danah boyd's The Biases of Links, seems to be a point of coalescence for a number of issues I occasionally snipe at: network structure in the blogosphere, blog crawling, aggregator design, etc. etc. Someone should really follow up on her informal analysis with a little more rigor to see if the patterns bear out.

Pondering Calacanis' challenge and Finkelstein's musings, I realized there's a certain flaw with Technorati's authority lists. As far as I can tell, they actually don't serve any purpose other than promotion. A close read of the original Google search engine paper indicates that PageRank was developed specifically to support improved relevance based upon traditional IR techniques. PageRank has a specific purpose in life.

If Technorati authority is supposed to help their keyword search, I'm missing the connection. Authority is "just" a promotional vehicle. If you want to reach people and influence minds then that has a lot of value if a lot of folks pay attention. In some sense, spending money to improve the rankings has to be balanced against promoting Technorati's corporate visibility.

Since I'm not as invested in the politics of the blogosphere as others (no blogroll, no comments, no trackback), what's my angle? Well, if aggregators are going to morph, they need good metadata about information ecologies, cleanly exposed. Mary Hodder and friends took a stab at cooking up a new ranking system, which is fraught with all sorts of issues (lots of noisy inputs, some quite hard to get, complex remixing) but thinking about engines that support such ideas is a wide open area of research.

PubSub is doing well over 1 trillion matches a day according to Bob Wyman. Essentially for a couple hundred thousand standing queries against about 14 million sources, watching a big chunk of the blogosphere is eminently doable.

In response to a snarky commenter ("Big deal! Small number of subs, and small number of events."), Wyman jogged me to think about two other things that might make the hype # more interesting: 1) latency between post discovery and completion of matching against all subs, 2) what the heck are people using PubSub for? Other than vanity searches.

As someone who's hacked up a custom editing interface through older editions of MovableType, Kevin Shay's BigPAPI plug-in is a welcome edition. Writing the backend of a user facing extension is relatively painless, but integrating into the standard MT interface was a relative pain. BigPAPI, as described in Anil Dash's overview, is "server-side Greasemonkey for Movable Type". SixApart's pro developer's network, also has a good BigPAPI tutorial, penned by Shay.

Jesse James Garrett conducts a nice, wide ranging interview about Flickr with Eric Costello, lead client UI developer for the photo sharing site. Just lots of good background material on where Flickr came from and how it operates. For example, some of Flickr's inspiration came from Neopets. I wrote about Neopets before Flickr even existed but didn't take the media objects as community focus angle.

Thumbs up to Garrett for usefully condensing what I'm sure was quite a bit of material.

From one of the loveliest cities Down Under, Sydney, Andrew Vande Moere puts together infosthetics. Appropriately elegantly designed, both visually and intellectually.

Hat tip to Smart Mobs who was propagating the fine thoughts of one Fernanda Vi

Jason Calacanis' $10,000 bounty for a better Technorati 100 is quite amusing. Not only for the cheekiness of the thrown gauntlet, but the comments which illustrate the many facets of this challenging problem.

As an added bonus (is there any other kind?) I stumbled upon PubSub's linkcounts, another interesting posting flow tool and their site stats feature. Neato stuff, but I bet since their data is feed based, you'd see significant variation from something like Technorati, whose rankings are based on scraping site front pages. That and PubSub seems to conflate blogs with last 2 elements of domain name, which is a bit problematic...

I'm wondering if there isn't a Google moment about to happen somewhere out there. A couple of enterprising computer science grad students pull in an idea from another academic discipline (percolation theory?), engineer the hell out of the implementation, execute cleanly on the user interface, and ride off to fame and fortune.

Mark Glaser has an in-depth interview with the brains behind Topix.net, Rich Skrenta and Chris Tolles. It's mildly entertaining, mainly for the economics of the business, but not a huge amount of tech insight. Still worth a peek to see Paul Graham's startup wisdom in action yet again.

I have to chastise Glaser though. I didn't know there were degrees of perfect.

I'm approaching the 180 webfeeds subscribed to mark and Bloglines is starting to get a bit creaky. Or maybe it's just my innate techno wanderlust. I've already got a Rojo account so maybe it's time to jump ship, although on the times I've tried it, I've found their interface irritating. Jack Vinson also has a few nuggets on Lektora and Blogbridge looks like it might have potential, despite my Java Webstart trepidations.

Dave Gifford and Lyn Turbak are shopping a book on modern programming language design principles. Targeted at CS grad students, the table of contents outlines topices in Design Concepts in Programming Languages that will make most people's heads explode. Including mine!! However, stuff like this is useful for every self-respecting CS graduate student and it's too bad we haven't been able to solidify our PLDI offerings here at NU.

So it's been a while since Flickr got swallowed by Yahoo! They've been mumbling about moving to the Bay Area, new data centers, improving the Web services API, yadda, yadda. I wish they'd get around to adding a new feature or two. Another month or two and they're going to lose their hipster cred.

Oh, wait. Never mind.

Clustering looks like a straightforward application of some well known information retrieval techniques. But an interesting thing about interestingness is that you now have automated egoboo. Let the gaming begin!!

Lucas Gonze is really on point with this thought:

Most of the doodads that people are dreaming up to help web audio and video to take off are beside the point. The challenge is making A/V more webby, not changing the web to accomodate A/V.

Going reductionist, if a human authorable URL scheme for specifying in/out points on continuous media became widely adopted, A/V would become first class on the Web.

Jon Udell touches on the potential for connecting different autonomous tagging systems. We both agree that a Wiki style, inter-system reference language has potential, but Udell adds in the utility of aggregators and meta-search engines.

There's probably enough instantations of these systems, c.f. Connotea, del.icio.us, CiteULike, that a watch/buzz engine specific to them would be useful.

Moment of silence for Gataga... Thank you very much.

When AOL announced that they were going to provide blogging tools to their members, there were apocalyptic pronouncements from the blogosphere both pro and con. My how times have changed.

AOL is beta releasing a new webfeed aggreagator, MyAOL, and receives not much more than cursory nods. While apparently similar to MyYahoo!, if AOL delivers on some of the coming attractions, MyAOL could be a nice product.

Konfabulator is a neat little desktop sandbox for running small JavaScript applications. Originally a MacOS application, it sort of got bigfooted by Apple's Dashboard. Yahoo recently bought Konfabulator, and made it free. Yeah!

I forsee a plethora of Konfabulator widgets exploiting the many Yahoo APIs.

Despite appearances to the contrary, the Northwestern Department of Computer Science no longer exists. Now I'm appointed in the Department of Electrical Engineering and Computer Science.

Living in interesting times.

Link parkin': kbsearch is Randy Charles Morin's meta-search engine of blog watch engines.

As a way of blowing off a little steam, a couple of Flickr employees launched a tag and pool for self-inflicted ugly photos, aptly named DoYourWorst.

Why am I pointing to this? Because while it may be a minority use, I maintain that tags as "social events/spaces" is an important use. In fact if I could figure out an objective means of classifying users tagging behavior as event or classification, the current high profile focus, an interesting study would be to look at the varied social tagging services, compare and contrast their features and see what style they encourage.

Even though Flickr and del.icio.us have very similar tag mechanisms there's something qualitatively different about how they're used. Something that's worth analyzing, quantifying, and understanding for future software designers.

A long time ago, I lauded Movable Type for supporting a variety of database backends. This feature allowed for a wide range of installation scenarios. Apparently this ability got lost in recent vintages of MT for the whizzy dynamic publishing capablities.

Anil Dash concurs with me (;-)) and reports that PostgreSQL and SQLite DBs can be used with dynamic publishing. Some of the GIS, XML, and extensibility features in PostgreSQL could be put to amazing effect for an enterprising MT extension hacker.

Thanks to Greg Linden's careful eye on Google Labs one can start to see some of the value of all those PhD's at Google. First, PLDI (amongst other things) expert Jeff Dean, along with Sanjay Ghemawat, cooks up a domain specific programming model and implementation that allows J. Random Hacker at Google to easily crunch lots of data across many machines. Next UNIX legend Rob Pike and team take the baton and start sloshing around petabytes of data at a time by building a new language on top of Dean's work.

When you have a toolbox with gadgets like this in it, tasks such as blog analysis and aggregation are cast in a much different light. High end toolmakers are major force multipliers in military speak.

According to Heather Champ, Flickr has a million members. Wow!! A good idea, solid execution, and a dash of agility (remember Game Never Ending and Flickr Chat), go a long way. Note that Flickr isn't much more than a year old.

Mary Hodder has a nice overview of how various blog tracking services work. Be sure to check out the feature comparison chart, but the key point to take away is that services like Bloglines, Blogpulse, Technorati, PubSub, Feedster et. al. are barely in the same business as Google, Yahoo!, and MSN. Sure various of them take stabs at keyword searching of posts, but their bread and butter is in trend tracking; mainly via link watching but with the potential for more sophisticated analyses down the road.

"MochiKit makes JavaScript suck less." Gotta love an opening line like that, along with testing, documentation, cheap Python connectivity through JSON, and an open source license.

Via Ryan Shaw's del.icio.us linkblog, which is pretty damn entertaining.

I'm noticing an upward trend in daily linkblog postings (probably aided and abetted by social bookmark services with decent APIs) in the feeds I read, and I'm really enjoying it. There's a high level of serendipity in an easy to read form factor.

On a related note, while aggregators bundle things together nicely, they also strip out the overwhelmingly horrific visual design plaguing websites today. Honestly, Angelfire has taken over the web.

As an experiment, take your top 5 favorite weblogs. Follow permalinks to a few archive posts and marvel at the amount of screen real estate occupied by non-content. As a mildly objective measure, navigate to the bottom of the archive page using the spacebar counting the number of chunks that contain absolutely nothing of the archived post.

End of rant, and yes I know where my petard and your hoist are.

Your humble author now explains the break in posting after 54 consecutive days of commentary.

Got an airfare/lodging deal for four evenings on the isle of Grand Cayman at a hotel where you can walk out of the lobby onto a nice stretch of the 7 Mile Beach. Lots of laying on the beach reading (knocked out Moneyball, 1/3 of Quicksilver), sampling Pina Coladas, nighttime surf strolling, and visiting Stingray City. Carribean islands in the low season are nice because the locals aren't so harried and have plenty of time to chat. Also, the Cayman Islands are still crawling out from under Hurricane Ivan so they're really happy to see your dollar. Highly recommended!!

We now return you to your regularly scheduled techno media hackery.

The Hype Machine is a watch engine that tracks MP3 file references in blog posts and presents them an easy to consume fashion. The system incorporates two spiffy technologies, NYU's experimental Content Distribution Network, Coral, and the open source XSPF Music Player.

Last month, I postulated on "podbytes", sound bites for podcasts. Jon Udell has been pushing the envelope and has a spit and baling wire, proof of concept for audio linkblogging. There are so many moving parts, a.k.a web services, involved it's looking a little Rube Goldbergian, but it seems to work. Now if someone can smooth out the bumps (I'd say Odeo but they've got enough to do), that could be a lucrative service.

At Foundcity, del.icio.us tagging meets Google Maps. Mayhem ensues. Kewl.

Vin Crosbie recently fired off a deep, well considered e-mail to the Online News mailing list. The message was regarding what to learn from what Lawrence Journal-World has been doing to receive so many industry kudos. I was going to ask for permission to liberally quote from it, but Vin saved me the trouble and posted the e-mail to the Digital Deliverance blog.

The 5 second recap: develop long term visions and strategies at the top, unleash good energetic people on the ground to make it happen

Crosbie also notes six trends that news organizations really need to capture in their strategic thinking: multimedia, unlimited depth, on-demand, depackaging, individualization, mobility and ubiquity. These are consistent with his running theme that infrequent, mass delivery of uniform content bundles is not a winning proposition in the long run.

Link parkin': CommonTimes has an api. Jeff Reifman calls for mashups.

Don't look now, but beep, currently holds the #7 ranking for the word "beep" on Google.

On another note, Your Mom, a project researched, conceived, pitched, launched, edited, and now being reengineered for citizen's journalism, by Medill students, made the front page of The Washington Post.

At Medill students get real opportunities to turn the academic experience into industry impact.

Strictly for PLDI wonks. Patrick Logan gives a capsule summary of Termite, a new Scheme based language for distributed computation.

Scheme macros + Erlang's concurrency + distributed continuations. Tasty!

Greg Linden spotlights Findory's API for searching their database of news articles and blog postings. REST in, RSS out, and you can even request personalized results.

I'm not into MySpace. Too old, not into pop music, etc. etc. But in a major development, Fox paid big bucks for MySpace's parent company. PaidContent.org has a great wrapup of the deal's details and implications. Those people do not sleep.

On a contra note, Danah Boyd connects the dots on the dissent commodification at work.

NeoPets, Flickr, Bloglines, del.icio.us, MySpace: if not a trend, I smell a trend story. The money's back jack.

Jython is an implementation of the Python programming language on top of the Java Virtual Machine. Since Jython runs on Java you could do things like stuff Python into applets or server side servlets. Think of it as ActionScript but more programmer oriented and without the all inclusive development environment of Macromedia Flash. Of course, applets haven't taken over the world but for small installations Jython based applets are eminently useful.

Unfortunately, Jython's language features had diverged from the ur-Python. Recently there's been a big effort to catch up, and Bill de h

Since it was just down the highway for him, J. D. Lasica attended IBM's New Paradigms for Using Computing conference. This was a small, invite only confab led by Daniel Russell, with whom Lasica snagged a five minute video interview. The main focus was on ultra-portable devices and mobile computing applications.

One thing I noted was the dismissal of the laptop as a mobile computing device. I'll grant that laptops are mainly nomadic, but they are fast approaching the signature consumer computing device. Desktops are so last century, and cell phones are for talking unless you're a bleeding edge geek or business person. And the combination of laptop portability with large (enough) screen real estate makes them decent, if not great, sociable devices.

More and more I'm seeing a laptop, (or two!), being a central element in student study group huddles. You know the scene in which 3-4 students commandeer a large table in a cafe or the library and hunker down for a few hours, books and papers spread everywhere.

To make a long story short, there's probably a lot of mileage in researching and engineering laptop technologies for the forseeable future. In my book, they're essentially the briefcase of the new millenium.

Even before finishing up the 32 Hot Features list, SixApart has unleashed the MovableType 3.2 Beta.

This probably isn't noteworthy to readers of this blog other than the hordes ;-/ of folks demanding comments. If I bite the bullet and rework NMH for MT 3.2, I'll give comments a shot. Warning though, you're dealing with a crusty old USENET type who used to read newsgroups (moment of silence) in Emacs.

Jack Vinson, follows up on my aggregator though exercise. He captures an issue for those of us who use aggregators for situational awareness (my definition): redundancy reduction.

For example, my Findory personalized feed picks up on items from feeds that I'm already subscribed to in Bloglines. Except that I'm already subscribed to them, so why do I need Findory to tell me about them? Having Findory keep track of that is too big a job, but my aggregator should be up to the task.

Memo to Jack. I'll buy next time at Mud.

Jeff Reifman, along with a small crew of volunteers and under-paid interns, recently launched CommonTimes.org. Capsule summary? CiteULike focused on news. Needs more mass but looks like it could be worth keeping an eye on. One neat little bit is the source stats breakdown in the left rail.

Interestingly, other than overall launch tone, I don't see anything that enforces so called "news" posting. Not that I'm advocating any such mechanisms, but an interesting study would be to look at a couple of tagging systems and see how much "topic drift" or abuse they receive. Define topic drift amongst yourselves.

The tagging concept is pretty much old hat. The big win so far has been in enabling social navigation. Next up, who can start to build really interesting analysis services on top of tagged content.

Watching David Heinemeier Hansson's Ruby on Rails Movie, I was struck by a number of things as the tutorial proceeded, herewith braindumped for posterity.

Building a blogging application has become the "Hello, World!" of web application and content management systems. This includes all of the pejorative connotations. Next up, del.icio.us knockoffs as "Your Second Application!".

Dennis' Corollary to Greenspun's Tenth Rule of Programming, applies. However, the Rails community has gone the extra mile on design details and marketing, which is not to be scoffed at. And the emphasis on unit testing mitigates the "bug ridden" factor.

Screencasting would be a valuable tool to teach computer science students. Having project oriented classes is great in theory, but a downside is working through the iterations to train students up on presentation skills. Presuming either really accessible tools or reasonable training, screencasts could be a great way to time and place shift demos for busy faculty and TAs. They also are a concrete artifact that students could put in a "portfolio" to convince employers of presentation skill along with design proficiency.

Heck, if I were an employer, I might start asking for screencasts as a highpass filter.

Somehow I managed to get back to the #1 Google rank for media hack. We'll see how long that lasts.