26 Oct 2003 @ 15:14, by Roger Eaton
This article is an overview of the blogosphere by a newcomer bringing a fresh eye to the scene -- the result of my researches how best to implement the voice of humanity vohP2P middleware program in a blog-friendly way. I am taking ming's comments about making blog metadata available through an interface seriously. The technoblogerati will want more than a metadata interface, though, before they are seriously interested, so it was delightful to stumble on an idea that really might get the voh project moving. I call it the MetaWeb; it is a simple, highly extensible idea for building a peer-to-peer web on top of the web, implementing among other things, the instant blogroll by category and the stigmergic ant trail in a big way - see too this article.
A stroll through the faqs for those who are new to the weblog party is a revelation. (The first thing to know is to say blog, not beelog, and the second thing is that the party might just be getting started.) From 1999 Jorn Barger's Web Resources FAQ looks back at where weblogs came from. Originally, the notion was to log and annotate other websites of interest. People who had their own websites made log entries manually.
A wave of online blog tools, also called blog builders or blog servers, made it possible to maintain a weblog without being very technical or having a website. Here is a September 2003 review of some of the top dog blog builders. From a more up to date perspective thanks to Andreas Ramos we can see that once blogging facilities were available to us masses, we began to blog freestyle. Online diaries have become a popular form and there are weblogs where original articles predominate rather than commentary. "Newslogs" track many stories, from Iraq to fashion. Blogs are generally still link intensive, but what makes a blog a blog is that it is entry driven, with new entries not overlaying the old, just pushing them down the stack eventually to be archived.
In a second breakthrough, the online tools adopted the practice of creating standardized RSS newsfeeds in XML to accompany each weblog. Don't ask what RSS means. Here is what an RSS newsfeed look like. XML is very machinable, so once these feeds were available, programmers began creating widgets for them. The "news aggregator" you set up on your home pc and with it you can subscribe to other people's newslogs. The aggregator automatically checks for new entries for you and displays them nicely in a resume list and maybe lets you click in to see the full article. If you have a list of favorite weblogs, this is the way to go. Another possibility is to display a newsfeed summary for someone else's weblog on your website. Here's one implementation at $15 a year. In addition, you can register your rss newsfeed at numerous sites and have your weblog entries automatically scanned and stored by keyword for others to find. Examples are Syndic8, or NewsIsFree.
Several of the main blog builders also established APIs (application programming interfaces) to connect straight into their blog database, where the webloggers' blog entries are stored, so you could have a client program running on your desktop and write your entries while offline if you want, conveniently, and then blast them all up when you go online instead of having to go through the web interface, slowly, slowly. Plus the same API could be used instead of the RSS newsfeed to aggregate entries.
One big problem with all this is that there are several flavors of RSS and several different API's, and none of them does what everyone wants. Much to their credit, therefore, the technical people in the community have gotten together to design a next-generation API that is certainly going to be adopted by the blog builders. In fairly short order the "atom" API was thoroughly talked over by the interested parties and consensus achieved on the new, improved and extensible design. Interestingly, although one of the RSS flavors was built on RDF (see an earlier article), the atom design team decided RDF was not appropriate for the new API, presumably because it complicates things without any near-term payback. Characteristically, the rejection of RDF was combined with a request to the RDF supporters to please come forward with any design factors that might be useful in adding an RDF extension later. When an either/or design decision was forced, the atom team remained inclusive and conciliatory in spirit. And as it turns out, the atom-->xslt-->rdf formula will likely fill the bill for those who want to work in rdf format.
Something to watch for as atom comes into play, is a period of rapid differentiation between the blog builders based on atom's extensibility. If I am not off the mark, weblog clients will soon be available that store the users' log entries on their home machines. This will make it possible to switch from one blog builder/server to another and carry all one's archives and active logs across to the new server. Therefore, the blog servers will add special features in an attempt to keep their users from leaving and to lure bloggers from other servers over. My guess is that community-building facilities will be important in these new extensions.
In a major move, google has recently acquired the largest blog server, Blogger. This acquisition occurred after Blogger committed to atom, and so far google has not changed course, nor will it, I expect. The weblogger community has no animus against google as far as I can tell and google, 900-pound gorilla physique notwithstanding, doesn't need online antagonists, particularly not of the hacker variety. However there is a tension built into the situation. As the blog servers differentiate, what if someone comes up with a peer to peer design based on atom that brings the server down much closer to the users, so that each blog server would be handling say 1 to 100 clients? Now what if this new peer-to-peer server provides an extension based on its peer-to-peer connection capability that google cannot match? This would be a classic top down vs bottom up duel, and it is not clear, considering that people will be able to vote with their feet by moving their blog from one system to another, who would win. Yes, very likely google, with all its money and other advantages of centralization, such as a "blog this" button on the google toolbar, but not necessarily so. There is a lot of sentiment for bottom up, and that is what google just cannot be, not really, though I expect they will try to mimic the effect.
Another blog development of some consequence is called "trackback" with its offshoot, "Topic Exchange". Trackback allows a weblogger to post a comment on her weblog to another trackback enabled weblog entry and have an introductory snippet (up to 256 characters) automatically displayed under the provoking entry. It is even possible to reply to a reply and so to have a public conversation that can be tracked by a third party tool. Although it is not really very hard for the user, still it can be a bit confusing at first and the snippet that displays is too short to say anything, so trackback has not caught on in a really big way. Not as far as I can tell anyway.
Trackback was designed to enable topical community weblogs sponsored by the blog server. Then individual bloggers that are trackback enabled can post topical comments to both their own weblog and the community weblog for that topic in one go. Replies can be posted three ways, to one's own weblog, to the weblog of the entry being replied to and to the common community weblog for the topic. Good idea, and it is happening to a degree, though again, it has not caught on fully, not yet anyway. Somehow, before long, the desire to aggregate the blogger community by topic and implement cross blog posting will find the technical answer that works.
Finally, in this quick catalog of blogosphere elements, the blogrolling phenom needs to be mentioned. Blogrolling is the practice of linking from a sidebar in one's own weblog to other weblogs that one thinks are particularly worth watching. Then several websites, blogrolling.com for example, which btw also offers a free blogroll widget for instant blogrolling, use the blogroll cross references to list the top 100 blogs – i.e. those that are linked to most often. Taking the blogroll a step further is the daypop search engine with such novel features as wordburst tracking and an improved top 100 list of blogs.
A problem with these top 100 lists is that they do not cover the hundreds of smaller groupings of bloggers, who cross post each other on particular topics. However, Feedster lets you search an extensive blog database and list the results by blogroll rank, which helps sort out groupings of weblogs. (If you search by the phrase "collective intelligence", you come up with a Winer! And that points to How to Decentralize Directories, which is somewhat like the voice of humanity structure that was described in a previous article.
So where does vohP2P, the voice of humanity middleware implementation, fit into this picture?
First of all, vohP2P must make a perfect back end for the perfect atom client/server. Therefore, it must be able to store and retrieve blog entries fast and flexibly.
Additionally, vohP2P should be able to:
2) get-from-web-atom-server, store and feed-to-web-atom-server all entries, including archives, for any given blog, using whatever specialized code is necessary, blogbuilder by blogbuilder. This is to implement the easy transfer of a blog from one blog server to another.
3) form a peer-to-peer group with other vohP2P instances.
4) use BitTorrent to flood a peer-to-peer group with large new files.
5) allow a specialized peer group gateway to pass off requests for your weblog to your atom server if you are online and to another online peer group server if you are offline. (Is this possible, sensible?)
6) store weblog owner information such as name, email address, age, sex and so forth.
7) keep track of a server owner controlled directory hierarchy - including adding new folders, moving folders, deleting folders, tracking folder moderators and so forth. This will implement flexible categories for blogs through an extension to the atom API.
8) allow a folder moderator to link that folder to a folder in another vohP2P server as described in a previous article, with upload/download rules.
9) provide an entry to the MetaWeb. The simple but oh so powerful idea here -- has no one else thought of it? -- is to have a cgi server that mediates between the ultimate user and the web. The MetaWeb will act like a website that captures a user and links out to the web for the user in a frame. But we don't want to use frames, of course. Instead we will wrap the page from the web with a header and a footer that we provide. A simple rule for using a url as a cgi parameter will invoke vohP2P in its MetaWeb server capacity.
The first link into the MetaWeb will likely be to the vohP2P owner's own weblog. vohP2P will then fetch the requested weblog page over the internet, wrap the page with a footer and header and change all the page links so they go back through the same vohP2P program instead of directly to the requested pages. In the header will be a MetaWeb address bar so the user can remain in the MetaWeb instead of using the browser's address bar. In the footer, we add navigation to the vohP2P directory for the current weblog owner, and to other vohP2P instances as designated by the owner. A javascript program lets us rate the page we are on for interest by pressing any numeric key 1 to 9, or to vote y or n on the page for yes/no, thumbs up/thumbs down, and the server stores this information. (Javascript can do this, right?) A 'b' blogrolls the page by the current category, also adding the page to the current category item list.
Of course we will want to mark the link trail stigmergically, and that means either a color scheme for links or possibly the adding of small gifs next to each link to indicate the average rating received by the page linked to. A great feature will be the ability to link into the MetaWeb from different communities, using rating averages from the current community to color code the links. Better yet, we will be able to switch communities without even leaving the MetaWeb by hopping to a different vohP2P server.
10) having the MetaWeb, with its ratings by known webloggers who have provided user information, we can apply the intelligent search capabilities of a previous article and build up the blog meta-data which we will share with other systems.
That has to be it for this article. There are loose ends here, to be sure. Next up is a series of articles about the bloggers who are in the conceptual neighborhood.
|
|