Just read an intriguing market study from Mike McGuire and Derek Slater predicting that by 2010, 25% of legal music purchases will be driven by taste sharing mechanisms, including (especially?) playlist sharing. The report was jointly commissioned through Harvard's Berkman Center and Gartner, so it trends towards the business/marketing end of things, but assuming the survey methodology is sound, points at the potential for another Web native form to emerge.

There's money to be made in them there playlists!

Of course Webjay's Lucas Gonze has been pushing, to good effect, on sharing of playlists that point to free streams. The McGuire and Slater paper documents a number of music file and playlist sharing mechanisms within commercial efforts like iTunes (iMix), Rhapsody (Playlist Central), MusicStrands, Mercora, etc. etc. I want to be skeptical of the potential since making playlists is work. However, there's enough evidence that enough people are motivated to do this that it could take off, if it hasn't already and I just don't know it.

The tricky bit is how to write transmittable, but rights safe, playlists. If a playlist points to licensed music, and I send the playlist to someone who doesn't have the same license, what happens? I was perusing the XSPF format and this appears to me to be fraught with a few technical issues: canonically identifying tracks, resolving how to retrieve them, yadda yadda.

But the DJ in me has a fascination with the potential for playlists as another social media.

FeedBurner is pushing the envelope in terms of making individual feed items useful with their new product FeedFlare. Since FeedBurner proxys feeds for publishers, they can easily add elements to a feed item such as related links (according to Technorati) or tags from del.icio.us. I wonder how far this can be pushed, because while a feed item is generally rendered as HTML, aggregators aren't quite the same as web browsers.

One downside though is that aggregators need to be able to remove the FeedFlare stuff before doing duplicate detection. Seems like Bloglines isn't doing this and let's just say the results are a bit irritating. More work for aggregator writers if this becomes popular.

Eric Smalley's Technology Research News interview with Jon Kleinberg summarizes why Kleinberg is such a genius.

A love the term usable exhaust which I ran across at Hypergene MediaBlog. The phrase captures the essence of the more academic, social navigation, in an accessible fashion with a dash of potential activity thrown in.

The subsequent analysis is a bit lacking though.

First off, Chris Willis and/or Shayne Bowman correctly point out that tagging isn't magic pixie dust, but as implemented tagging is extremely important as usable exhaust. In both del.icio.us and Flickr each tag emits an RSS feed, which makes it easy to monitor a topic's exhaust. Even better, in both systems you can watch a given person's tag, so you can actually focus on people (also groups in Flickr) that you care about. Plus, you can discover new tags through the feeds as well. The del.icio.us popular page is probably a bad example as it often includes stale links and for many del.icio.us users the popular URLs aren't all that useful.

Also, the veneration of APIs is undeserved. In both systems, the number of applications that use the API, introduce users to each other, and make it easy to find new, interesting content is vanishingly small. Of those I bet they are used by a relatively small minority of the community. Heck, about the only thing the del.icio.us API is really good for is backing up your bookmarks, since you can only see your own and can't get any global information.

I could be convinced that the Butt-Brush factor is a real issue, but I'd counter with the amazing number of ways social connections can be made in Flickr if you choose to be public: groups, group photo pools, comments in groups, comments on photos, favorites, interestingness, contacts. Also, on every photo the tags, sets, and pools it participates in are visible and navigable, so the more exhaust you generate the more you grease the skids for other people to get sucked into Flickr. Finally, there's a white hot core of community on Flickr that probably generates the most traffic through socializing and providing the best content. Luckily Joe Just Get Photos Online can ease in with little hassle, but he makes up the weak periphery.

The authors are right that RSS is easily overlooked. They did it themselves. Again, RSS isn't just an add-on, it's an essential mechanism by which people can monitor the activities of others. It's how people see your exhaust!

Other than that, great post!

Link parkin': A little late to the party, but the September 2005 issue of ACM Transactions on CHI, edited by John Riedl and Paul Dourish looks to have some interesting papers on recommender systems and social navigation.

Nothing like a little light reading over the holidays!!

Now we've got rankings of blog networks. Given a relentless focus on advertising, where size does matter, will the commercial blogosphere eventually congeal into an oligarchy of of massive corporate concerns? Meet the new boss, same as the old one!!

Anyhoo, as these efforts to analyze largish (between 100s and 10s of thousands) sets of weblogs proliferate, I wonder if we'll start to see some open source tools for experimentation and independent verification. I'm not an open source zealot, but for independent researchers, it's hard to trust results based on closed systems. Maybe this year's Weblog Ecosystem Workshop will provide something in this area.

Tough crowd. Marc Hedlund, and commenters, aren't too positive on Aardvark'd, the film documentary of Fog Creek Software's, and by extension Joel Spolsky's, summer interns. Sounds like there's not a lot of Joel and not enough SW development, which is what many people were expecting.

I had some high hopes for it as promotional material for aspiring CS majors. I'll reserve commentary, but it still might be entertaining and useful to those who come to it without familiarity with Joel and his oeurve.

Apropos of Elle Driver: "You know I've always liked that word... oeuvre.. so rarely get to use it in a sentence."

That was fast. Yahoo! bought del.icio.us. Here's Joshua Schachter's announcement

Predicted consequences:

• Lookout for the Y! single sign-on and hail of stories regarding disgruntled users leaving

• Ads on del.icio.us
• Reduced downtime
• The long promised rollout of group/privacy features
• One more entry on the Y! toolbar

One thing that would be really entertaining is some integration with Yahoo!'s media offerings. Imagine Y! Music Engine with a good bookmarking service inside.

Just a minor followup on my next laptop given the the dead Thinkpad, because I know you were waiting with baited breath on the decision.

I've decided to go laptopless for a while. I'm something of a quiet contrarian which motivates the decision. Once upon a time, I'd be one of a small handfull of folks who'd pull into the cafe and crank away at getting stuff done. Now you've got gaggles of teenyboppers copping all the good tablespace, not to mention outlets, with 5 laptops for 3 people. I kid you not. Time to move away from the herd a bit so I can regain that smug feeling of superiority.

Also, I've been laptop primary for close to 5 years now, and a lot has changed for desktops in that timeframe. Heck, mongo flat panel LCDs are cheap now!!

Lastly, I really want to get back to MacOS for the next laptop and figure I'll slice a few fingertips off as a frontline adopter when the Intel based PowerBooks come out.

Oh and no smilies for the humor impaired.

Marketing wiz Seth Godin and a few friends have launched Squidoo, a site for "lensmasters". As best I can tell, Squidoo combines a number of well polished Web authoring techniques into a nice package that allows people to easily develop lenses, a.k.a. rubrics or precis, on specific topics. Toss in LensRank, a ranking mechanism of unknown origin at least to those of us not enrolled in Squidu, a clear focus on what people are supposed to do, along with the potential for lensmasters to receive AdSense revenue, and you've got an evolutionary step in web authoring past blogging.

For those who've got a lot to say, or bleeding edge types, they're probably better off just doing a good job maintaining a focused weblog+wiki combo. But for time constrained people who just want to cash in on their expertise Squidoo might be the way to go. I'm interested to see the "hundreds upon hundres of modules" advertised in the FAQ. Why the need for so many lens building blocks and how does a small company build so many?

Apropos of nothing, I'm not sure it's a good or bad sign that when my post-warranty, unbacked-up Thinkpad T41 (nice machine) flat croaked with only 48 hours of flakiniess, I had no serious thoughts of going ballistic on Lenovo, nee IBM.

I said a prayer for the harddrive's survival. I got out the tools, and extracted said harddrive. I put the harddrive in a $50 IDE to USB enclosure, picked up at a nearby Best Buy, and plugged the dang thing into my desktop. Didn't lose a bit of data.

The Thinkpad HW? The LCD, Centrino Chip, 768 MB SDRAM, Keyboard, Trackpad, Wi-Fi Chip, XGA adapter, CD-RW, 100 MB Etherenet chip, et. al.? Ain't nothin but dirt to me baby. I figure it'll make a nice doorstop, or gag machine. Oh my god, I dropped my laptop just before the presentation.

And there might be a little value in the DRAM, but other than that...

Here endeth the catharsis.

Link parkin: Ho John Lee's short list of social ranking technical papers, including a deep, comprehensive look at PageRank by Langville and Meyer

I'm enjoying Lucas Gonze's intellectual development of the lightnet concept. To summarize, lightnets are Weblike (based?) networks for the distribution, linking, and editing of open continuous media, e.g. video and audio. Gonze has more technical definition, a strawman, and some examples of what contributes to lightnets and darknets.

A key emphasis is on "open" in both the media and tools. There are plenty of challenges here, especially in the face of massive, deep pocket, incumbents who already have a stranglehold on much of our culture. But this is a commons worth championing.

Then again, as Jon Udell finds out in his continuing tilts against the windmill, technical issues with making video and audio of the web are not to be ignored either.

Somewhat stale, but a recent post on the official Yahoo! Blog points out that there are over 30 search shortcuts you can use on the site. I don't know how or if they combine well with advanced search, but that's a mini-programming language.

Ob. geek. Clearly it's not Turing complete and probably not formally develped but therein lies a challenge for an enterprising domain specific language builder. Develop a tiny, accessible, extensible language for expressing search queries. Majority of language power has to available from one-liners of less than 40 characters or so.

The first visible fruits of Adrian Holovaty doing his thing for The Washington Post have appeared. Working with Derek Willis, Adrian used Django to build an interactive database of congressional votes, House and Senate, going back to 1991. Over and above general availability, the site also lets you monitor your congress critt, errr, representatives, using RSS feeds. Brilliant!!

Crafty news organizations will increasingly see a big part of their business as developing high local profile (of course this is US national, but it's beneath Google/Yahoo!/MS scale) software sites and services for the public good. Given the continuing importance of online advertising, such efforts are one way to differentiate from the usual ad network suspects.

Oh, and wash, rinse, repeat, for any public records database, which newspapers have a wealth of experience digging into. Speaking of which, Adrian's old boss Rob Curley, pulled off a similar feat when he was running online in Topeka, if I remember correctly.

I haven't visited the Left Coast in a while, and was searching for some good motivation. ETech 2006 seems nerdly and influential enough that it might be worth a peek. Earlier in the fall I was digging around looking around for previous prices on O'Reilly's Emerging Technology Conference, but was having a hard time finding any mention on the web. Now I know why. Sticker shock.

I still may see if I can scrape something together.

D-Lib magazine recently published two articles on social bookmarking services. One was a general overview of the current landscape, the second was an indepth look at Connotea. These are handy citations for lending authenticity to your next peer reviewed paper describing some whizzy new social bookmarking research.

Combine with Golder and Huberman's paper and there's the nascent kernel of a rigorous research community.

Maybe it's because I visit bloglines from a multiplicity of machines, but I seem to be getting repeat items quite often. Highly irritating. I'm just about ready to punt the service. Not to mention that Bloglines is a tad insecure.

Phillip Keller asks whether the tagging revolution is stuck, and while rambling through a number of tagging features, posits some ways tagging systems can be improved.

Generally, I'm not a big fan of tagging as categorization. As implemented, one of the best things del.icio.us does is support social navigation, which Keller notes in passing. However, the tail end of his commentary hits on some juicy territory. Most of the tagging systems are pretty poor at creating other things out of a pile of tagged items. Keller's example is creating simple ordered lists. I'll one up it and say a meta-application of del.icio.us would be the ability to create hierarchical outlines by choosing from sets of tagged items. Such an app would be good for doing things like reading lists, wish lists, and playlists.

Link parkin': the processing discussion board births a discussion that generates a number of good links to information visualization readings.

Link parkin': Fabrice Bellard's QEMU project provides an open source solution to virtual machine emulation on Win32. Interestingly, unlike Microsoft Virtual PC and VMWare, QEMU can emulate other processors, e.g. PowerPC, ARM, on an x86 machine, if I read the status page correctly.

Finally, something worth writing about. Three weblog relevant workshop calls for papers came across the transom.

First off, at WWW 2006, the International Workshop in the Weblogging Ecosystem, makes its third appearance. This time Intelliseek is providing a large pile o' data for people to munge as they see fit.

As an added attraction for WWW 2006, there's now a Workshop on Collaborative Tagging. Out of this CFP comes interesting notes on folks I sort of know. Looks like Soctt Golder has taken time off from (ditched?) The Media Lab and is now working at HP Labs. The tagging paper he wrote with Bernardo Huberman is good stuff and I'm glad to see it get published somewhere. Also, Andrew Tomkins, of Web Fountain fame, has definitely ditched IBM and is now working for Yahoo! Research. Very interesting.

Lastly, Fernanda Viegas and Karrie Karahalios are organizing a CHI 2006 workshop on Social Visualization. Not only will the workshop address visualization of textual social media, e.g. blogs, e-mail, but will be looking at video and audio as well.

Time to get back to work!!

A few folks are slagging the roster of experts behind the American Press Institute's Newspaper Next effort, but there's a couple of energetic and interesting names in there. Skrenta and Wyman are definitely outsiders, while Curley, Yelvington, and Youngman have been up to interesting things from within a traditional newspaper framework.

Personally, I'm starting to find the revolutionary rhetoric a bit shrill. Newspapers and news orgs are too big to up and disappear/transform overnight. It'll be a slow morph ala the US auto industry, a.k.a. not world dominant but still a major factor in a new form.

Link parkin': I didn't realize that the Nature Publishing Group provided Connotea under the GPL. Connotea is a Web based, scientific article, bookmark manager with tagging include.

Speaking of boyd, if you're interested in sociable media and software spaces, you'll want to grovel over her writings, especially if you're an ubergeek and think it's all about the software.

I've talked about Jeffrey Heer's, prefuse toolkit. Written in Java, prefuse supports various interesting information visualization techniques and is particularly good at animation. For example, Heer and danah boyd conspired to implement visualization and navigation of Friendster.

Heer, Card, and Landay have a paper in CHI 2005 that describes the toolkit from a research perspective. Meanwhile, Heer's MS report digs into the internals of prefuse a bit more. Interestingly the toolkit is actually fairly simple in design concepts and should be easily portable to other GUI toolkits. For example, building a prefuse like toolkit in Python+wxWindows should be straightforward, given the paper and the prefuse source code. This would be worthwhile to get a hold of some benefits of Python relative to Java.

a new Lithium-Ion battery for my laptop, which can hold it's charge for more than 45 - 60 minutes.

Link parkin': A good overview of the new Yahoo! Maps APIs for busy developers, by Simon Willison.

Are you a weblog researcher looking for clean data to analyze? Ask and ye shall receive.

Steven Cohen is the Library List editor for PubSub which currently has a total of four Community Lists. A community list is essentially a blogroll, but augmented with some analysis provided by PubSub's LinkRanks tool. The analysis gives you a sense of the activity of the feeds in a list of sources. Also, I'm assuming the lists are easily turned into PubSub queries which generate an aggregate headline feed. With PubSub though, you could also get a meta-feed of posts that point to posts captured by the blogroll.

Some observations and thoughts. Since the analyses are dynamic, it would be powerful if they could be delivered as RSS. Similar mechanisms should be baked into webfeed aggregators. You could cobble something like this out of PubSub queries if there was a decent interface for managing and editing them.

I was about to blame this on PubSub's use of XMPP and lack of a Web services API, but there is no such lack. Apparently, I missed out on the PubSub REST API announcement. Conveniently this API deals with subscription management into PubSub and HTTP based distribution of announcements from PubSub. Wrapping all of this in a nice Web interface is just a small matter of programming.

Via Tara Calishain's Research Buzz

Just a couple of half baked thoughts, inspired by the discovery of io and recent social network hacking. First, implementing a library for graph representation and large scale, complex network experimentation, think networkx for Python, would make a good benchmark application for learning a new language. Such a library engages concepts that should be easy to understand, is easy to map on common datatypes, enables a bunch of fun algorithms and can exercise interactive facilities like GUI toolkits. Add in the large scale aspect, e.g. implementing a set of important social network analysis algorithms/processes on 100,000 node graphs, and you'd get a decent non-numeric performance benchmark.

Congruent to that, I wonder how interesting a whole intro curriculum that exclusively used complex networks for motivating examples could be? Again, I think the science of complex networks could be made quite accesible, yet programming simulations of such networks or derivative applications could easily cover all of a CS1 curriculum.

It'd be a hoot for the students to keep progressing in a CS degree and saying, "Hey, didn't we talk about that in Intro. Programming!" And who doesn't want to know why Kevin Bacon is (not) the center of the acting universe.

Linux's user extensible file system architecture is the gift that keeps on giving. Manish Rai Jain has hacked up a Linux filesystem interface to the Flickr API. Now you can grovel over the contents of Flickr using the same old scripting techniques you have for plain old text files.

Other than answering the question "why?", Steve Dekorte's io programming language looks like a lot of fun. A radically simple mixture of Smalltalk, Self, Actor, and Lisp concepts, the cute bits are that the implementation plays really well with C, meaning io is easy to embed in other programs and easy to develop bindings to C libraries for. Out of the box looks like you can get a lot done. OOPSLA paper and slides for more details.

Could be useful as a pedagogical tool. A relatively pure OOP language would be interesting to explore in a PLDI class.

Kevin Burton's latest baby joins Flock as a "industry term who's name will not be spoken" application that decloaks. TailRank is now in public beta, meaning you don't need a personal invitation to start using his new Web based aggregation tool.

Let's hope Burton's release goes a little smoother than Flock's did.

After looking at the pile of "marked as new" articles I have in Bloglines, I realized most aggregators are pretty poor at helping me forage, pull the meaty bits out of, the information they collect. Adding tagging is a start, but get the dang things out of my, at least that's the problem with Bloglines. Don't get me started on the laboriousness of "clips" in Bloglines.

NetNewsWire is a bit nicer in that it'll pile marked articles in a special folder, but it's still only one pile. And while I know I can drive NetNewsWire through AppleScript to act on items in that pile, action in aggregators is still a second class citizen.

One aspect of NSF that gets underpromoted is the funding of small workshops where experts get together and chart new territories. Noodling around the University of Maryland HCIL site, I ran across the results of the Workshop on Creativity Support Tools.

Led by Ben Schneiderman and Gerhard Fischer, the workshop roster is pretty impressive, including folks like Richard Florida, Peter Freeman, Brad Meyers, Randy Pausch, John Maeda, and Mary Czerwinski, just to name a few. Definitely something to sink the teeth into.

Joel Spolsky takes relentless self-promotion to new levels. In a good way. Who else could turn the excitement of summer geek internships into a documentary with a modicum of Hollywood production values.

Actually, this could be worthy of most CS departments in the country buying a copy to give their students a better look at what real SW development looks like. Seems like a no brainer for any course with Project and Management in the title.

Thanks to some travel I'm a little late to the party, but I was tipped off by Topix.net CEO Rich Skrenta (entrepreneur, NU alum) that, Topix has added blogs to its news mix. Topix automatically tags/categorizes content coming out of continuously updated sources like news feeds and provides search against those labeled chunks.

One of the interesting things about Skrenta's post is his analysis of how much crap there is out there in the blogosphere. To summarize, 90% crap approaching 95% useless, e.g. not frequently updated.

Let's see if I'm in the Topix index.

Link parkin': Anjo Ajewierden discusses the details of mapping a community of webloggers. For future empirical weblog investigators, his list of weblogs crawled could serve as a reference set to compare against.

Wonder if you can get the raw data set?

Nicely played Heather Champ. NMH appreciates a good play on words.

And any information on large, scale LAMP implementations is always welcome.