(look at slide notes for full talk transcript)
Imagine a world 8 years from now where instead of a warehouse, Amazon is a factory, where products are made in small quantities based on direct input from users to designers. In this world design directly drives product creation, and where data informs design.
(special thanks to Joel Truher for many of the ideas and Alex Chaffee for the Amazon example)
3. THE NEW PRODUCT
DEVELOPMENT ECOSYSTEM
HOW WE WILL SKETCH
PRODUCTS AND REINVENT
MANUFACTURING IN THE
PROCESS
Mike Kuniavsky
July 20, 2012
Portland, Oregon
Good morning! Thank you all for coming. When people ask me what they should talk about at Sketching, I usually say that it should be something that you’re passionate about, and ideally something that you’ ve never talked about before. In the past, I ’ ve exempted myself from that because, well, I ’ m the organizer, but I really wanted to do that this year. This talk is based on ideas I’ve been thinking about for the 8 or 9 months, and much of it is based on you, on things I’ve heard at Sketching and on the specific work of the people in this room. So, first of all, thank you.
Imagine Amazon 8 years from now. It looks like this. Yes, it looks exactly like the Amazon today. It has all of the familiar ways to discover new products, to compare them, to see what people think of them, to see what goes with what. It has wish lists, Gold Boxes, the whole thing. But there’s a crucial difference. Instead of Amazon being the front end to a fulfillment system, as it is today, the Amazon of 2020 is the front end to a set of factories. The back end didn’t look like UPS, but Ford Motor Company. When you click on on buy you start a manufacturing process at the factory nearest you, instead of a delivery process from a warehouse far away.
I know what you’re thinking: “Mike just saw a MakerBot and got all excited. We’ve heard this all before, it’s called mass customization, and it’s never worked out.” Why talk about this again? Because I think that the presentation of mass customization as “configurators for everything” missed the point. That totally gets the user motivation wrong: most people don’t want to be designers, they want to be consumers. They have better things to do than figure out what colors and patterns look good together, what is in fashion, what functionality they need to include in the firmware. They’re busy. They want someone who is a professional to do that research, to think really hard about what they need, to be really fluent in the tools that make it good, then to create a solution.
I’m also not talking about desktop manufacturing. As much as all us geeks want a Star Trek replicator, it’s not that useful in practice. We just don’t need that much new stuff all the time. Paper printers are useful because they represent high density information that fits into a rich existing culture of information use, and even they’re not used nearly as much as ecommerce sites. Outside of work, people probably shop a lot more than they print.
I think more importantly, both mass customization and desktop fabrication imagine a new world that’s different than ours. I have nothing against envisioning new worlds and working toward their creation, but creating new worlds, changing the behavior of millions of people, is really hard and takes a really long time. If we look to a world 8 years into the future, odds are that it’s not going to have changed that much. I’m talking about a new world that from the outside looks and works exactly like our world today. It’s still driven by the thrill of finding something awesome when you’re bored surfing the Internet and then making it yours by buying it. The relationship between the consumer and designer are still intact. Designers still design, ecommerce sites still help people find stuff they like, people still buy. However, there will be a crucial difference behind the scenes, and it will be this difference that changes our world of centralized warehouses to a world of distributed factories.
The difference is analytics. When you order from the Amazon of 2020 a counter is incremented that registers that you, a human being with a set of well-known behaviors and a demographic background, decided to buy this specific version of this specific idea. Moreover, since the world of 2020 is a world of ubiquitous computing, every product has a small bit of digital hardware in it that tracks how the product is used and, with your implicit permission, sends that information back to a central server, which aggregates and anonymizes the results. This is of course exactly how large-scale Web design works right now, but now we map it to all products.
Every design decision is the result of hypotheses about people’s behavior, preferences and interests, tested by observing actual behavior with competing designs that are rapidly deployed and whose use is closely monitored. Big web sites embed the scientific process of hypothesis generation, testing and validation into nearly every design decision in what is probably the most scientific innovation process in history. This diagram shows really big changes, but in reality the changes are tiny. It’s what Google was doing when that entire “43 shades of blue” controversy came out, which was really just them taking a useful tool too far.
When this approach coupled with all of those digital manufacturing technologies we love—CNC machines, laser sintering, Makerbots, etc.—we see a new way of creating products. When you have rapid, cheap, distributed manufacturing capability AND real-time analytics you have a new way of designing products. You can take those Industrial Age design processes that took years to react to people’s buying patterns and you can speed them up by orders of magnitude. Image: http://commons.wikimedia.org/wiki/File:Tailfins-evolution-1957-1959.jpg
This tight loop between an idea and market validation of that idea is also the core of the Lean Startup philosophy that has been so successful in creating a wide variety of Web products and mobile apps over the last several years. The core concept is that you use the lightest weight method you possibly can to test an idea in the marketplace, that you make as little new technology as you possibly can to test a hypothesis. This is a slide from Steve Blank’s 2008 on customer development that illustrates this basic idea. My vision, my hypothesis, is that it’s possible to do this with ANYTHING, but applying the ideas, practices and technologies we developed for the Net to everything else.
The Sketching in Hardware tools that we have been talking about, developing and designing with form the infrastructure of this ecosystem. They provide a core set of tools and practices for rapidly taking abstract ideas and making them physical things. It’s not surprising that so many people in this room have crossed back and forth between manufacturing and electronics: today manufacturing is almost a subset of creating electronics So many products are digital, or are becoming digital, that the tools and concepts have begun to overlap.
To me, the whole ecosystem looks like this. Here come the buzz words, so excuse me in advance. Digital fabrication, we know what that is. It will allow us to make all kinds of things in small batches. Ubiquitous Computing and the Internet of Things is leading to everyday objects that send a stream of telemetry when we bring them home. They have an information shadow in the cloud that can be data mined. Big Data Analytics crunches all of that data to create information about people’s behavior. Social commerce creates sales channels that sell small numbers of products by finding niche markets and letting them market to each other. Kickstarter is currently a social commerce catalog for stuff that doesn’t exist until enough people want it. And finally, cloud-based design tools will allow designers and engineers to collaborate on the distributed development of physical products. And I don’t just mean cloud IDEs, I mean tools like Dave’s Arduino hack last year that connect physical prototypes to cloud-based collaborative systems. This is my ecosystem vision: a world where design directly drives product creation, and where data informs design. This is a world where products are made in small numbers only when they are requested. They are made locally, with local materials and workers, while at the same time being able to use design and engineering talent from anywhere on earth. In other words, they use the best qualities of both atoms and bits: atoms are available everywhere, bits travel fast. Designers in this vision add hypotheses against the actual market to their toolbox of design methods. In the full pipe dream, this means we use fewer natural resources, take full advantage of talented people wherever they are, create only products in large quantities that people need and want, meet the needs of tiny niche audiences, while still taking advantage of the infinite variety implicit in digital manufacturing technologies. Whew!
The key piece, and this is where I would like to end this talk, is collaborative design tools. In moving to a system that can rapidly iterate on ideas we essentially shift power from the replicators and movers of things to designers of things, to groups of collaborative creators. Unfortunately our tools for collaborating on the design of products are still in their infancy. CAD systems are huge and incredibly difficult to learn. Product Lifecycle Management systems assume that you’re always building a commercial airplane, and are also insanely complex. Moreover, none of these tools seem to recognize the possibilities of the cloud in bringing people together to collaborate. Github got to where it is through an evolution of tools and practices that began with makefiles. The physical world isn’t even at the makefile stage.
This group is developing a generation of tools and design practices that take collaboration ideas from software and from the Web and apply them to the development of physical products. Digital electronics are the focus of the meeting, but the underlying theme is how we support people in creating new physical things, all kinds of things: products, artworks, environments. I intend to make this vision my next focus as a designer and entrepreneur and I need your help: over the next three days, tell me where I’m wrong, tell me what I don’t know, tell me who I should talk to and where the opportunities are. I think this will change the world, I want to change the world, and I need your help. Image: Legend Performance Cheer box prototype by Abraham Peters
Thank you.
The packaged goods industry has been using a similar technology for years. It’s called conjoint analysis. The way it works is that when you try to figure out what attributes of a particular product are most important to customers—is it more important that it’s vegan, or that the portion sizes are big? That’s a hard question that people can’t just answer easily—and you mash them up in a controlled way. Then you offer test subjects from a specific demographic cross-section multiple variations of a complete-looking product and watch which they choose. You do enough of this and there’s some well-defined math that lets you extract what’s important not from what people tell you, but from what choices they make.