Core77

Design for Emergent Systems

By Steve Portigal - October 23, 2006

in Events

A recent BayCHI panel on Designing systems with emergent behavior featured Tim Brown (IDEO), Peter Merholz (Adaptive Path), Larry Cornett (Yahoo), and Joy Mountford (Yahoo), and was moderated by Rashmi Sinha. It was a deliberately free-flowing panel; no prepared remarks but rather discussion and some banter. My notes (after the jump) are similarly unstructured. It's hard to capture everything that comes up conversationally, so most of the apparent non-sequiturs are artifacts of my note-taking. The panelists were smart and articulate; this medium may make them appear less so. Please send in any corrections or additions if you were there. [Peter's thoughts on the event are here]Tim Brown reframes the issue of designing for emergent systems as designing for communities. He offers an example from their work with Kaiser Permanente: a way of creating more innovation in healthcare environment. They took teams of doctors and nurses and ran them through a whole process of UCD. They go through the process, find projects and create innovations, and then the IDEO team moves along to the next group. They are designing tools, processes, methodologies and ideas. That's what they designed: helped other people to design. Eventually 40 different innovations had been created Larry Cornett used the example of Yahoo Answers: Besides the value of information and emotional benefit of answering questions, there is fun and entertainment in looking at the site. There was a change in how they looked at design. The key is ratings and reputation. It's not all high quality content and there's no way that Yahoo can moderate that content, so it's critical to give the community tools, but you do not know what will happen. You have normal human behaviors: greed, retaliation, anger, altruism. Their approach was different: getting a solid system and not over designing it and watching what happens. Peter Merholz cited Epinions where he worked 7 years ago. It had much of the web 2.0 functionality, didn't really take root. Joy Mountford described the redesign of the Yahoo home page. Wiith the huge amount of users and the established behaviors, many of which they had to go and discover (the heavy user, the light user) led to features like the new light-customization module, she terms this a "living breathing interface." The first ever, it leaves a bit of you behind when you go elsewhere. Rashmi Sinha then formally introduced the session by relating how Tim told her at a party that "MySpace is not designed" so she asked if he would come to BayCHI and say that. She wonders if the tools and methods we have enable us to work in this new paradigm, when we have to ask what are you designing? Is this even design? The Facebook fiasco and subsequent uproar begs the question, that if those were good usability decisions, how would you anticipate these reactions? Tim: Contrast that with physical design where you have more chances to test prototypes, with rapidly changing software, it's too easy to do something new. Seems like a new feature got launched before a design process happened. Maybe they didn't get to test it a little bit. Not referring to Beta, in videogames they are always testing all the time. It's part of the design process. He prefers that to the classic Alpha Beta approach Peter: Facebook doesn't have a lot of empathy or condescension historically. Initially they were "but this is so much better. This was similar to the backlash with opening up Facebook. Feels like a lack of appreciation on the lack of how users are going to respond. A significant portion will always freak out, but at what point can you ignore hem and move on? This is a contradiction, he acknowledges. Larry: At eBay you have to buy and sell, and in his group at Yahoo you have to be on Answers. You have to be part of the community. You can start to anticipate if it doesn't feel good. Joy: Everyone should just go for it; anarchy is where it's at. She agrees with Larry, though, with "know your user." She finds it difficult to know if she's going to a site with 500 or 500 million users and what she expects from those sites. She remembers HyperCard they worked on "oh no, anyone can do an interface" but they still did a guidelines stack and thinks that hopefully the bad stuff goes away on is own and maybe even more quickly than they used to. General discussion about "what made YouTube so successful?" Infringement? If we knew we'd be rich, it's simple, Lazy Sunday video. Making video embedded It's content (and that breaks a lot of comparisons like Alexa lets you do). Video as communication, a new brand that is free of the constraints of existing brands, and what happens next? Rashmi: How do you balance "don't listen to them" but "watch them", i.e., make stalking harder? Our methods are too oriented to the individual. How do you understand the group? There are no design research methods to be able to understand these larger reactions. How to do social design, when 100 people watch 100 people do something, how do you do understand that? Tim: Agrees, it's very hard to look at social phenomena, especially non-visual behaviors to look at visual artifacts you will be creating. They've run very abstracted live prototypes that can be explored. It's not clear what it's about to the user but people can start doing stuff. They might have to start building larger software models, not sure they can do say, massively parallel user testing, say. Larry: It is the produce-and-release process, not the design process. You have to do more modeling of risk. What if people do his or that? You have to think about it so you have a plan. Tim: Be dystopian as well as utopian. Design is inherently optimistic. It's important to step back and look at the dystopian way. Writers are better than designers in creating a scenario that is intentionally pessimistic, this helps you explore the design a bit (and they've brought in writers to do this for them). Larry: With Yahoo Answers, in the first release the ratings didn't work, on purpose. They released it and saw what happened and then made changes, essentially a test release. Joy: Discoverable interfaces. This is not like industrial design, it's way more complex. The teaching is not keeping up. She doesn't know how to teach any more, we don't have apprentices learning what people do any more. Gehry used to not use computers, even; she has heard they are now. He was a great teacher. Who are the great interaction designers? She doesn't know any. Peter: Tracing genres through organizations book. The first two chapters are great, ignore the rest. It challenges the dogma of UCD of user as victim. As in "it's so hard for these poor users to stalk, let's make it easier for them" Maybe there is something that is desirable about those current behaviors. It's easy for designers to see themselves as superheroes and users as victims; maybe how we think about users needs to change. Joy: Bring in actors. At Interval they called it Informance. Think out of the box. Rashmi: When you have millions of users, and you have logs you will find every possible usage, incredible amounts of data that is locked away in spreadsheets and is hard to understand. We need the help of designers, in terms of tools. A second point is that Gmail was used internally a long, extended period of time before being launched. When you've got forums with hundreds of messages, how do you mine those without extensive resources? Peter: Create APIs, tools that let more people do more stuff. Facilitate an ecosystem. Joy: In Second Life people are contributing to the new design of W hotel. Will they take any of those suggestions for real are is it just PR? This is the beginning of a trend, perhaps [SP: beginning of a trend? Co-creation is such an overblown buzzword, I think the trend is well on its way!] Larry: The evolution of reputation points. Sometimes people were gaming the system and Yahoo took that ability away and people were okay. In other cases, people said "no we want that, so please put it back" and they did. You'd be surprised how addictive a point-based system becomes. Some people answer questions all day. Some of those they hire. Users have to answer questions to get points, asking questions costs points. This makes more people offer stuff. Voting helps sort out answers and you get points for voting. Joy: The story of project on Amazon's Mechanical Turk asking people to draw sheep. They paid 2 cents per sheep but in the end sold 'em for 4 cents. They have no actual value and they aren't well done, but the fact of how this project evolved is fascinating. [and with that she left] Rashmi: With the rapid iteration and things remaining in Beta, is that a way of never saying I'm sorry? Are we going to see finished products in the social space or will companies start adopting an approach of "we're going to be hands off in case there is a problem?" Tim: Stuff isn't ever perfect and it isn't usually perfect, either physical or software. We go public earlier than we used to because our manufacturing (on the web) is so much more accessible than big factories with machines. He doesn't see a whole lot of difference. Larry: It's okay that it's never done. It's a good thing, it's great that it's evolving, allowing them to adapt to users, to the culture. Peter: As consultants we do see this as a delivery model. Clients get something and keep doing it. As designers how do we deliver value ? Rashmi: Aren't you designing a "medium" when you design an emergent system? Things get defined by a user group. She remembers Instant Music for the Amiga that allowed you to compose music at various levels of freedom or constraint, i.e. any note or only notes that would work. That's one of the things that you are trying to do in emergent systems; to figure out what useful sets of constraints would be. But also style, there are styles of use that you might be able to find with some analysis, if you could capture it you could publicize it and help people find the right style and adapt to the styles. Tim: "Tools and Rules." Designers are having a hard time with the identity as enablers rather than architects of specific experiences. Our education needs help. We need to create examples. In things like Photoshop you have always seen that, always great to see examples of what you could do with it. We'll se more of that, we'll need to see it as design. Peter: This predates the web. Read Steven Johson's Emergence hIt considers evolutionary systems, cities, urban planning, and what we can learn from them. He's not sure it's a medium, maybe more of a space. How buildings learn is appreciated by people in this space. Rashmi: In grad school, seems like they all knew that social science had punted on group interactions, the combinatorics of that stuff were difficult, most people in experiments were confederates, "social excrement" in weblogs, so is this a golden age. But we don't have a good understanding of what is the metric for a good organization; we just know some organizations work and some die. You could get never get a European and American to agree on style of government, there's not a vocabulary for evaluating organizations. Chris Alexander has a gnomic set of books about patterns, but nothing that is quantitative and trackable. Maybe there are opportunities now to study this. Tim: Karen Stevenson is a professor somewhere in southern California. Steelcase built something off her work, a social networking pattern recognition/tracking system, manually, with surveys. They did it at IDEO to see what would happen, got amazing diagrams of what their social network looked like. The folks who helped them do this told IDEO that they were off the charts in some aspects (i.e., most socially networked organization they had seen) but found negatives too like their devices for sharing knowledge were low. They didn't know how to measure; they don't know what good is anymore. Roger Martin talks about the difference between organizations striving for validity (right idea) or reliability (efficiency). What am I as an organization, before you can even measure you have to ask and answer those questions. Rashmi: What happens with the aggregated content that is growing over time? Larry: They are exploring that, they realize it's a big deal. One simple answer is to index it back to web searches. Stuff that core web search would fail at before, they are seeing their Answers content show up, and now Yahoo Answers can benefit you in search. Peter: That's how he ends up at Wikipedia. Amazon.com stuns him with how no one has been able to take advantage of emergent behavior. Consider the feature shows what customers ultimately buy after looking at this item. Although demo doesn't work in this instance he explains that last time he looked at Jane Jacobs "Life and Death of the American City" there was a reference to "Fear and Loathing in Las Vegas" - and no librarian would ever have offered you that, but the system helped them find that reference in cases where it would be relevant (i.e. what were they really looking for versus what they were able to search for). Question: A post on Creating passionate users about mom and daughter and MySpace, lead to someone asking his Wells Fargo colleagues about what generation Y would do with online banking. How do we design more "serious" applications like finance and health? Tim: These folks will be customers of online banks before too long. This will start to change many of the systems they are using, email systems, anything. Larry: (joking) Why not have a small release of a page that shows the checking account balances of all your closest friends and see how they react! But there is more application design going on that has traditional use cases, and maybe smaller cycles and reaction to stuff could get more feedback in more quickly. Peter: The role of visual design in these systems...the more polish MySpace and Facebook has would limit a user's comfort. Something about the appeal of the user created design that lends it authenticity. Appropriateness is important (i.e., a black background and shot of the ocean from the bank?). Problems in trying to mix and mach these models, i.e. that these teens won't use email when they go to work, it's bullshit. Email is the best tool for work but not great for at home and chatting with their friends, but teenagers don't have anything value to communicate at 7 pm. Be careful with how we assume those trends are going to apply. Companies that stay true to themselves. When Barnes and Noble tried to be Amazon, it didn't work, Tim: Cell phones:nobody else but teenagers is interested in sticking stuff on their cell phones. [Rashmi looks very very sheepish, shows her phone with stickers, to big laughs] Rashmi: How can academic researchers contribute to the discussion? Peter and Larry emphasize the softer side of social science, such as danah boyd's work. Rashmi asks why so few academic researchers are reaching out to designers. Papers are published in locked journals and no one reads it, instead now she writes it on her blog and people can come and read it. Why aren't academics reaching out? Rashmi asks about potential for abuse of information in Wikipedia, an emerging emergent system. Power will be multiplied to come. There are blips like Lonelygirl15, but there is a lot of power that these systems can wield. Who are the gatekeepers? We have stable editors in the print world. These new realms have tremendous credibility but there is concern over the lack of power and the vetting. Tim: That hasn't that always been the case. Isn't the so-called stability of the print media an illusion? In the UK you couldn't believe what the papers said; you knew that they represented different issues and you had to decide what to read. He doesn't believe it's a new issue. In the end, the effectiveness of the sites will be based on how much trust there is, so that people will go somewhere else if there is a better job at being trusted done somewhere else. Larry: Being more open allows for moderation and more self-correction Peter: You can see all of the editing decisions in Wikipedia versus the NYT. The NYT has an aura of authenticity for a long time, this isn't anything new. Journalists don't have a special line to truthsaying that the rest of us don't have. Tim: Kids are getting more skeptical. Rashmi: Does the system create the behavior or does the behavior already exist but you have opportunity to do it, through the system? Larry: Yahoo Answers has lowered the barrier for people who don't want to blog or don't know how to create a web page or don't know how to blog. It gives them a quick way to offer up a little snippet of knowledge. It's been there always. These systems are facilitators for existing social behavior, they don't introduce new or artificial ways of behaving. Peter: To start you need to facilitate something that already exists, i.e. fmous YouTube story of frustration in sharing wedding video. What then happens is that then you start seeing people figure out other things to do with the. People take advantage of that, i.e., taking advantage of comments and reviews (gaming the system) is not a behavior that was intended. Larry: Those are existing human behaviors but the system has allowed them to twist the tools in a new way. Tim: Often the most successful sites start off with a successful behavior and they evolve and we evolve with them, they have a life cycle. They reach a point that new users can't start using them. Even the Treo in the product world has evolved to that point, it has a life, and other products will evolve and attract new users. Sites will have to start figuring out a way to attract new users after a while or cut back to a level of simplicity again. These complex behaviors that evolve get so complex that new users can't get into them. -- My unasked questions were around how these systems change or could change or should change when they got adopted in large numbers by unexpected types of users. Tribe and the Burning Man and other alternative cultures, Orkut and Brazillians and Indians. These create unexpected revenue, design, interaction and other challenges. What about Spore or other Will Wright efforts that are designed purely to explore with few goals, or where the exploration outweighs the achievement of goals, at least? Where do they fit into this discussion?

o

Favorite This
Q
1

Comment

Comments

Simon King

18 years ago

Z

Z

Reply

Thanks so much for posting this. Do you know if the panel was recorded as well? I'm thinking a lot lately about how products evolve through use and this is really related.
!Report as spam