I usually think blog memes are pretty lame–if you can’t think of anything to write, how about just not writing?–but this one, spotted at Blackbeltjones, caught my imagination:

What are the titles of the posts you have in your draft folder of shame? The things where you’ve just thought of the title, but written nothing to back it up? The momentary points of self-deluded genius that in the cold light of day you thought better of?

Here’s mine:

Think Tank: Greeting Big Brother With Open Arms
Spontaneous Screen Videos
Visual Radio
Video Sniffing
braintag: why i don’t think videoblogging is ready for alpha release.
The Bastard Child of Art and Commerce
Movieoke
Ghetto Brawls
Fully Digital Hollywood
Me Too!

Kent Bye says, “Playlists are to Music as Edit Decision Lists are to Film.” Yep, exactly.

More and more of my important data is stored on the web: my bookmarks at del.icio.us, my citations at CiteULike, and my notes at Backpack and Writeboard. I love these services, but I am fully aware that they could disappear at any time, taking my precious data with them. Fortunately, they all make it easy for me to back up my data to the location of my choice (in this case the SIMS servers, which are themselves backed up regularly).

Backing up Backpack data:

#! /bin/sh

curl -s -H 'X-POST_DATA_FORMAT: xml' 
    -o "/home/ryanshaw/backup/backpack/$$.xml" 
    -d '<request><token>[My Backpack API token]</token></request>' 
    http://rybesh.backpackit.com/ws/account/export

Backing up CiteULike data:

#! /bin/sh

URL="http://www.citeulike.org/bibtex/user"
DIR="/home/ryanshaw/backup/citeulike"

cd "$DIR/ryanshaw"
mv -f *.xml backup/
curl -s "$URL/ryanshaw" | /home/ryanshaw/bin/bib2xml -s 2> /dev/null

Backing up del.icio.us data:

#! /bin/sh

curl -s --user rybesh:[my password] 
    -o "/home/ryanshaw/backup/delicious/$$.xml" 
    http://del.icio.us/api/posts/all

Backing up Writeboard data:

#! /bin/sh

url="http://123.writeboard.com"
dir="/home/ryanshaw/backup/writeboard"
cookies="$dir/cookies.txt"

# args: writeboard_id, password
export_writeboard () {
    [ -d "$dir/$1" ] || mkdir "$dir/$1"
    curl -s -L -c $cookies -o /dev/null -d "password=$2" "$url/$1/login"
    curl -s -b $cookies -o "$dir/$1/$$.html" "$url/$1/v/export?format=html"
    rm -f $cookies
}

export_writeboard [writeboard ID from URL] [writeboard password]

The final day of the main program of ACM Multimedia was satisfying enough to convince me to try to come back next year. Highlight of the morning was a presentation by researchers from Japan’s National Institute of Information and Communications Technology, who demoed a system which analyzed chat logs to create highlight summaries of live broadcast events. Basically, they looked at chatroom activity during live broadcasts of American football games, and looked for spikes in the rate at which messages were posted. They also recognized emoticons to determine positive or negative affect. They used the resulting data to pick out segments of the broadcast program which corresponded to these positive or negative spikes. By doing so they could create overall highlight summaries or ones geared to particular social groups (my team’s highlight may be your team’s lowlight). A simple idea, but not one I’ve seen actually implemented before.

Photo courtesy of Ramesh Jain’s Flickr feed.

In the afternoon Hongjiang Zhang of Microsoft Research Asia, Ramesh Jain of UC Irvine, Alberto Del Bimbo of the University of Firenze, my advisor Marc Davis, and Rainer Leinhart of the University of Augsburg (shown from left to right, not including Dr. Zhang) debated about the foundations of multimedia computing. This was interesting to me not so much because of the what was being discussed (though that was interesting too), but because it was an opportunity to observe the dynamics of a particular research community. Like my “home” field of information science, the multimedia community struggles with the challenges and opportunities of “interdisciplinary” work, and many of the concerns raised and possible solutions offered at the panel echoed things I hear around SIMS. Ramesh also commented on the panel at his blog, suggesting that the multimedia community is in its “teenage” years.

Later I had tea at the Raffles Hotel with Frank Nack, Dick Bulterman, Wolf-Tilo Balke, and Milena Radenkovic. I was especially glad to have had the chance to talk to Milena a bit about her projects at the Mixed Reality Lab, including a new effort to involve millions of cameraphone users as frontline collectors of scientific data. And of course it was great to have a chance to catch up with Frank again.

All in all it was a good conference and worth the trip. I’ll definitely try to make it to next year’s conference in Santa Barbara, hopefully to present a long paper…

Well, Day 2 of the conference wasn’t nearly as interesting as Day 1, unfortunately, and I don’t really have any particular highlights to point to. I did however have a nice long chat with Lynda Hardman of CWI, who in addition to being involved in some fascinating “multimedia meets the Semantic Web” research also turns out to be very cool. Oh, and the Garage Cinema Research MMM2 video won the award for Best Video of the conference.

I’m currently in Singapore for the ACM Multimedia conference, presenting some of my work on anime music video editors (“Toward Emergent Representations for Video”) and Organum (“Individual Presence through Collaborative Play”). Today was the opening day of the conference, and already I’ve met enough interesting people to have made the trip worthwhile.

This morning was the Brave New Topics session on Multimedia Challenges for Planetary Scale Applications, which took a very broad view of the design space around multimedia sensor networks. Phil Gibbons of Intel Research kicked things off by discussing IrisNet, an open-source software infrastructure for turning ordinary networked webcams into global monitoring systems. Next a representative from Nottingham’s Mixed Reality Lab discussed the challenges raised by pervasive gaming.

What struck me about the these two presentations was how closely the IrisNet system architecture mirrored the social structure of pervasive gaming events proposed by the Mixed Reality Lab folks. IrisNet separates nodes into a large number of “sensors” and a smaller number of “organizers,” with the former gathering data at the edges of the network and passing it up to the latter for further processing. Similarly, the MRL designers see the roles of pervasive gamers as a pyramid structure, with wide but shallow public participation at the bottom, a smaller number of more engaged interest groups or enthusiasts in the middle, and a small number of experts at the top. Clearly this model applies to more than just pervasive games: similar patterns have been reported in open source software projects and Wikipedia. Even unmediated.org can be viewed in these terms: I am a del.icio.us-addicted “sensor” madly recording all that interests me, while Kenyatta is an “organizer” curating feeds for wider consumption.

The highlight of the afternoon was the presentation by Atau Tanaka of Sony CSL Paris, who talked about his work on Malleable Mobile Music and a web-based tool for collaboratively remixing Creative Commons-licensed audio. I had read about Tanaka’s work before, but he was even more interesting in person and seems to be deeply interested in the social dynamics of groups interacting with media–somewhat of a rarity at ACM Multimedia, unfortunately. Definitely one to watch.

Garage Cinema Research and Yahoo! Research Berkeley were well represented the first day as well: in addition to my posters, there were tech demos of PhotoRouter and Photo LOI and a video presentation of the MMM2 project. In the evening Yahoo! paid for beers at a poolside reception in which we madly tried to recruit the best and brightest of the ACM MM crowd, a shameless gambit which will hopefully pay off in a great crop of interns next summer.

I’ll be posting highlights from the remainder of the conference over the next few days. In the meantime you can check out the Flickr photostream for the conference, which is looking a bit anemic at the moment, but should fill out as people get some time to upload their photos.