.comment-link {margin-left:.6em;}

Sometimes I Wish That It Would Rain Here

Friday, February 23, 2007

where lies the uncanny valley?

a while back, when the PS3 was first being launched, I had a bit of a rant about the over emphasis of graphical fidelity in games and lack of emphasis on the behavioral components. I think that the comparison between the sales of the Wii and PS3 are a testament to the fact that gameplay ultimately trumps graphics and pretty volumetric lighting. however, because making faster, better, prettier graphics is relatively easy, as compared to making dynamic characters that truly interact with the player, people are doing just that. in a recent BBC article, David Knunkley of Obsidian Entertainment says that game characters are currently in the uncanny valley; "they're too close to real, but not quite real." graphically, I think he's probably right.

however, I would argue that the horizontal axis of "likeness" in Mori's uncanney valley graph is not a single dimension but is very high dimensional. it certainly involves visual appearance, but there's a lot more to "humanness" than appearance. the ability to interact with characters in a non-scripted way, the ability of characters to have believable emotional reactions both to PCs' and NPCs' actions, the ability of the character to make autonomous decisions that can impact the player, etc. in some regards, this begs the question of how close games lie to an interactive narrative. are you playing through a story, or is a story evolving based on the ways that you interact with the characters? furthermore, is humanness the quality towards which character developers should strive? consider the success of the WII, powered by what is essentially an overclocked Game Cube processor. the graphics are far from stunning, and HD support is limited, and yet you can easily acquire a PS3, while getting a hold of a Wii is a slight bit more challenging. point being, super amazing, photorealistic graphics are cool, but they are not the end all and be all of games, and the Wii's success is just one example of a focus on developing new player experiences.

Tuesday, February 06, 2007

notes from Marc Davis

last week, I went to this panel at the HumaniTech conference on Text and Image : From Book History to “The Book is History.” the panel included talks from Daniel J. Clancey on Google’s book search, David S.H. Rosenthal about Stanford Library’s LOCKSS project, Marc Davis on Yahoo! Research’s projects about social mobile computing, and Ramesh Jain on organic books. unfortunately, I didn’t catch Ramesh Jain’s talk, but I did get to see the others, and all were quite interesting, if for very different reasons. I was actually struck the most by Marc Davis’ talk, because he essentially got up and said, “all this talk about text and books is cool, but we’re ignoring the ‘text’ [in the post-structuralist sense] that people author everyday of their lives as they create temporal patterns from their movement through physical and social space.” (this isn’t an exact quote, but it’s largely what I took away from the talk). see as my research of late has been focusing on some computational linguistic stuff, which is obviously about text as written words, I must admit I had a bit of a gut reaction to this. on the one hand, I think the stuff Marc Davis and others are doing on mobile social computing is not only fascinating but really important. I agree that we should be hesitant about the amount of importance we place on written texts; there’s certainly a lot about physical embodiment that they don’t capture. however, at the same time, written words as text are not going away, and I don’t think in our newfound fervor for traces spatio-social-temporal traces we should ignore textual analysis.

what I’d like to do here is just post my stream-of-conscious style notes from the Marc Davis’s talk, partially to summarize and partially as an exercise for myself to go back over his talk. I have a bunch of comments about this, but I wanted to contextualize my comments by giving a synopsis of the talk first. however, trying to do that all in one post would be a bit too monstrous. so, for now, here are my notes from the talk, with my analysis soon to follow.

Marc Davis – Social Media Guru from Yahoo! Research Berkeley
studied with Wolfgang Iser, who wrote The Act of Reading
shaped by post-structuralism and reader response theory
how many people read or wrote text today? how many people looked at multimedia (photo, audio, and/or video) today? how many people authored multimedia today? trying to change that, so that authoring multimedia is just as much a part of daily life as is authoring text.
image and text are very related
Eisenstein (“father” montage) – a sequence of images as a form of writing
[computer] programs as a form of writing, computational forms of writing
moving from garage cinema (title of a previous ACM multimedia paper) to social media
not a web of documents or text
it’s a book “made out of people”
this is an entirely different set of business models, where in your customers are your suppliers
Yahoo answers – you don’t want to search for some string, you want somebody to answer your question. trying to connection question askers to answerers. connect everyone in a large knowledge network.
citing a figure about the amount of information on the web and the number of people in the world, “you have more than 10 stories to tell.” connecting people via the stories the tell.
“not just information, but knowledge.”
from cybernetics to social media
cybernetics designs a network to connect human and computer elements
HCI comes in part from human factors, focuses on the human part of the network
the “PC interlude,” a focus on the interface between a single computer and a single user
social media, web 2.0, billions of people, 2 billion cell phones, all connected together in a large cybernetic type network
the challenge is how to incentivize human participation in this network
citing B.J. Fogg, how do we use mobile phones to influence human behavior?
with the time and energy people collectively spend playing solitaire online everyday, we could build the empire state building. how do you harness this human and rhetorical power?
there’s a lot of talk about the internet, but there’s also this exciting thing called the world that’s going on
the world and the internet coming together
cell phones in a data and media network
more phones being sold than laptops
to think about humans and media, you have to think about phones
not just media consumption, but media production – the camera phone is the most important invention since cinema
it’s also a sensor, from which we can gather a wealth of data
by and large, a cell phone is a unique identifier for you as an individual human
Roland Barthes – “From Work to Text”
there is a binarism to our notion of what text is and what language is
a book itself is not real, the process of making the book real happens between the book and the mind
text is only experienced in an activity of production
what is in the library? is there knowledge in the library? is there information in the library?
the text is plural
text is a social space that coincides only with the practice of writing
“From Text to Web”
makes the invisible visible
record explicit and implicit human activity, large scale activity
analyze and interpolate to predict about missing information
people like to believe that they are unique, but sociologists know that we’re not all that different and that we’re very predictable, especially in groups
Yahoo logs 22 terabytes of data daily in user activity alone
social media metadata – where + when + who + what
from phone, you can get spatial, temporal, social (who I am and, via Bluetooth, who’s around me)
construct a path of human attention through spatio-temporal-social space
you can graph this in a 3 dimensional space, and you get nodes of attention, knots of overlap
at this panel, we’re all here together at a time and place with a shared activity and shared attention
while you may be thinking about other things, you are “attending” to this talk here and now with these other people
these are now nodal points of human activity that can be found from devices
we know when, where, who, and what (but not why)
this is a new form of writing, of text, the paths of people through data space is a new form of inadvertent, implicit, subconscious authorship
but what is this collective document we are authoring, and what can be done with it?
time is linear, but human activity has periodicity and cycles
you can track periodic pulses and cycles in people’s lives via the photos they take
if you want to predict who’s in the photo, you’re better off looking at patterns of behavior rather than the actual image
a photo taken right now during this panel might likely include your co-workers but not likely your children
photo LOI (level of interest), capturing (shared) attention
understand what’s interesting/important in a photo – shared attention likely indicates what’s important
with space, we’re used to thinking about maps, but we can also do hierarchical structure
social network structure of human relations becoming visible
can automatically suggest tags for images based on spatio-temporal-social metadata
uniquely identified humans via their phones, uniquely identified events, predict events to automatically add metadata
almost 11 million geo-coded flickr photos
can show a map at different zoom levels and overlay tags used in that area
evidence of human attention
like the Jewish Book of Life, creating a book of life of billions of people on the planet
but is it a panopticon?
it’s creating a new text in the Barthes sense
flickr is the “eyes of the world” that can see each other
this is the text being written everyday that will be read by future historians

that's it for my notes. stay tuned for my comments.