Douglas Thain has concoted the Fault Tolerant Shell, an experiment in merging scripting (high level, process oriented programming) with distributed computing. Apparently th e Fault Tolerant Shell is the residue of a new philosophy regarding grid computing, known as the Ethernet approach.

I wonder how the Fault Tolerant Shell compares with scsh as a scripting language.

Link parkin': The Project for Excellence in Journalism has released a voluminous report on "The State of the News Media". Vultures flocking even as I speak.

Link parkin': Pre-release of the first chapter of "Making the News", Dan Gillmor's upcoming book on the media in a world when everyone can be a journalist.

Link parkin': Harvard's Kennedy School of Government recently released a case study of how weblogs were influential in the Trent Lott affair.

I'm kicking the tires on Brad Rhine's Frequency desktop blog posting application. Currently, I'm liking it better than w.bloggar which, as far as I can tell, is the cream of the current crop.

Minor nit, Frequency needs a refresh button on the edit posts list.

Update There's a menu selection, Posts : Refresh List, and key combo, Ctrl-R, to take care of the above nit. Thanks to Brad Rhine

Ask and ye shall receive. E-Media Tidbits now has an RSS feed.

Many thanks to Adrian Holovaty, ace Web hacker and all around good guy. Maybe he can help out the Online Journalism Review folks next.

Sparklines is a concept of visual display Edward Tufte is cooking up for his next book. A sparkline is an, "intense, simple, word-sized" graphic. Damn cool.

Seems like a directory, or categorization, of weblogs could stand to use a device like sparklines as a brief quantitative summary of the regularity of posting. Sort of like the calendar except denser and more informative.

Mark Newman, of the University of Michigan, was in town to give a seminar. Newman is one of the new breed of physicists/applied mathematicians who are heavy into graph theory and social networking. Newman worked closely with Duncan Watts on characterizing various properties of large simulated and empirical networks.

For the most part, the talk was an overview to anyone who's been following the area. Towards the end though, he got into recent developments in finding communities within networks. Social network analysts, and clustering afficianados, have tools that work on relatively small graphs. But in offline discussion, Mark seems to think that his work and extensions by others could scale to sparse networks of millions of nodes.

That's using one desktop machine. With a lot of RAM. But still.

For perspective, Technorati is "only" watching somewhat over 1.8 million weblogs. blo.gs claims to be monitoring about 1.4 million. Throw a small server farm at the task and you could easily apply Newman's algorithms on a reasonable time scale. The algorithms aren't particularly difficult either. You can grab the papers from his site. If you do, there's a slightly tricky hidden point, I think, involving the determination of biconnected components, but it's actually a minor issue.

Instead of these uninformative power law graphs maybe we can start to get a better picture of what the blogosphere looks like.

A nice Python templating system like Divmod.Org Nevow would be nice, but the inhale on the package is just way to high. I've got to pull in a whole event driven Internet protocol/server/client framework just to do templates?

GNU Screen is a terminal multiplexing program that I should be using on a daily basis. The neato trick is that you can detach from a screen and reattch at a later date or even from another machine. Very useful for long running processes.

So it figures that I take two months off, and somehow my Google juice goes up. I've gotten more external contact about my blog in the past month than I did in the previous year of blogging. Not that that was a particularly high bar.

Oh yeah, now I have comment spam too. The sheer stupidity of which boggles the mind. I don't even take and display comments on this site. So it's not like the comment spammers get any Google-boost out of it.

Sheesh!

MyStack is a new hosted linklist service being provided by PubSub.com. Sort of like Blogrolling.com except for link lists.

A new twist though is that a stack can be a canned keyword search that generates links. This search runs against PubSub's search engine of blogs, newsgroups, and news feeds. This lightweight hosted Web service idea is catching on.

Here's a thought. When will indivdual weblogs morph into platforms for integrating services? Maybe that old "personal portal" idea wasn't so stupid after all. The key would be that if they were closer to writing and syndication tools, people would be come more attached and thereby more interested in pulling in services. The hard part is keeping the programming barrier to entry pretty low though.

Vinay Sajip has put together a configurable logging module for Python. Probably something I should check out given all of the Python code I'm writing these days.

Originally uploaded by riddle.

Flickr a chat, community, and networking tool centered around images, allows you to easily post to your own weblog regarding an image on Flickr. Great idea, making connections to tools people already have easy. Well, we'll see

One interesting thing about Flickr, is that users have a limited amount of image storage: only 50 photos total. Not horrible, but not much in an age where images are really cheap. Then again, even though storage is cheap, images can probably chew it up pretty quick.

And the actual formatting leaves something to be desired, but hey, the price is right.

Vin Crosbie finally cuts loose with his previously promised take on a succesful strategy for online news. Along the way, he takes the current news industry, online and offline, to task for poor judgement, lack of vision, and general all around myopia.

The key to his strategy? Focusing on the service that news organizations provide as the real value add and using technology to make hyper-relevant personal editions.

RSS/aggregator zealots will now get all worked up about how this specific technology is what Vin is talking about. Take a deep breath folks. RSS has a long way to go before it can meet the goals he discussed. But it's a good start.

And somebody please tell OJR to get on the bandwagon and provide an RSS feed

Soumen Chakrabarti's book, Mining the Web: Analysis of Hypertext and Semi Structured Data should be required reading for anyone who claims to be doing any sort of large scale Web page/site analysis. Especially all those folks trying to do bad PageRank knockoffs.

Mainly for my future reference, I need to stash a link to the Weblog Ecosystem Workshop at the WWW 2004 conference. It'll be interesting to see how rigorous this first edition is. Besides the folks at IBM Almaden and the search engine companies, who's taking weblogs seriously as an academic pursuit from the technical side?

Must ... make ... submission ... deadline

Flickr came to life during my hiatus. It's a social environment for sharing photos. I think it would be cool if you could easily embed links in images and create photo meshes. Think of the network analysis possiblities in that!

One two, one two!

Is this thing on!

Steve Makofsky, aka The Furrygoat, generated an interesting visualization of RSS feed content. I think he undersells the potential of this technique a bit. A bit stale, but mainly interesting for the link to a .Net TreeMap control.

For those scoring at home, yes I'm alive, but recurring bouts with flu/cold are severely slowing me down.

Slightly stale, but an America Online press release indicates the media giant is pushing forward in integrating the activities of its membership with its news creation and distribution mechanisms. Of particular interest is connecting social activity with news search and interactive features.

John J. Lee has put together python-spidermonkey, a way to bridge the Mozilla SpiderMonkey JavaScript (written in C) into a Python interpreter.

I actually saw this a few days ago in the Daily Python URL, but didn't really grasp the significance. Simon Willison pointed out how this might be useful as an embedded interpreter within Python, without having to write a complete language implementation. This could be useful for Python apps that want to expose a scripting mechanism, but can't quite sandbox Python to satisfaction. Sounds wacky, embedding a scripting engine in a scripting engine, but SpiderMonkey is a bit better behaved for restricted execution. Heck, it's burned in through thousands (millions?) of hours of use in Mozilla.

Bryar is a Perl based blogging tool cooked up by Simon Cozens. It seems to occupy an intermediate position between Blosxom on the one hand and Movable Type on the other, in the simplicity/flexibility/completeness spectrum. Probably worth kicking the tires on.

Mostly link parkin: Surf*Mind*Musings on some Php image galleries.

Need to find RSS feeds from a Python script? Put Mark Pilgrim's RSS locating module in your crawler and smoke 'em.

Digging around for some papers on network growth models, I ran across a preprint of a paper by Michael Mitzenmacher scheduled for the new journal Internet Mathematics. Fan Chung (Graham?) is the managing editor of a seriously kick ass editorial board.

By the way, if you really want to understand the literature (across a number of fields) regarding growth models that generate power law distributions, you'll want to read Mitzenmacher's brief history of the area. It's not too hard to find using Google.

And of course it figures that the essence of the model was presented in 1955 by (Herb?) Simon. There really isn't much new in computer science.

Correction: I know think it's just Fan Chung and the journal web page has a typo.

Mystery solved. Chauvinist that I am, I didn't think that it might have been a woman with a professional and a married name. Mae culpa. I'll just leave it at Chung though.

Okay, I think I've just about crafted a halfway decent environment under Windows XP. As opposed to MacOS X, I had to pull a lot of stuff from the 'Net. Basically I've papered over XP to make it feel more UNIXy. Here's what I'm currently using to replace a combo of NetNewsWire, Omni Outliner, and Safari, plus a baked in *BSD substrate:

• Cygwin (UNIX replacement) Back to the Future
• XEmacs (Text editor) Baked into my genes
• FeedDemon (RSS aggregator) Enough keyboard customization to make it livable
• Mozilla Firebird (Web browser) Must have tabbed browsing!!
• Mozilla Thunderbird (e-mail) Need a change from Outlook/Entourage
• ActionOutliner (outliner) Nice, simple keyboard controls, and lets you put URLs in the outline

Of course it's a heck of a lot more work downloading and installing all of this stuff, then buying Jaguar and installing 2 more apps.

So FeedDemon isn't quite the savior of RSS aggregation for me under Windows XP, but I've configured it well enough that the actual task of reading doesn't drive me up a wall. Now if I can get some semblance of the hierarchical outline that NetNewsWire has I'll be happy.

The key thing about NetNewsWire's outline and how it presents folders is that it provides so much context on progress made. I can quickly see how many unviewed items in aggregate are left, as well as in particular groupings.

You really don't appreciate some things until they're gone.

Link parkin': Mark Bernstein simulates the effects of advertising in the blogosphere. Massive clustering of readership occurs. Sez any decent programmer should be able to recreate and validate the experiment.

Hmmmm.....

With a spate of Web based, RSS aggregation services coming online, it's most appropriate that William Grosso deliver an analysis of the pros and cons of desktop versus server aggregators.

Synopsis? The desktop will be the home of high cost per user services, e.g. tight UI interaction. Server side will rule where aggregating users is important, e.g. metasyndication services.

I'll be interested to see which side wins the bleeding edge extension race. With a decent plug-in architecture, a popular desktop platform would have many more developers. Then again, a server side implementation that presented a really good Web services API (ala Amazon and Google) could be a killer.

Thinking outloud? What if there was a standardized way to represent the reading state of a blogroll? Then you could sync a desktop and a server side aggregator to get the best of both worlds.

Work and a flakey TiBook have slowed down posting here. The laptop inconsistency has also forced me to work on a Windows Tablet PC for a bit. In trying to recreate my MacOS X environment I've been kicking the tires on Win32 RSS aggregators.

Jeez, they suck.

Alright, maybe I'm being a little harsh, but relative to NetNewsWire, they drive me up a wall.

It's actually a couple of minor details that kill. First, many, not all but many, Win32 RSS aggregators don't do folders, or do them nearly as elegantly as NetNewsWire. Second, spacebar reading, if it even exists often doesn't work right. For example, in Outlook based aggregators such as NewsGator or IntraVnews, if you read all of the posts for a given feed, the spacebar doesn't advance you to the next unread post. You have to navigate out to the group level, pick a group, and then start reading again. Usually without the support of any keyboard shortcuts.

And lastly, the arrow navigation in NetNewsWire is to die for. With one hand I can maneuver through feeds, groups of feeds, posts, and then to where posts link to.

The beauty is that NetNewsWire's structuring gives me better perspective on the reading process, especially for 100 or so busy feeds. I can't imagine dealing with a 300 or so post pile of mush when I actually get a chance to visit my aggregator.

Previously, I really didn't see what all the fuss Dave Winer and Scott Rosenberg were making about all-in-one aggregation views. After having suffered through a host of bad navigation options, I can see their point.

FeedDemon may be my only hope.

But NetNewsWire is still cuter than a bunny's bottom.

From the Dead Parrot Society, Ryan Pitts writes of how one newsroom plans to incorporate weblogs into its coverage of the Saddam Hussein capture. To summarize, excerpts from foreign blogs will be used as react material.

Random thought. Weblogs, to the extent that they submit to standards of verification, are an open source of reporting. There's no reason that every other paper in the country couldn't be mining local, national, and international weblogs to get personal stories for their own stories.

Don't quite know where I'm going with that, but something to think about.

John Unsworth penned a Not-so-Modest Proposal, discussing the future of scholarly communication. The major theme is that humanities organizations should start thinking of collecting digital articles as a default (Caveat from a colleague Michell Citron, digital doesn't preserve as well as analog).

Unsworth was the founder of UVA's Institute for Advanced Technology in the Humanities, so he's got a few chops. He's also a very nice guy, as I met him at a luncheon when he came to NU to consult with the dean of the College of Arts and Sciences about technology in our humanities programs.

Just for the record, Chicagoland has a great opportunity for a moon shot in this direction with Northwestern, University of Chicago, The Field Museum, The Chicago Historical Society, The Newberry Library and The Art Institute of Chicago within a tiny geographic area.

Oh, did I mention that they're all within easy reach of a new Chicago area optical network? What's a few gigabytes between friends?

The guys at the Abusable Technologies Awareness Center might be one of the best collections of security experts ever.

Everybody going's gaga over David McNab's Linux user level file system implemented in Python, but fuse looks just as good.

And besides, it's not quite as sweet a hack as writing Linux kernel modules in Scheme.

Giles Turnbull tells you how to make my favorite outliner (OmniOutliner) stand on its head and do all sorts of fun export tricks.

I'm sure someone else has thought of the term already, but in the spirit of Waypath's meta-weblog services, I thought of a generalization: metasyndication services. Eventually there'll be quite a lot more information flowing through RSS, the majority of which might not come from weblogs. But there will be meta services that aggregate, analyze, and distribute that content and even operate on larger scale, complex dynamics of those feeds.

Heck there's only a few references on Google to the term "metasyndication" and nothing for "metasyndication services".

I just zipped past post 500 and didn't even notice it.

Must be getting the hang of this thing.

Chris Wright, publishing in gnovis, documents how blogs gave the Trent Lott story enough initial legs for traditional media to finally latch on.

Somehow I bet this'll be cited in quite a few grant proposals. And is it just me, or are there a bazillion peer reviewed online journals dealing with communication, society, and technology? I didn't know there were enough academics in the states to do all that reviewing.

Actually, it's not that type of networks.

Intel is BSD licensing a library for doing machine learning based on graph models, e.g. Bayesian networks.

Thanks to Ted Leung