If you’ve been reading this site for a while you may recall my earlier post on MusicPlasma, a cool site that uses Amazon’s web services to build a visual search engine for music based on collaborative filtering data. It shows bands as “orbs” or planets, each with their own “solar system” of related bands. (Play with it, it’s pretty cool.)
Recently a colleague contacted me and asked if I had anything interesting to say about blogs and how they might shape the media world in the next year or so. My initial thought was “Why of course I do!” – but the fact is, it’s not easy to have something interesting to say about blogs that doesn’t require a hell of a lot of throat clearing, groundwork laying, and general hand waving. Try to explain to an intelligent layperson the power of blogs – it’s not easy. The perfect piece has yet to be written on the true power and impact of blogs; at least, I haven’t seen it.
Sure, the examples are there – from the tsunami coverage to Trent Lott. But my colleague was looking for a visual high order bit – a way to see what the big deal was, after all. I thought about Dave Sifry’s slides from Web 2.0, but that was still too inside-the-blogway.
The I thought of MusicPlasma. The thing I like about it is how intuitive it is – put in the name of a band you like, and you find more that you might like but had never heard of.
Hey, I thought, what if we did that with blogs, and instead of Amazon data, we used Technorati cosmos data, or Feedster data, or Findory, or Bloglines, or some combination of all of that plus more? “Folks who read this blog also read that one,” for example. Or “Blogs who link to this blog also link to that one.” If we put a sophisticated interface with some dials and levers, it could really be a neat tool for exploring relationships in the blogosphere. I could imagine some cool slices that might parse this wildly growing ecosystem in interesting ways. (I’ve always been fascinated by the visualization of data, I was the force behind the Standard’s metrics section, if any of you recall that.)
So I think I’m going to try to do it. But the honest truth is, I have no idea how to. I’ve contact the folks behind the various sites listed above, and they all stand ready to help. I just need a technical lead, and ideally, to talk with the MusicPlasma guys, to see if we might share their skin, so to speak. Anyone know them?
What do you all think? Would this tool be a valuable addition to the conversation?
Great idea!
You should be able to accomplish this with a tool like Grokker. If you talk to Richard Ault of Technorati it would not surprise me if he already knows of some reporting tools similar to what you have described in the works.
Neat idea! Its time we went beyond case-studies of what a blog can achieve to step back and look at a bigger picture of its potential and prospects.
Brilliant idea! It would make “finding the others” so much easier…
Can’t wait…great idea!
And, you’re right, it’s hard to explain blogs, I’ve tried it twice this week and I believe they are still clueless 🙂
Unfortunately, MusicPlasma seems to have vanished. However, for a similiar type of function for blogs, check out TouchGraph’s implementation of a LiveJournal visualizer. http://www.touchgraph.com/TG_LJ_Browser.html
If anyone knows where MusicPlasma went, or how to get in contact with the owners, please let me know.
similar to musicplasma, is musicmap http://www.music-map.com, part of the gnod network.
I vaguely recall seeing a system that did just this a few years ago, but unfortunately I can’t dig out any details.
Using viz tools to help people better understand their external business environment/bands/romantic partners/blog popularity is a fascinating idea.
ToughGraph is really rough as a tool and Musicplasma looks nice but it’s slow and simplistic. Nothing new past what ThinkMap did. Boy did they waste an opportunity, their visual thesaurus is great, too bad the tools lacked a certain amount of polish. BrainForest comes to mind as well. I just got done emailing Jeff Heer, who has developed Prefuse. Been playing with that and building the online dating industry ecosystem- entities like people, companies, money, many types of relationships, events (fired, hired, funded), all tied to live data feeds like blogs and press releases. Jung looks like a good tool to build the graph and let Prefuse render it. Java is much faster than Flash for this sort of thing.
Been doing this as a visual aid to help with my consulting gigs when I come in and do the big brain dump. It’s all just arcs, nodes and edges, the rendering mechanism will continue to improve, it’s the back end thats important, and classifying all the different types of relationships. Have you looked at the FOAF spec and all it’s permutations? Yikes.
I worked on a tool called Contexta a few years ago that “was” going to do all of what you mention, which is the tip of the iceberg. Data-driven viz based on user role and context of use. Free tastes, pay for deeper meaning when you start parsing blog content and re-tagging it’s RDF according to your toolbox of patented algorithm. throw in del.icio.us and flikr and you’ve got something.
Great idea. Will every blog in the world point back to Radiohead like Music Plasma does?
For a jumpstart on processing blog relations, check out the blogging ecosystem:
http://www.myelin.co.nz/ecosystem/dataset.php
Most systems I’ve seen for graphically visualizing large interconnected datasets suffer from being very slow. I think that to be successful, it will need to do some sort of intelligent caching to cut down on the amount of processing.
This is actually an easy application from the back-end side, Technorati has API calls to do exactly this – the “cosmos” call and the “outbound” call to help do this kind of easy discovery and link visualization. All you need is someone to plug in the visualization code into our API calls and it is done.
If anyone is interested in doing this, drop me a line, we’ll up your API calls to handle the traffic spikes it would involve…
Dave
You need to have the debate over what GUI technology to base the tool on. I would say Flash beats Java beats SVG in terms of numbers of existing enabled clients out there but in terms of ease of development Java would win.
There’s a bunch of open-source Java graph layout toolkits listed here: http://www.manageability.org/blog/stuff/open-source-graph-network-visualization-in-java/view
I’ve tried Touchgraph and Prefuse in SNA apps and found both to be OK but they do occasionally lock up. I’ve also found that their layouts can get very cluttered too. Treebolic looks nice but doesn’t show strength of relationships between nodes too well.
TomSawyer software do nothing but sophisticated layout tools but their products are commercial offerings.
I actually found Music Plasma mediocre. The sizes of artists felt wrong, and the connections didn’t fit. It actually felt too subjective to me.
But for something more topical and categorizable like blogs, such a system sounds cool. But what about using something like Clusty or, more graphically, Mooter? This seems closer to the current text-centered tools, and thus possibly more easily adapted.
What a truly great idea!
The data from Share your OPML at http://feeds.scripting.com/ could also be useful in building this. There’s an SDK to access the data: http://feeds.scripting.com/sdk
I’ve found that the best way to explain blogging to people is to demonstrate it – set them up with a blog and an aggregator and demonstrate commenting and trackbacks and linking. It seems that only when people see trackback in action do they suddenly realise that the power of blogging is in the network you create rather than the posts you publish. I’ve done this both with the Big Blog Company’s boot camps for journalists, and on a one-to-one basis and it works very well. Of course, a nice way to visualise the network would help, but it remains an abstraction that some people still won’t ‘get’.
the blinx search engine/desktop search tool has taken a whack at this, although it’s not exactly what you are describing. enter your blog and click the “viz” button – or try http://searchus.blinkx.com/BlinkxBroadband/vis/visualize.jsp?q=searchblog&lang=&sf=
Anybody remember the Social Network Explorer that used to be part of Blogdex in the old days?
Musicplasma could be more useful by showing meta information about each artist and link on a mouse hover… for example a summary about the genre and background for each artist, and more details about why a particular artist is linked to another. How many people bought both artists together… etc.
The same would apply to a visual map for blog relationships.
In my former life, I used Tom Sawyer Software’s Visualization http://www.tomsawyer.com/home/products.php and Inxight’s VizServer, http://www.inxight.com/products/vizserver/, to prototype visualizations of large social networks.
Both work fairly well.
Just was over on Technorati and see one of the runners up to their developer’s contest made just such a tool. Check out Michael Dale’s Touchgraph which is a downloadable today but Michael says will be web-enabled soon:
http://hybrid.ucsc.edu/~dale/technorati/
You’re essentially just talking about graph visualization.
It’s certainly very interesting but the issue is that unless the UI is done VERY well then its not much use.
Its just too much data…. its pretty but just too much data.
http://www.peerfear.org/rss/permalink/2003/02/19/1045648437-Reputation_in_NewsMonster_and_Blogs.shtml
and
http://www.peerfear.org/rss/permalink/2003/02/19/1045684244-Blogstreets_Blog_Graph.shtml
You may also want to check out what the people at tribe.net are doing.
Have you seen the visual search engine
http://www.kartoo.com
It is a “metasearch engine with visual display interfaces. When you click on OK, KartOO launches the query to a set of search engines, gathers the results, compiles them and represents them in a series of interactive maps through a proprietary algorithm”. Requires flash
This is pretty much the basis of the
Keep in mind other ways of mapping in addition to the standard lines and connectors.
=============================================
Maybe something with -tabs- along the top for showing subsets of the “map”, with one tab for the overall view. Keeps it from being to busy and overwhelming.
=============================================
Perhaps use standard mapping color theories and practices. It’s been fine-tuned over generations…
http://www.xspace.net/
This was done by apple many years ago, and it never caught on.
There’s an SDK to access the data: http://feeds.scripting.com/sdk
can’t open this url. Do u have a mirror one?