the challenge here is turning hypothesis and models and questions into something we can instrument. You are looking for evidence of something having happened.
How do you work with people to frame questions and form hypothesis?
Some clients we work with have very sophisticated understanding of the products they make, the people that use their stuff, etc. They come to us to get closer to those models – to validate the model they have or change the model – to see something in a new way.
Looking at complex patterns and making sense of them.
Some clients do not have a sophisticated model and we are developing them. It takes some degree of trust for them to step into this process with us.
Both are trying to further advance their models in order to innovate.
There is something about the kinds of data you are collecting and the kinds of things that can then be designed. Can you talk about that?
People come in and ask, “what can I do with an accelerometer?” They are focused on the instrumentation and not on the question that the instrumentation should serve.
One of the metrics of a successful question is how many more questions that it produces.
TJ, one might say that asking what an accelerometer might be capable of doing for them would lend itself to asking LOTS more questions… can you say something about the valuation of those question?
In working with a client, coming up with a way to interrogate whether you have come up with a successful question from the business viewpoint – what are they trying to know or understand. We work in the sweet spot where we can have a hypothesis and can then refine the hypothesis and build instrumentation to test it out and see what emerges. If we are working with a retail client around engagement in the store – we can come up with a model of what engagement looks like. The question might be something like, “This is my design intent. I made a change to the arrangement in the store. Is it more engaging?” I can now see more clearly or closely what I thought was engagement, so I notice things about engagement that I wasn’t seeing before. And that leads to richer questions around engagement.
How are you developing and refining the hypothesis, TJ. I think this is something that sparked our early conversation – the art of asking good questions. Do you have a process for that?
Put yourself in the position in being an alien – how do I get outside existing expectations? How do I escape the culture I am embedded in with the assumptions it is already operating within? How do I cultivate the alien view?
Multiple perspectives. Lots of evidence that you can look for around technology supporting activity in the workplace, and we try and ask questions from many different vantage points to discover what is useful.
Can you explain more about the processes for doing that?
We have four different techniques.
How and Why
One is simply the game your kids play with you. Kid asks why is the sky blue. You give them an answer, and they ask why. Why do you want to know how many people are in the room? They give an answer. And we ask again, why do you want to know that? Why are you interested in activities in the workplace? They answer, we ask why are you interested in workplace performance. You are trying to extract out of them what really is the core value that they are testing. Then you can drill down into how are we going to get into workplace performance with groups of people. Yes, we may want to know about how many people, but also instant messaging, emailing, etc that help us understand what is happening between the people in the room.
Basic word games
You are still trying to tease apart the high level goals and the evidence of that high level goal into something that you can actually measure. These are like crowbars to force a careful articulation of your model using a linguistic structural technique. We are trying to get to a very granular description of a high level concept. We want it to be granular enough to be measurable. So say a company creates a lounge area – and they have a concept of what they want to be to the world. There is a design intent – lounges to hang out and work together and drink more coffee, as if it is an offsite workplace. So, what does an offsite workplace mean? Maybe at the first pass we say ‘coworkers come to hang out.’ What way would I be able to see, touch, or otherwise sense that verb, hang out or lounging, happening? So maybe Lounging is signified by people sitting. Lounging is signified by people sitting, having conversations. Well, how do we know what that is? How would we know that is happening? Conversations that have multiple people talking. They are taking turns while talking. We often look for a key – not a single thing that happens but a collections of things that are happening to let us know we are, say, having work conversations.
Metis
We are developing a tool internally around structuring a hypothesis into something we can instrument. For example, “Is my dog happy when I am not home? We can ask, “Then what is a happy dog? How do we know? How do we know he is not happy? Wagging tail, eating, sleeping all the time might be indicators. Presence of tail wagging and absence of sleeping all day long. Then we can ask how we instrument for these things.
Concept Maps
In their most rough state are a diagram where you are working with someone who has a belief about the world, and you help them draw literal connections between that idea and things they associate with it. We borrow from Hugh Doubblerly on concept maps and models in general. Concept maps show you relationships between things but may not always give you direct evidence.
We also work on broadening the instrumentation – not just a single sensor – getting enough width to allow yourself to be alien – and find things that you didn’t expect to find. And what else can I notice that might be relevant to our innovation or might be irrelevant to the current question but useful in some other way or approach.
We may mix these methods together, and do a concept map to show related issues and then drill down with the how why game or the word game to get clearer on the evidence we want to look at.
It is useful to both work on a hypothesis and also to go beyond what data will affirm your hypothesis – so also looking for a bit of the unexpected.
We set up instrumentation that allows you to see – more vantage points into the question at hand and more items that you might not have considered.
We are looking for evidence of their map – a description – an empirical description and map of their world, help them see it, and then test their belief about it.
People then use the data you generate to design and innovate?
Not really data driven design, as if data is in the lead. It is inquiry, hypothesis, then instrumentation to have evidence that then increases your understanding that helps develop an insight into innovate with.
Horse-less carriage – having to describe something from the past.
Design is decision making. You make decisions based on something — probably data.
We don’t really do design. We are involved in understanding the context in which design might happen. Helping to develop the intuition for designing… developing the understanding of how to project into that environment.
And that is not about say, designing the bottle. It might be more like designing the system that produced the bottle. The whole flow of how it comes together and how it flows from here.
How people engage with things, their perception of it… we offer a way to begin to do that. People places or things – our client work is about the complex interdependencies of the perception of the thing, the lifespan of the thing, and the system it is a part of.
I think a goal is to help people see more clearly the relationship the people places and things to engage – for them to make it better on their terms.
I think we are seeing more and more..
Often, when we have talked about this work, I get these many dimensional visions. How are you mapping and visualizing multi-dimensional data?
Because we start with their description of their model of the world, we can use that model to present the data back to them. We can take their qualitative view and make it quantitative for them. There are different ways of looking at that information. Sometimes it is a pie graph. Time is often a backbone for glueing the data together. We can learn where the position of a product in a grocery store may influence buying behavior.
We often have 6 or more kinds of information linked together through several different methods of presenting the information. Say we have time and a quantity changing over time, and we tie that to a physical map of the quantity of people moving through the store. We might take a series of people walking through a path, and then make it a topographical map based on the count through that path. Then we consider how much time they were on that path and develop a heat map overlay on that topography. We discover that there are different types of people on the path based on where they go and how long their trip is.
We would do a very different visualization if we were focused on a product than on a space – who touches it, when, for how long. We would use a third compilation of visualizations if we were to focus on a person – what they do during the course of their day, what is their path, what do they interact with, and who do they interact with.
What do you see this process and the data making possible in the world? How do you see it helping to expand the option space for organizations?
I feel increasingly in control of my engagement of the world. I am no longer flipping through the Sears catalog for holiday gifts. I am flipping through kickstarter to see what more I want to see in the world. At the end of the day here is some person involved. You can look at the bottle, but there is a person using it. To allow us to get a closer view of human behavior.
Itunes model – restructures music environment – changed how people are thinking about how they engage in music and I think it gives people more agency. A ubiquitous digital presence, scary or not, getting closer to people is naturally going to enable people having more agency – more ability to make choices that work for them.