New York landscape populated with status updates

The visualization continues to take shape (see these earlier posts for context). We are now parsing live data from Twitter and image tiles from the Google Maps API for the surface mesh. The user interface remains the point of focus at this point in time, though we are beginning to look for data parsing solutions to help construct the semantic pathways between status updates. Below are a few images of the latest progress.

Close-up of landscape populated with updates

Narrative pathway connecting several updates

Narrative pathway connecting several updates

Sketch of geolocated tweets, manipulating the mesh surface below

Joining the project is Liangjie Xia, a programmer and media artist based in New York and a recent graduate from ITP.

The current project focus is on a data landscape comprised of geocoded social updates, forming narrative pathways according to themes (we are evaluating Twitter, Flickr and Foursquare as data sources). As they occur, updates add points to a basemap of Taipei, cumulatively changing the elevation of the landscape. Updates, represented as nodes, are selectable, and articulate the landscape based on other thematically related updates. Finally, we are exploring toggling between two views: the surface mesh outlined above, and a view exchanging the mesh for narrative pathways, represented by hairlines connecting nodes related by topic and time.

Below are a few early explorations that begin to depict the direction.

Concept diagram of thematically-related social updates on a basemap of Taipei

Early concept model of a narrative pathway, utilizing the z-axis

Side elevation of surface mesh wireframe

Sketch of chronologically-connected, geolocated Tweets

Articulated surface mesh mapped with satellite image of Taipei

“Memories are often linked to places. Through data visualization, I am proposing to explore the connection points between collective memory and their geographic context, relative to the city of Taipei. Collective memory could unfold as a time-based narrative, constantly changing yet always representing a continuum. Places are mapped to the memories they correspond to, in the process resulting in a new kind of city geography in which the urban landscape is reframed through narrative. In this new representation of Taipei, places will exist in the proximity of other places, related not through geography but through shared thematic associations. As it unfolds, the landscape could reveal new points of intersection and invert our expected view of the city.

The piece will utilize existing social platforms, such as Twitter or Plurk, to gather geocoded written statements that will be added to a growing database of city memories. The resulting visualization will continue to grow in specificity and complexity as more data is added to the system.”

Below are a selection of data visualization projects that seemed relevant for our process.

Cover of Guide Psychogeographique de Paris (Guy Debord)

Emotion Mapping (Christian Nold)

Pastiche (Christian Marc Schmidt and Ivan Safrin)

Digg Labs Big Spy (Stamen)

Twitter visualization of Obama’s inauguration day

We Feel Fine (Jonathan Harris)

History Flow (Martin Wattenberg)

Gothamberg (Martin Wattenberg)

Cabspotting (Stamen)

Many think that the biggest competition to search is presented by social networks. Mark Pincus of Zynga talks about a public and a private web, with an unauthenticated mode addressing a general public, and an authenticated mode based on individual social networks. Websites offering a password-protected, personalized experience will increasingly need to take advantage of social networks to incentivize people to sign up, either by creating new social networks or leveraging existing ones. This evolution will simply serve to further integrate content providers and social platforms, making it easier to recommend and share content with one’s followers or friends. Here, we ourselves are becoming curators for our social network—an authority regarding content that others might find relevant and interesting.

This makes search less important than it used to be, though it still remains the primary authority for the public web. Of course, there are many authoritative and reputable brands among online content providers, from traditional media companies that have translated their offering online, to new content providers that have emerged in the online space, to individual bloggers. The sheer breadth of available content explains why we are increasingly looking towards our social networks to help us regularly digest information. But it also explains why search engines have established themselves as premier authorities for content in the realm of the public web.

Clay Shirk recently wrote a piece about what he calls algorithmic authority: “Authority thus performs a dual function; looking to authorities is a way of increasing the likelihood of being right, and of reducing the penalty for being wrong.  [...] Algorithmic authority is the decision to regard as authoritative an unmanaged process of extracting value from diverse, untrustworthy sources, without any human standing beside the result saying ‘Trust this because you trust me.’”

The first company we think of as having established and capitalized on the notion of algorithmic authority is Google. Google has built its reputation on a perception of neutrality via its PageRank algorithm, which has succeeded phenomenally in providing relevant search results. So while individual recommendations are important relative to the private web, the algorithm has become an equivalent authority for the public web.

Seen in this light, it is no stretch to think of search as moving from a mere navigational tool, to becoming an editorialized space for surfacing content, within the framework of the search algorithm. Search engines are in the unique position to feature specific topics that aggregate a multitude of sources, lending it potentially higher credibility than any single source could assume.

Google News was one of the first attempts at creating an editorialized space by leveraging its search algorithm. A more recent development is Google’s Living Stories, which combines news articles from various sources, over time, to form complete coverage on a particular topic. As Living Stories shows, editorialized search is primarily about aggregation and combination, rather than the decontextualization (the traditional search model).

Microsoft’s Bing features particular pieces of content related to each-other through search queries. The daily image on its homepage links to various search queries corresponding to relevant content types (web sites, videos, images, maps, etc.), showing for example the location of the photograph on a map, in addition to a Wikipedia article and an image search on the subject. Bing’s Visual Search involves linking objects, related by object type and various attributes, to specific search queries, in essence allowing their comparison through quantitative visualization. What makes this approach novel is that it provides an editorialized framework for corresponding information from various sources, via the search algorithm.

While these are all early steps, they compellingly demonstrate the potential of search as a content platform, and point to the future of search engines as providing an alternative, authoritative source to individual recommendations.

In respect to the first mode, the design of websites appears to be converging towards certain conventions, from the typical 3-column layout that most blogs adopt, to arrays of content represented by lists (Craigslist, Delicious) or grids (YouTube, Flickr). Many sites use any combination of the above. One could say that these conventions are the result of the attributes of certain established formats, such as the web browser, the particularities of HTML, and predominant screen sizes and resolutions. Other formats, such as mobile phones and touch displays, have accordingly resulted in their own conventions.

With these conventions comes a certain genericization of structure. One instance is the homogeneous grid, used for arrays of normalized content. The grid has become the default solution in any situation where page real-estate, flat hierarchies and the automation of content are relevant factors. With these considerations in mind, it is hard to imagine a more functional vehicle for content delivery. A grid of normalized content pieces also allows for quantitative visualization opportunities, demonstrated effectively by Pivot, a recent project of Microsoft Live Labs.

Pivot

The homogeneous grid aside, other automatable patterns exist in situations where the content is more nuanced and hierarchical. Good examples are the Times Skimmer and the Times Reader, both of which emphasize featured articles in a flexible grid arrangement.

Times Reader

Grids and lists are a pure expression of a development that I think of as object-oriented design, in which taxonomies are derived from recurring content elements, for the purpose of automation. These forms of organization pertain when all content elements are of equal weight. Not coincidentally, they are also the way in which we are accustomed to viewing content in the case of aggregation—the second mode of content consumption.

One could make the argument that the influence of RSS and search engines, through their respective ease-of-use, has in fact provoked a shift in the design of websites towards normalization, which is to say that the design of websites is converging according to the organizational principles established by aggregation software.

If website structures are becoming more generic, wherein lies the role of design relative to the web?

We are, of course, accustomed to websites existing in various states and levels of detail. Many websites are already platforms with multiple touch points—the site itself; the RSS feed; the mobile or desktop application. Website design could increasingly revolve around individual content elements, rather than the larger, aggregating framework of the site or platform itself.

A “widgetization” of websites would be consistent with the decentralizing tendencies we are already observing, and would have the potential to preserve the identity of the content source at a micro-level, down to the individual piece of media. Of course, this is already happening in some areas: Embeddable content often conveys the identity of the original source. One has only to compare an embedded video from YouTube with a video from Vimeo or similar service. Tweets are also differentiated by design, due to their 140 character limitation.

Applied to all websites and forms of content, this change in focus could become truly paradigm-shifting, allowing for a decentralized content presence without a cumbersome and interchangeable main corpus. Might this in effect lead to a design strategy for linked data?