A Map of the History of Science Fiction

According to the artist, “History of Science Fiction” is a graphic chronology mapping the genre from its roots in mythology and fantasy to the space operas of today:

The movement of years is from left to right, tracing the figure of a tentacled beast, derived from H.G. Wells’ War of the Worlds Martians. Science Fiction is seen as the offspring of the collision of the Enlightenment (providing science) and Romanticism, which birthed gothic fiction, source of not only SciFi, but crime novels, horror, westerns, and fantasy (all of which can be seen exiting through wormholes to their own diagrams, elsewhere). Science fiction progressed through a number of distinct periods, which are charted, citing hundreds of the most important works and authors. Film and television are covered as well.

The image was inadvertently leaked onto the Internet a couple weeks ago, going viral enough to prompt Shelley to start collecting emails from fans interested in a possible poster version of the map. In addition to being a visual delight to explore, it is also a great reading-viewing to-do list for a complete science fiction education.

Shelley—a Brooklyn artist specializing in large paintings and sculpture installations—has also done paintings depicting the evolution of Avant Garde, the New York downtown scene, media role models, and Fluxus.

Deb Roy wanted to understand children learned language. What makes this collection of captured sounds and images revolutionary, however, is the techniques his team has developed to track, visualize and analyze the everyday connections we humans make by living in the world.

By the time his baby-makes-three moment finally arrived, Roy had wired up his house with cameras intending to catch (most) every moment of his son’s life. The nine rooms of his house were tricked out with a birdseye lens looking down on the activity below. Unlike a lab setting, where observations of interactions are artificial and prompted, Roy’s recorders could find thousands of key unsolicited, natural moments of childhood development. After scrubbing the artifacts to make provisions for privacy, he handed the files over to his Cognitive Machines research group at MIT. Together, they turned the detailed home movies into a massive data set of language development.

The little Roy had picked up a vocabulary of 503 words by his second birthday. The TED audience was treated to an audio montage of the evolution of the boy’s understanding of water, moving from “gaga” to a an articulated “water.” In that clip, you can hear the cognitive spurts and experimentation until the happy ending when he owns the correct word.

In his talk, Roy describes the stages of meaning-making the researches progressed through to find some interesting insights about the relationship between context and learning. The first step was to create “space-time worms” using motion analysis to track movement in a room as a function of time. This allowed the researchers to focus on activity in the data the revolved around Roy’s son. The result was about 7 million words of transcripts.

The team then studied the relationship of words and their use in the context of the world. By mapping the individual players in space, Roy found “social hotspots”—where a more activity was spent together—and “solo hotspots” showing the boy interacting alone with his surroundings. These trails were mapped in the house as a function of time, creating wordscapes—hotspots rise like mountains on the map. For “water,” the peaks are in the kitchen. For “bye,” they are near the door.

After examining the caregiver speech in the key moments when a new word was acquired, researchers found that caregiver speech would change. The language the child heard from caregivers was simplified in those critical moments, allowing the learning to take place before gradually building it back up to complex language.

Beyond the boon to language researchers, Roy’s work has two other big impacts.

This same method of working with massive datasets can be applied to how we interact with the mass media—specifically, television—through social media. Applying the language-mapping techniques to media led to the founding of Blue Fin Labs, Roy’s agency that measures engagement with mass media. In near real-time, they can track how tweets and other individual digital footprints are tied to specific shared events on television. They mine about 3 million comments a month on relevant topics, linking them to an event and thus allowing a wordscape to be generated.

The second impact is one of personal history and reflection. Digital artifacts can capture moments in early childhood that the brain can’t possibly remember from a perspective no one has experienced, a kind of God’s eye view of history. To underscore what it means to be able to share your own developmental milestones with your children, Roy rolls back the 3D rendering of the video camera footage when his son took his first few steps. In response, I can only echo the word shared by father and son in that moment: Wow.

Screenshots from Pennant

Pennant is an interactive visualization depicting Major League Baseball from 1951 through 2010. It is a gorgeous visual treatment of the kind of statistical data I enjoy browsing, from over 115,000 baseball games. You can view any team’s complete history, any given season, or re-play-by-play individual games. Since this is an interactive application drawing from time-based data, this is a great tool for exploring how the fate of teams changed over the course of a year or game. It costs $4.99 in the App Store, but it is great to have as a demonstration of what the iPad can be (as well as anticipating future updates).

The application was initially developed as a thesis project at Parsons by Steve Varga. Pennant was launched courtesy of data drawn from Retrosheet and The Baseball Databank. Varga created his own API to manipulate the raw data, to leverage new calculations that support the visualizations. He may release that massaged data to other applications in the future as an open API. More of Varga’s work is available at Vargatron.

The application isn’t perfect, of course:

• When searching the 1983 season for a game I went to, I was annoyed that going back to the season timeline always reset me to the middle of the season, not the game I was viewing.
• Navigation is clunky if you want to move laterally. There isn’t a way to look at the detail for a game and simply flick to the next game in that view (there should be).
• Keyword search is non-existent, making it impossible to search for references to Roger Clemens pitching against my White Sox and explore just those games.
• There is no player tracking. The data is available to follow an individual player through his career, like you can with an entire team. Imagine re-living the now tainted 1998 home run race between Mark McGwire and Sammy Sosa.
• Customization isn’t supported, nor are personal histories or favorite games. I would love to go through my old stack of scorecards and ticket stubs and curate all of the games I attended in person. When player timelines are supported, it would be a great tool for documenting my annual fantasy baseball rosters.

Pennant is currently only available on the iPad and requires a wireless connection, so it may not yet be practical for taking to ball games to look up stats. It is, however, quite satisfying for replaying games I attended, like the time the White Sox knocked a young Roger Clemens out in the fifth inning or the “Ryne Sandberg” game on June 23, 1984.

Most of my applications are for work productivity, games, or news aggregation. I like the possibilities of specialized apps that allow easy and enjoyable exploration of large data sets in a specific domain. The tablet is wonderful for this kind quick interaction and could be quite useful to support journalism, tourism and attending conferences. This one is particularly well done, and I hope some updates are coming in time for the 2011 regular season.

Due to the topic and depending on your own experiences in the world thus far, the links might lead to insensitive, inappropriate, or even traumatic commentary. I’m mostly going to focus on this as an example of negative participation in public discourse.

The previously-on-The-Internet scenes that are relevant include:

• Courtney Stanton is a producer for DINO Interactive Studios, a storytelling and game application publisher in Boston. She also graduated from Indiana University back in 2002, so yay!
• Penny Arcade is a webcomic about video games and video game culture. Creators Jerry Holkins and Mike Krahulik also put together an “semi-annual” conference (PAX).
• In August 2010, Penny Arcade publishes “The Sixth Sense,” which references rape by creatures called dickwolves.
• Some people objected to the frivolous use of rape as humor, and then a bunch more objected to PA’s sarcastic response.
• Mike draws a dickwolf at PAX (NSFW), leading the following month to a t-shirt of the creature for sale on PA.
• Courtney objects, offering a survivor’s guild t-shirt for sale.
• A few months later, Courtney is asked to speak at PAX and declines, citing all the above. She is immediately griefed on social media channels, most notably her blog

Although the story continues, that was how Stanton got her data.

Setting aside the important catalyst for all this activity, the high-level (but presumably tedious and probably painful) analysis Stanton conducted on her blog comments led to some interesting data-driven insights. She divided all of the comments up into groups, first separating based on agreement with her post and the level of aggression shown by the commenter. The final grouping looked at whether the comment tried to share information or hurl personal attacks. The result was a definition:

Troll (n.): 1. One who contributes nothing but noise to a conversation. Often disagrees aggressively and insultingly using personal attacks.*

Stanton acknowledged that the comments she is most likely to want to block are the ones attacking her.

When looking at how the comments were threaded and directed, she discovered that two-thirds of all comments were responses to other comments while only 17% of the comments by trolls were focused on others. In other words, most blog conversation doesn’t involve the original poster, yet the overwhelming majority of trolls ignore the conversation entirely.

The text visualizations are also informative. In a word net, rape is the most used word among all comments, but it is a word with peers (including: penny, arcade, culture, people, dickwolves, people, comic, shirt and more). For trolls, however, the message is all the same: It is overwhelmingly a one-word message. The word trees for “fat” and “I hope” are equally supportive of Stanton’s definition of troll. “The trolls only have one hope,” she said in her talk, “and it’s like they’re reading from a script.”

Jiwon Shin (Columbia University) theorizes that it is the absence of an online morality reinforced through education that allows a troll to exist. Susan Herring and other researchers examined the strategies of trolls almost a decade ago. Their work pointed to a few suggested interventions to mitigate the effectiveness of trolls. These included killfiles or other controls that allow one member to mask the contributions of another, and education, to improve members’ ability to recognize trolling tactics. Strong and clearly-worded policies also can establish a framework for discourse that establishes the boundaries of acceptable behavior.

* One of Stanton’s commenters pointed out that trolls might also agree with the original poster. Jaishankar and Sankary might classify such people as “Love Rats.” If not trolls in the way described above, then perhaps we can call them jötnar for their wisdom, or think of them with a flop of pink hair. Of course, the correct etymology refers not to the Norse creature but fishing, as in “trolling for newbies.” In that sense, both positive and negative are acts of trolling, albeit for different fish.

The map of my LinkedIn professional network

The map is interactive (to you). When you click on one of the nodes in your network, that person’s profile is displayed in the right sidebar and the connections you each share are highlighted. It is easy to see who the influencers and connectors are among the people you know. This has some practical value for better using the LinkedIn service:

You can use those insights to measure your own impact or influence, or create opportunities for someone else. So, you might see two distinct groups that you could introduce to become one. Or, you might leverage one person to connect them to someone else.

Source: Visualize your LinkedIn networkwith InMaps, 1/24/11

My use of LinkedIn has been largely limited to my academic connections made during my time in the Indiana University School of Informatics and Computing. Much of my initial exploration of social networks occurred as I prepared to graduate with a Masters degree in 2006, motivated by a desire to stay connected to those in my class and the two classes before and after mine. This is clearly represented in the dark blue cluster. Since that initial growth, I have mostly responded to connection requests but very rarely seek out new connections. This has led to a second large cluster of Informatics connections acquired during my quest for a Ph.D.

Other groups of note reflect my local Bloomington community (rose colored) and my old high school friends from Woodstock, Illinois (light orange). I also have a couple small clusters based on connections in Silicon Valley (maroon), mostly acquired through my Twitter API book, and my pre-academic job at TicketsNow (light blue). The most interesting color coding is the deep orange, scattered throughout all of the clusters. These people seem to be Indiana University students and faculty who are not Informatics folk.

While this is an interesting reflective exercise—both to consider who I know and how I’ve used LinkedIn—it would be more useful if I could also see the second connections, the people my people know. Understanding the next-step reach of my connections might inspire some expansion into clusters of interest and to know how the people I know are known outside of my network. I’ve never understood why this information isn’t easier to get to on LinkedIn, especially since I can visit anyone’s profile and it suggests a path to connecting to that person.

This service is free, provided your network is large enough to map (at least 50 connections and 75 percent of your profile completed).

This graphic—created by David McCandless—was inspired by Rohit Bhargava’s analysis of the Sysomos study, released in June. Ignoring the fact that the David’s viz now has only 95 people, not 100 (he hacked off an extra row), the idea behind the graphic is appealing: reduce Twitter to a number we can wrap our heads around.

The problem with doing so—and adding the snarky comments about these sub-groups—is that the value of Twitter is being pre-determined by the most obvious metrics, followers and posts. It is also misrepresenting the actual numbers in the Sysomos report.

Let’s critique David’s graphic based on the groups he identifies using color coding.

5 with more than 100 followers (blue)

On the surface, so many small networks seems bad. There are many blog posts and several Twitter applications that are geared specifically toward cultivating lots of followers. The biggest Twitter membership drive to date was fueled by a race to 1 million followers. However, there are two important things to realize about Twitter that soften the shock of this figure.

First, social networks follow a natural growth based on age of the account. Merely being around the service long enough is bound to find some random attachments that accumulate over time. Almost three-fourths (72.5%) of all Twitter accounts were created in 2009. It would be interesting to re-examine the data trying to normalize for length of membership.

Second, the implication is that only large networks are valuable. This is a decidedly blogger mentality, where readership means impact. With Twitter, value comes in many forms. An oft-overlooked value is strengthening small, known, local connections. I consider the small overall percentage with large networks to be a sign of strength of the service, not a weakness.

Technically, the body count in the graphic should be 6 (not 5). Sysomos found that 93.6% of the 11.5 million Twitter profiles they examined had fewer than 100 followers.

5 creating 75% of the content (purple)

When the uneven distributions of content was disclosed in analysis earlier this year, the popular implication was there was some problem with Twitter. While I haven’t seen a formal comparison of Twitter’s distribution of activity with that of other communities (both online and offline), I suspect that these numbers match up well with what we know about group participation.

At a certain size, the group dynamics influenced by signal-noise and turn taking will start to encourage inactivity in the majority of the group. When in the same room, there is a scarcity of attention that tends to be owned by the person doing the talking. Some people are more predisposed to be that talker, and others the listener. Online, the same is true with posting and consumption of information. That doesn’t mean the group is dysfunctional.

There is also a relationship between network size and posting activity. According to Sysomos, as Twitter users attract more followers, they tend to Tweet more often. At 1,000 followers, the average daily posting rate climbs from three to six tweets. Above 1,750 followers, that number rises to 10. There is no effort made in this study to parse out the quality or type of tweet. Replies, for instance, increase with network size simply because there is more consumption of social artifacts around which to communicate. It may also be true that those most likely to have large networks are the kinds of users who believe Twitter to be about broadcast (news media, A-list bloggers, celebrities)

20 with empty accounts (red)

Here again, the graphic is a bit off. The Sysomos data suggests 21 out of 100 are empty accounts, never having posted a single tweet. It is not clear from the report, however, if all 21% also have no network (i.e. following no one, being followed by no one).

Empty accounts can be placeholders or test accounts, as Rohit speculates, but they can also be requirements to use desktop clients or web/mobile applications that allow for consumption of other content. The absence of posts is not necessarily an indication of an unused account. I guarantee there are lurkers in the Twitter community, as much as any other. Lurkers are active in a different way, not “dead.”

50 with no tweets in the last week (green)

“Active” is a very ambiguous term. Most commonly, it is defined by the available statistics that can easily prove a member is using the system. There are no such statistics available for readership, for instance, or to take into account specific use cases or network size. One man’s active is another’s dormant, or—in David’s eyes—”lazy.”

It is also unclear, with so many new users, what percentage of these will grow into more active usage. There are a number of accounts that would have been classified as dormant a year ago that are now active. For some people, it takes time to learn to integrate Twitter into a routine.

20 with a small following and recent content (gray)

This is the most ambiguous group. It is viewed on the map as “leftovers” without considering how these previous groups might overlap.

For starters, the flip side to the previous statistic is that about half of all twitter accounts are using the service at least on a weekly basis and about 70% are posting at least once a month. The implication from the graphic is that none of the 50% who haven’t tweeted within the past week are the same as the 20% who have never tweeted. With so many new accounts, however, it is likely that many of the dead accounts fall into the Green group. So rather than a 30-70 split of dormant to active, it’s actually 50-50, making this gray group considerably larger.

The same is true with overlaps with Purple (lots of content) and Blue (lots of followers). As pointed out earlier, there is some relationship between tweeting and network size. It wouldn’t surprise me if most of Purple and Blue groups were the same people. Recognizing that makes the Gray group larger. In fact, since Sysomos found that 85.3% of all Twitter users post less than once per day, about 35% of all accounts fall between daily and weekly use.

I’d love to see someone take another crack at this visualization with a deeper understanding of how the statistics overlap. Perhaps even turn three of those male icons into women to further improve its precision—the Sysomos survey discovered there are more women on Twitter (53%) than men (47%).

Social Collider, a Chrome Experiment in action

The project is several months in the making. The @socialcollider Twitter account posted the first update about a revision in early January. Social Collider is one of the nineteen initial Chrome Experiments, which showcase interesting uses of JavaScript and web browsers created by designers and programmers from around the world. In the description of their project, Schmidt and Pohflepp describe the information problem:

With the Internet’s promise of instant and absolute connectedness, two things appear to be curiously underrepresented: both temporal and lateral perspective of our data-trails. Yet, the amount of data we are constantly producing provides a whole world of contexts, many of which can reveal astonishing relationships if only looked at through time.

The solution is inspired by visualizations for particle colliders, drawing pictures of how the subatomic matter interacts. The tweets that resonated with others connect horizontally with those of other users writing about relevant topics. Sometimes the connections are direct and obvious, but this visualization also reveals more subtle relationships amongst all the Twitter activity.

To start the visualization, you define two options. First, the keyword search can be done on a username, a phrase, or the trends tracked by Twitter. Second, determine the duration of the data set, from one day up to one month. Once submitted, the querying begins to populate a graph with tweets. The most recent tweets are represented as dots at the top of the screen, and each user in the network appears as its own column of dots—identified by topic, link, or username at the bottom of the screen.

That’s when the magic happens.

Social Collider shows how my tweets are related to others

Using connector cues, such as @username references and shared links, the visualizations starts tracing the connections made between tweets. Every colored track is a new related topic; You can trace the patterns of how the content and people are connected by following any line. A flash of retweets might be a tight circle contained near the top of the page, whereas more persistent topics run the height of the screen. Conversations spiral around between users, easily identified from the mass of dots.

A topic search for my the initial announcement of my book

The AIG scandal is tracked as a Twitter trend

The weaknesses of Social Collider are few but important. Each tweet can be revealed by mousing over any dot, but the target is so small that this becomes a difficult task, impeding exploration. At least on my MacBook, loading this tool results in that jet engine sound that accompanies a heavy processor load. I was unable to leave the visualization, too, without the browser asking me to manually stop the script from running.

I was impressed, however, with the patterns that did emerge. The user search appears to be the easiest to comprehend, showing how my most recent tweets on topics like Star Trek and AIG fit in with or sparked other posts. My new book, “Twitter API: Up and Running,” got a plug and a bunch of retweets a week ago. Those show up as a swirl and a red line crossing several users. My test of the Twitter trend “AIG” was surprisingly segregated, with not much activity running horizontally. I’m not certain what that means, but I do appreciate that the visual patterns are made easy to identify should I want to investigate.

Give it a try. This kind of insight and originality is sure to characterize Twitter visualizations in 2009.

Jeff Clark, a Canadian programmer and visualization fanboy, released a neat visualization Tuesday leveraging Summize‘s API of Twitter content. Twitter StreamGraph examines the last 200 tweets containing a given keyword or @username, parsing the rest of the words in those messages into a time-arranged stacked graph.

Twitter StreamGraph for @kmakice

Four streamgraphs show completely different trends. The keywords used for these searches were: informatics, Bob Schneider, makice and Bloomington.

The 200 tweets are graphed on a horizontal timeline from oldest to most recent, scaled to fit in the available space. This creates some interesting visual meaning, as content scrunched toward the left side might indicate a dormant topic. Each of the other words are sorted for frequency and stacked vertically near a center line. Each band reflects the ebb and flow of use of that work in conjunction with the search keyword, creating a wavy pattern. Tall parts might mean many different words or many uses of the same words.

In his search results, Clark allows you to focus on a single word from the list of secondary terms in the visualization by clicking on it. The corresponding band turns red and reveals some of the specific tweets that use both words (the secondary term and the search keyword). Exploration is facilitated by making words in the tweets clickable themselves, spawning a new StreamGraph.

To test the tool, I tried a few personally relevant searches, leading to some interesting insights:

• makice—My last name is not used in isolation very frequently. I wonder if a future iteration by Clark might allow this search to include content for both Amy and myself.
• informatics—Realized I most commonly refer to my school as “SOI” or “IU,” not “informatics.” Most of the informatics conversation on Twitter is done by Europeans with an emphais on biomedical and health informatics. There were a few interesting spikes I remember about ambient informatics and personal informatics, too.
• Bloomington—Much of this content is due to an Illinois newspaper using Twitter as an RSS broadcaster. There was also a big spike when gas price watching on the west coast found a station nearing $5 per gallon. This term is too noisy to be a good Twitter search, thanks to Bloomingtons in at least three other states. This is where a filtering mechanism would be helpful.
• Bob Schneider—Our favorite rocker has some definite patterns of activity over weekends, when the Austin singer is giving concerts. Quite a few song titles are have recognizable parts in this stream.
• @kmakice—I have a wide stream, with up to about 15 tweets in a day.

Clark is also responsible for two other appealing visualizations of Twitter: TwitArcs and Twitter Spectrum. The visualizations are all composed in Processing, a development tool that is making it easier for programmers and non-programmers to create interactive visuals.

Signs of good health for Twitter
The StreamGraph release comes on the heels of some encouraging news about everyone’s favorite microblog whipping boy. Twitter, which experienced some crippling downtime in May and June, appears to have turned a corner toward stability without losing their growth in membership. Now that third-party applications are coming back online as well, Twitter is starting to add new features again.

After officially acquiring one of those applications—Summize—Twitter announced embedded search tools built on that new platform. The new search is nice because it keeps track of new content even as you are perusing the previous results, updating the page title with a growing count in much the same way Gmail does for new email messages.

NOTES:
As people are walking in to what will be a packed room, Ben is pulling up old photo galleries showing profs in the 1960s.

3p start

Creativity support tools: Accelerating discovery and innovation
Ben Shneiderman
* founding director (since ’83) of HCI lab
* his first job was at IU
* gave a talk last night (hmmm … SOI didn’t let us know about it)
* background in databases, growing into a perspective of “20% social psychologist”
* impressed with how the Informatics program is working here
* authored “Designing the User Interface”
* worked on “embedded menus” (which Tim Berners-Lee would eventually call “hot links”)

Spotfire
* gender and age differentiated births in DC … filter by age, birth weight, plurality (how many kids in same birth, i.e. twins) … sociologist not surprised by this data, but interesting for Ben
* not a walk-up-and-use tool … a collaborative tool

Treemap
* stock market, clustered by industry … market falls steeply on Feb 27, 2007 with one exception (small green square in map of red)
* used for discovery … detect insurance fraud, supply chain management, patterns of oil wells (underproduction), newsmap (overview of news stories), gene ontology
* scalability … illustrated by server IP map

Reasserts the human as the discoverer, rather than the machine

SciViz vs. InfoViz

Patternfinder
* patient histories … spotfire-esque controls to filter by age, turn on/off genders … filter by events, like white blood cell count -> purple dots indicate matches in the field … two-criteria -> matches in “ball and chain” to further filter data sets, by stringing together more criteria
* lots of temporal constraints that are difficult to specify otherwise
* 2006 paper published in Visual Analytics conference
* Patternfinder 3 -> flash based, but created new problems … coming to understand what kinds of queries we can do in seconds, minutes or days
- interface for specifying queries (past SQL)
- execution of these things
- display of the result set (find patterns in a large data set very easy)

how do you prove that this leads to discovery?
* surrounded by skeptics who question the HCI endeavor
* http://www.cs.umd.edu/hcil/CC2007

NSF provided $50K for an art exhibit

trying to transform the discipline … visualization is not a far out thing, but something we should do as computer scientists

Creativity Support Tools:
GOALS:
- more people, more creative, more of the time … multi-disciplinary endeavor: software engineers, scientists, architects, product/graphic designers, educators, students, humanists, new media artists, musicians, composers, writers, poets, screenwriters … open the space up (Flash allows for many more people to produce films on YouTube; Dramatica Pro – brilliant implementation of Theory of the Story, by Robert McKie, guru of Hollywood screenwriters … 250 questions, to build structure of his story)
- Looking for “normal science” (evolutionary, product design, engineering, music & art) … Not revolutionary breakthroughs/paradigm shifts … nor impromptu everyday creativity … the things we do can be helped by providing better tools

Key Sources:
* Creativity (1996) [based on interviews with many creative people] and Finding Flow (1997) … Csikszentmihalyi
* Sternberg – Handbook of Creativity (1999) and International Handbook of Creativity (2006) … computer, HCI doesn’t appear in the index -> opportunity for us to make a contribution (how the tools might reshape creativity)
* National academy of science – Beyond Productivity: information technology, innovation and creativity (2003)
* Richard Florida – Rise of the Creative Class (2002) [communities that work with creativity have to work in certain ways - controversial], Flight of the Creative Class (2005)
* eric von Hippel – Democratizing Innovation (2005) [support open source]

1) Structuralists: A plan, method, process
- Dan Couger (1006) reviews 22 “creative problem solving methods” … preparation, incubation, illumination, verification -> French words, 1945 book on a mat mentor, cataloging the creative methods he used
- Treemap solution came as an aha moment (3 days for 6 lines of recursive code after that)
- Atman’s design steps: problem definition/identify need, gather info, generate ideas, modeling, feasibility analysis, evaluation, decision, communication, implementation (from Design thinking research symposium 2003)
- combinatoric exploration, structured problem solving (Russian TRIZ, Arrowsmith), self-help books, business consultants

2) Inspirationalists: Aha, Aha, Aha!
- free associations (brainstorming, ideation, Thesauri, photo collages, random stimuli, inkblots)
- breaking set (getting away to different locations, working on other problems, mediating, sleeping, walking)
- visualization (concept maps/2D networks of ideas, sketching)

3) Situationalists: context, community, collaboration (takes a long time, depend on social group)
- Personal history – family history, parents, siblings, challenging teachers, inspirational mentors
- Consultation – experts and friends, information and empathic support, early/middle/late stages
- Motivations – fame, legacy, admiration, contribution and competition (finding right motivations can facilitate collaboration)
- CREATIVE PEOPLE need: mentors and support/cheerleader (empathic support)

Csikszentmihalyi’s Creativity (1993)
1) Domain – “consists of a set of symbols, rules and procedures” (i.e. biology)
2) Field – “the individuals who act as gatekeepers to the domain … decide whether a new idea, performance, or product should be included” – your creativity does not mean anything until you gain recognition from the domain, understand and please members of the field
- most rejections fail to reference the right people in the field
3) Individual – creativity is “when a person … has a new idea or sees a new pattern, and when this novelty is selected by the appropriate field for inclusion in the relevant domain”
- early stages -> fear of ridicule, stolen ideas
- aggregate individual work, but it happens “in between the ears”

not quite a requirements doc for a computer project, or a research statement, so …

Eight Activities:
* searching and browsing digital libraries – look at your own work and find where Google doesn’t work (Google is great at finding single bits of information)
* consulting with peers and mentors
- facilitate getting the right information (how do we track collaborations)
- spotfire’s success is based on the ability to collaboration
* visualizing data and processes
* thinking by free associations
- expand horizons, make surprising connections
- ex. Arrowsmith
* exploring solutions (what if tools)
* composing artifacts and performances
- starting points: exemplars, templates, processes
* reviewing and replaying session histories
- replay should be a natural part of design
- ctrl-Z allows you to navigate and go back, but you can view/search/send
* disseminating results
- journal publication is slow, email doesn’t help too much
- I would like to send emails to referenced authors (~200) and send paper for review, confirmation
- I would like to know people who cite, download papers and communicate with them

(from “Creating creativity: User interfaces for supporting innovation” – ACM TOCHI, March 2000)

SideViews (Terry and Mynatt 2002) – instead of trial and error, get a picture of many versions at once to select

Guidelines for creativity support tools
- support exploration & collaboration
- support many paths and many styles
- low threshold, high ceiling and wide walls (… and more)
- initiate by exemplars, templates and processes

Evaluation methods:
- ethnographic observational situated … multi-dimensional, in-depth, long-term, case studies (domain experts doing their own work for weeks and months) -> MILCs

MILC
* evaluate socialAction
- focused on integrating statistics and visualization
- 4 case studies, 4-8 weeks (journalist, bibliometrician, terrorist analyst, organizational analyst)
- identified desired features, gave strong positive feedback about benefits of integration
* paper accepted to CHI 2008 (Perer & Shneiderman, 2007)
- very interesting visualization showing senators with bills they voted together … first separates into party lines, then shows cohesion among Democrats as the number of supported bills gets higher
- discovery by Chris Anderson, user of the tool (US News and World Report)

Creativity challenges:
1) evolve new theories and evaluations
2) understand creativity across disciplines

propose innovative:
- individual (creativity support tools)
- group (socio-technical environments) -> need to think about number of collaborations, there are some special things that make social networks succeed

great room for HCI and computer scientists

HCIL 25th Annual Symposium – May 29-30, 2008 (University of Maryland)

Science 1.0 -> reductionist, controlled experiments, replicability, laboratory, natural world … hypothesis testing, predictive hteories, replications
Science 2.0 -> integrated, case studies, validity, situated, made world … hypothesis testing, predictive theories

Abstract:
Creativity Support Tools is a research topic with high risk but potentially very high payoff. The goal is to develop improved software and user interfaces that empower diverse users in the sciences and arts to go beyond productivity and be more creative. Potential users include a combination of software and other engineers, diverse scientists, product and graphic designers, and architects, as well as writers, poets, musicians, new media artists, and many others. Enhanced interfaces could enable more effective searching of intellectual resources, improved collaboration among teams, and more rapid discovery processes. These advanced interfaces should also provide potent support in goal setting, speedier exploration of alternatives, improved understanding through visualization, and better dissemination of results (demos will be shown). For creative endeavors that require composition of novel artifacts (computer programs, engineering diagrams, symphonies, animations, artwork), enhanced interfaces could facilitate rapid exploration of alternatives, prevent unproductive choices, and enable easy backtracking. This talk provides a framework for systematic study of creativity. Two key issues are (1) Formulation of guidelines for design of creativity support tools (2) Novel research methods to assess creativity support tools. These issues were addressed at the June 2007 Conference on Creativity and Cognition (http://www.cs.umd.edu/hcil/CC2007).

Biography:
Ben Shneiderman (http://www.cs.umd.edu/~ben) is a Professor in the Department of Computer Science and Founding Director (1983-2000) of the Human-Computer Interaction Laboratory (http://www.cs.umd.edu/hcil/) at the University of Maryland. He was elected as a Fellow of the Association for Computing (ACM ) in 1997 and a Fellow of the American Association for the Advancement of Science (AAAS) in 2001. He received the ACM SIGCHI Lifetime Achievement Award in 2001. Ben is the author of Software Psychology: Human Factors in Computer and Information Systems (1980) and Designing the User Interface: Strategies for Effective Human-Computer Interaction (4th ed. 2004) http://www.awl.com/DTUI/ . He pioneered the highlighted textual link in 1983, and it became part of Hyperties, a precursor to the web. His move into information visualization helped spawn the successful company Spotfire http://www.spotfire.com/ . He is a technical advisor for the HiveGroup and ILOG. With S. Card and J. Mackinlay, he co-authored Readings in Information Visualization: Using Vision to Think (1999). His books include Leonardo’s Laptop: Human Needs and the New Computing Technologies (MIT Press), which won the IEEE Distinguished Literary Contribution award in 2004.

Kyle Gabler’s Human Brain Cloud is a masterful time suck.

This massively multi-player game has been in the to-be-blogged bin for a while. Thanks to Josh Catone of the great blog Read/Write Web for profiling the web site on Tuesday, bringing it back to mind.

Although billed as multi-player, HBC is really an individual exercise—one that can quickly turn minutes into hours. You see a word or short phrase from the 530,000 already suggested by other players, and you enter the first response that enters your head. Although you type alone, you benefit from the wisdom of the masses. The word associations you make are scanned for matches and listed with the rest of your recent work. You can view the cloud, which is really a visual thesaurus kind of weighted network showing the connections to a given term.

There are also statistics showing leaders among both word associations and participants (most of whom show up as the default “human####” identifier one gets when failing to notice or ignoring the field to change your user handle). Stats are also kept on how you match up with the rest of the collective in terms of originality or conformity. The current record for a single session is 7321 associations by user I_have_no_life, who was 56% original.

HBC was a distraction project for creator Kyle Gabler while he helps finish another game (World of Goo) due to be released by 2Dboy in February. The network started from a single word—”volcano“—and has grown to a dictionary of more than a half-million terms and ten times that many connections. Every association made either strengthens an existing connection or adds a new node to the network. I am now one of 356,025 participants contributing to the experiment, although that number likely includes a percentage of return players, too.

Gabler includes a disclaimer about the value of the data—”This isn’t academically rigorous or anything, so set your expectations accordingly”—but there are potentially some interesting data mining and design inquiry that could benefit from the Human Brain Cloud as a resource. In his review, Catone pointed to some related academic work:

Researchers at the University of California recently conducted a study in which they found evidence to suggest that our brains catalog and rate the relevance of information by forming connections between data. The researchers compared the brain’s system to Google’s PageRank algorithm, but there are obvious similarities to the massive word association map that the Human Brain Cloud is compiling as well.

Last July, Gabler published some early statistics when the network was about one-fifth its current size. The Human Brain Cloud has the potential to produce enormously valuable information. Marketers could conceivably enter a term and check the related words, in part to avoid ambiguity and semantic competition, and partly to steer clear of easy targets for derogatory associations. Designers might peer into the brain of specific users to see how they make sense of certain words.

Although Twitter has not yet issued a similar challenge, their open API is already being used by some developers to examine the information stream in new ways.

Paul Rand describes the language of form

The language of form
Late graphic designer Paul Rand is best known for his corporate logo work for entities like Westinghouse, UPS, ABC, Next Computer, Yale University, Cummins Engine and IBM. A tribute film, archived on Adland, was created for his posthumous induction to the One Club Hall of Fame in 2007. In the four-minute short film, Rand talks about the relationship between content and form:

When you say design, everybody has their definition that doesn’t correspond to yours. There are many good definitions. One is the synthesis of form and content. In other words, without content there is no form, and without form there is no content.

A work of art is realized when form and content are indistinguishable. When form predominates, meaning is blunted. But when content predominates, interest lags. The genius comes in when both of these things fuse.
source: transcript of One Show – Paul Rand Tribute Film (2007)

Rand also goes on to describe the language of form: order, variety, contrast, symmetry, tension, balance, scale, texture, space, shape, light, shade, and color. These are the core building blocks that go into visual design. While the Twitter community has not yet reached a point where this language has been mastered, the more effective visualizations that are out tend to use these blocks to create or reveal new meaning.

Understanding the value of tweets
When it comes to Twitter, the content are the 140-character posts about the details of life—tweets—and the form is the information stream. The nature of the stream points to opportunities for good visualization to add value.

Microblogging content for a single individual comprises nothing more than a longitudinal diary. For an author filling her own stream, everything is already known before it is posted. There is no new information to be gained. The relevance is high, but the information entropy is low. By comparison, the public stream—containing all member tweets—is noisy and lacks context. Since almost every tweet contains new information, the entropy is high. However, the relevance is low.

The sweet spot in between is the personal information stream, comprised only of a member’s own tweets and those of the people she chooses to follow. Particularly when the in-degree (followers) and out-degree (following) are comparable, the tweets in the personal information stream are highly relevant with high information entropy. In other words, any investment in time to acknowledge new information will be seen as worthwhile. This is where most users spend their interaction resources on the site, whether by checking the Twitter web site or using a third-party access tool, like Twitterrific.

The biggest opportunities for visualization, therefore, involve finding ways to draw new information out of one’s own known personal tweets and extracting relevant information out of the torrent of public tweets.

TwitterMatrix views recent tweets from a Neo perspective.

Drawing new information
Among the earliest visualizations of the public timeline was TwitterVision. It uses a world map with the location information in the authors’ profiles to situate tweets in geography. TwitterFaces, Twitter Planet and Twitter Earth do this as well, albeit with different map platforms. Twitter Map places a permanent pin on Google Map, showing only the most recent tweet of each member. This was interesting because it leveraged an available but hidden bit of information—author location—and presented it in a way that was more engaging. Reading the individual tweet content requires a lot of filtering to find the relevance, but it is inherently easier to identify with a place. The map visualizations, though, require constant attention in order to benefit from the revealed information.

Twitter Poster is an example of an attempt to create new information through visualization. This site keeps track of the most central members of the community and uses their profile icons to create a poster graphic. Different sizes are used to indicate how important the author is in the twitosphere. This aggregation of user statistics has been done in several ways by other services, but those all use a spreadsheet or table presentation rather than letting the top member images do the talking.

Another example is Twitterverse, a web site that generates a tag cloud reflecting the current pulse of recent author posts. Twitterverse creates such clouds for both single word and two-word phrases. These kinds of visualizations invite short user sessions with return visits, rather than being something to monitor constantly.

In-house development by Twitter is limited but high quality. In late August, TechCrunch got early screenshots of Blocks, a Twitter visualization created by people who worked on Digg visualizations. Blocks displays the local neighborhood of personal information streams that connect to recent posters among those you follow. The top surfaces of the stacks are color-coded to differentiate between you and your followers. By clicking and dragging the screen, the entire block structure can be spun and examined from all angles. Each branch can be traversed to see the individual tweets of your neighbors, and other twitterers can be clicked to regenerated a new stream neighborhood built from that member. Blocks is a highly interactive tool for discovery of both content and other members.

TwitterMosaic uses member profile images as material for mosaic art.

Relevance of content
There are several services and resources leveraging the API access to improve personal relevance in the noisy public timeline. Twitter only allows filtering of the public information stream in one way: by deciding who to follow. Absent are ways to arrange content for specific contexts by building streams for specific groups or topics. That task is left to third-party development.

Twitter Digest allows personal information streams to be created from a list of usernames. For instance, the Makice family has a digest that is being used to help archive our microblogging by creating a daily snapshot of what we posted. This service addresses the issue of grouping users and control of the timeframe for viewing content, but otherwise it doesn’t add much value over the following function already in Twitter.

TwitterWhere, on the other hand, creates RSS feeds based on location of the authors. There is no way to do that through the Twitter site tools. Following Bloomington, Indiana, for example, is a way of discovering new local users, provided member profiles are first updated with a valid location. The in-house tools that come closest to providing this functionality are search and tracking. Search is a manual process that forces one to wade through pages of results to look for new matches. Tracking is currently only alerting a member to keyword matches through an IM channel, which is typically not the preferred means of interacting with Twitter.

There are a number of tools focusing on the text of the tweets. Initially launched in early October, Random Tweets uses a combination of computer randomness and human oversight to identify the ten best tweets of the day. TwitterBuzz and Twitigg look at the links authors embed in their posts, aggregating the appearances of popular URLs and allowing them to rise to the top of the list. Twitigg extracts links, resolving the shortened URLs first to show what links are popular by the hour and day. The site also separates forms of media. TwitterBuzz is largely viewing noise, however, given that TinyURL—the built-in integration that shortens long URLs posted in tweets—shows up at the top of the list.

TweetVolume is a simple bar chart comparing the frequency of keywords in tweets.

Situated visualizations
Artist Ben Rubin and UCLA professor and statistician Mark Hansen created a multi-media permanent art installation for the The New York Times headquarters building on Eighth Avenue and 41st Street. Dubbed “Moveable Type,” this visualization is about the physical space as much as the content flowing on its 560 small screens.

Two walls of monitors have been programmed to display snippets from both the Times archives and live feeds going into publication that day. Casual passers-by may notice the refreshing of text in patterns without understanding how the content relates to each other. That kind of understanding is only possible by moving in closer an spending time with the screens. There is also a clickety-clack of old teletype machines, bringing a notable ambiance to the newspaper offices. Because the content reflects the activity, both historic and current, taking place in the building, there is a sense about the physical space that might not be understood without the visual and audible presentation.

Twitter development is possibly moving in that direction. The high relevance-entropy of small streams is ideal for connecting local communities in a meaningful way. These connections will undoubtedly take place in different locations around the same town, perhaps revolving around smaller groups of Twitter authors posting from the same physical spaces. Visualizations—like the one we are currently designing for the IU School of Informatics—will have the dual role of connecting the larger community through the exchange of status awareness and imbuing a specific building with an identity unique to the collection of people working there.

Situated visualizations are meant to be seen by those attached enough to the room to visit, and therefore both the content and form should aim for the highest possible relevance for that particular context.

The New York Times display, Moveable Type, pays attention to physical spaces.

The Human Calendar involves 40 of Craig’s friends roped into a photo shoot in his garage, capturing a total of 3,992 pictures. Each face is recorded next to a number or label helping comprise the calendar. The gaze of every person will follow the current date. The site also offers some widgets to bring this unique presentation to web pages everywhere.


Tell time with human analog and digital clocks

The Human Clock is a collection of individual photographs containing the current time information. There are multiple photos for each minute, rotating the images in some random fashion. The analog version merely represents all 1440 minutes in the day, but the digital clock consists of 15,356 photos. Fans of the clock have expanded on the initial set of photos to include images that include appropriate numbers captured in the wild. Some times have acquired additional cultural interest, and the clock photos even became a communication channel.

At the risk of re-inventing the same wheel, this approach to time might be useful in community building. Particularly in a university setting, where students and faculty come and go, a personalized version of the Human Clock and Human Calendar on display in building lobbies or on desktop monitors as screen savers might go a long way to improving recognizability and connection in a small program. I can imagine a photo shoot conducted as part of the incoming student orientation week at the IU School of Informatics, substituting the departing faces with new ones.

The Human Clock project has some social network cache, with a presence on MySpace. Craig—who has a fine appreciation of quality archival computers—has covered his bases with the domain name, too, having registered a similar but distinctive URL in case of a typo.

See also: the other human clock.

Human faces gaze at the current date in the human calendar

McLeod’s simple rubric for assessing technology usage concentrates on two dimensions. First, the level of technology usage can be visualized as a circle, where larger circles equate to greater use. This is determined for two key domains, home and work. Second, the use of technology in different domains sometimes overlaps to varying degrees. The differently sized circles can be arranged to reflect a self-assessed pattern of use.

Some examples of how the two dimensions of size and overlap can paint a picture of individual use of technology.

Use of diagrams in surveys has potential barriers and enticements. One likely problem might include people being unsure about how to express themselves in a non-traditional manner. Surveys usually look like a page of checkboxes. Asking for a drawn or assembled diagram may be intimidating. One the plus side, though, there is more information to gain from watching how a person assembles such a diagram. In the above case, there would be an early decision made to assess either home or work domains first. The order may contain important information that would not be captured in a Likert scale. Also, turning the data into diagrams is a designerly way to communicate the results anyway, so having the participant compose the data visually might save a step.

I have a user study for Kosmix planned for Saturday at Bloomington Bagel Company, as I try to buy about 10-15 minutes of paper prototype assessment from 15 non-techie patrons (each participant gets a $5 gift certificate to BBC). As I make final preparations, I have decided to try and incorporate this diagramming technique into the simple demographic survey each person will complete.