posts are the atomic element of weblogs

Anatomic weblog elements: an idea. But also a URL, a certain emotion or feeling, a respond. Similar to IM, many blog posts seem to be spontaneous. So maybe, yet another atomic element might be the trigger.

Ah the long-awaited follow-up to your Microcontent Client article! Great to see.

Have you been following the k-collector beta? Paulo and Matt are doing some fantastic work organizing weblog posts by topic. This is something I'd like to see in RSS Aggregators, an ability to subscribe to "topic feeds" - in addition to subscribing to an author.

Regards, Richard

Posts may be the atomic element of weblogs, but most real atoms spend a lot of time colliding with other atoms, making cool stuff happen. We (the Sexxi team) are trying to fix what I guess you could neologize as the "lonely atoms" problem. Categories aren't good enough for weblogs; they just aren't natural for the way many people blog (excepting perhaps journalists and people who really like categorization). So rather than categorizing and ending up with your typical unconnected list of chronological posts, we're after a means to automatically link together keywords in posts, both on your own blog and between other remote blogs. See our site for more info, but the end result is that those lonely atoms won't be lonely much longer...

Well, a few quick reactions.

1. part of why chronologic order matters and is important, is that humans (readers) are very bad at diffing at a glance - i.e. I know I last read dashes.com/anil on Friday, so I can easily see what's new - but if instead it was buried deeply into categories.. much harder to read and enjoy.

2. Blogs serve as a historical record of sorts - categories and other ways of sorting can be useful in some contexts, but for enjoyment, for keeping current, chronological order is quite hard (if not impossible) to beat.

3. Consider your side column - while the list of links + commentary is of some interest as a set of individual units (and you might see the same links organized ala a bookmark file into categories); part of what makes it a blog and what makes it of great interest to the reader is the groupings and ordering of the links by date - i.e. that you, Anil, found the following interesting stories in this particular order

(note as well, I've see you have comments that refer to other links from that day i.e. "another one" type notes, which lose meaning outside of the context)

4. Personally rather than another step I have to take, I think the "solution" will be when search engines (ala Google perhaps) begin to recognize "posts" as search results, and also start deeply indexing most blogs vs. just the A list blogs - then some really interesting stuff might start to happen with RSS feeds, search results which return not whole sites but atomic units of posts etc.

just some thoughts/reactions

While the reverse chronological ordering of atomic weblog entries will always be important in the context of the implicit social contract of which you speak, category classification taxonomies will become important as the semantic web emerges. In the future you will see smart weblog technologies that employ sophisticated automatic cataloging algorithms to provide contextual indexes and category menus with or without author intervention. These will produce classified and categorized �molecules� from the �atoms� of weblog entries that will be linked and cross linked to similar topics across the global Internet.

I don't think the reverse chronological order is an absolute requirement of considering something a blog. For my genealogy blog, I wrote a content management system that allows users to read it in forward or reverse chronological order, and to select the family or families they're interested in. The default is all families, reverse order, so that it's like a traditional blog, but you can read it in the order it was posted too, like a timestamped narrative. I think the former is useful for regular visitors, the latter for people who stumble across the site. But then, that blog has a somewhat different purpose from traditional blogs.

Ways I'd love to see Anil organize his daily links:

- By date Anil published the link (as it stands now)
- By date post author published linked post

- By domain
(ie, all of Anil's links to scribbling.net)

- By link popularity
(**least popular, especially - I want to see the things Anil found interesting that everyone else did *not*)

- By topic
How topic is determined, whether by post author or Anil or some standards body, I don't know.

- By cross-reference (internal and external)
All the posts by the post author which references the post Anil linked
All the posts not by the post author which references the post Anil linked (ala Technorati)

There are more, for sure. But that's a start.

Because, you know, categories are not enough.

http://www.scribbling.net/entry/386/

I'm investigating some alternate ways of displaying weblog data. I stopped prominently displaying the post date some time ago. I don't find them especially relevant and they steal focus from the content, IMO.

I'm tempted to use categories but I haven't seen/discovered a purpose that suits my interests — I have an idea though. Keywords, on the other hand, are a big mystery to me.

"- By topic
How topic is determined, whether by post author or Anil or some standards body, I don't know."

I had a brief synaptic firing that said, "Hey, rather than the Weblogger categorizing it, let the reader do that if they so wish." I suck at categorizing stuff when writing; feels like I'm limiting the random synapses of my head when I try to make a decision at the fore or even at the end.

And yeah, categorization is important for the semantic Web. By date works for individual sites [vertical], but to go across the Web [horizontal] needs something else, unless you're just meme-following. ;)

I like that idea - of the reader categorizing posts. I don't like leaving it up to Anil or the post author or some standards body to categorize for me.

So then, in my weblog reader (aggregator, whatever), I can create categories with keywords attached. Like, my Web Dev category can contain the words "PHP CSS XHTML MySQL SQL etc etc", and then if I wanted to view all of Anil's links or posts related to Web Development, my weblog reader software includes all of Anil's posts/links which contain those words in the Web Dev topic.

It'd be nice if my reader was smart enough to figure out that the word "Javascript" occurs over 70% of the time in all the posts that contain the other words in my Web Dev category, and then it could recommend that I add "Javascript" to my keyword criteria for that topic.

But I digress.

What can existing tools do to facilitate a reader-defined topic view of a site?

Should Necho include *more* metadata per post to allow for this kind of thing?

I, personally, don't think so. Let the post speak for itself. We could probably kill ourselves (and our servers) with meta-data.

Yep. I think a reader is going to be smarter than an author, because the reader has perspective. Most Weblogs are truly self-published; I've done enough writing on the personal and the professional level to want an editor at times; this is a situation where I think it could work.

Now, of course, this is a CMS problem, really, because a CMS is going to have to be able to allow that, isn't it? But then there's probably a way to make it work in the metadata as well ... but it'll take a lag of what, 2-48 hours [depending on the site popularity] for it to get properly categorized?

You'd want something where the author doesn't have to go back and edit stuff.

Random synapses ... for once, a reasonable idea. Most of the rest of them aren't, heh. ;)

auto-categorization? bayesian filtering.

I think categories are by and large a red herring when it comes to personal weblogs.

When I relaunched my personal site on Movable Type, one of the reasons is that I wanted categories. And so I started to categorize my posts (internally... I wanted to get a sense of where it was going before I went public with them).

And what I found was that each post was it's own category. Or that each post touched on so many potential topics as to be not meaningful to categorize.

This exercise further cemented my belief that, with personal weblogs at least, the most meaningful organizing point is the *person*. The person is what gives all those posts context, and to a very fundamental degree, it really doesn't matter what the topic is.

When you talk with a friend, do you stay, "on topic"?

Now, there are blogs that benefit from categories. (I run one. But they tend to have more of a zine or publishing bent to them, tend to have a definable mission.

My site is just my thoughts. Why bother categorizing them?

But since you can assign multiple categories to a post in MT, what about viewing "categories" instead as facets? Wouldn't you want to be able to expose your posts through those different facets, not just date? Wouldn't your readers want to browse by similar facets across posts?

(I know the keywords field in MT doesn't do that yet, but there's probably a plugin for it.)

I don't stay on topic when talking to friends. But I do see conversations with friends that branch when we reach familiar or previously-discussed ideas.

Sure, you'd want to be able to browse that way. I mean, consider the Weblog of a college kid Figuring It All Out. That would be fun reading material, in a weird, anthropological way, woudln't it?

I'd love to have a compendium of thoughts of my transition from arch-conservative to slightly-right-of-center. That would be really fun reading for me, and maybe someone else would get a kick out of it. -shrug-

The concept of history is important, and yeah, you want to tie things together. If nothing else ... couldn't you set up a CMS to allow trackback/pingback similar posts in a Weblog?

I'll be honest here and admit that I'm thinking of Noah Grey's old archives, wanting to categorize them in a way but also show his progression.

[This feels like throwing spaghetti at the wall, but maybe a strand will stick.]

I've thought many a time that blogs should be displayable in reverse or forward chronological order. Nothing about blogs drives me crazier than finding a blog that I've never seen before, and wanting to read it from the beginning, like a book. Most blogs are set up so that it's very awkward to do this. Reversing the sort order is so trivial for a computer, why aren't more blogs set up so that I have that option?

Even worse are the archiving systems that generate links for periods of time that don't even have entries. Or systems where I only get one entry at a time from the archives and I have to navigate through many clicks just to get the previous or next entry in the archive. Blogs tend to be pretty light data-wise, why can't I read a month's or 20 entries (which ever is greater) at a time in forward or reverse order?

The reverse-chronological order is just a presentational conceit--it's not what makes a blog a blog. Although I like the idea of it for the front page, I've been thinking of making my archives read in chronological order, for just the reason Simon mentions.

At some point, we're going to get to auto-discovery of semi-objective/universal categories. I've tried to hash out some ideas on the subject. Once we have this, it'll be much easier to aggregate like postings from people you've never heard of.

When you write a review isn't as important as categorizing it. When you have a conversation is important, but cross-referecing it probably more important. Media - clearly - should be based upon categories, but the individual song, movie or TV show meta-daqta is probably more important then it's overall category.

Each kind of micro-content has it's own priorities. I think the reverse chron thing is a unique weblog thing.

This is all quite amusing to me, in an I've-been-on-the-internet-longer-than-you kind of way (don't be put off, I would never hold your youth and inexperience against you). But the later evolution of Usenet newsgroups reached a point where it was taken very seriously as a "new medium" and was experiencing growing pains as a result: some pretty savvy people wanted to break free of the bonds that held Usenet to earth and achieve its full potential as they saw it, just as internet use was exploding and the web was making its presence known.

The problem limiting Usenet, then, was seen as the artificial construct called the "newsgroup". Rather than being a container into which people posted, the newsgroup was seen as a topic denotation that was very granular and open to abuse and neglect : crossposting, topic drift, social discussion in technical groups, as well as all the meta-behaviors noted in Shirky's recent article -- e.g. sophisticated social hacking along the lines of alt.syntax.tactical. There were endless fights over newsgroup creation, either through the clumsy, imperfect, elite-dominated voting system in the "big 7" hierarchy, or through the "rogue" group creation in the anarchistic alt hierarchy. There were also some long-running and deeply entrenched battles over moderation and issues surrounding group charters, and the advent of newsgroup spam led to so-called "rogue cancels" (posts benevolently cancelled by persons not their author), and resultant arguments over the nature and scope of "censorship".

(Sorry for all the scare quotes and stage-setting. Forthwith:)

Thus some solons of Usenet came up with a massively re-engineered Usenet, sometimes called "Usenet 2". This Usenet would not be hampered by artifical topic firewalls: it would have no newsgroups. It would not be subject to external or internal censorship and the inevitable political battles: all posts would be killfiled (or not) by the reader's software, based on sophisticated filtering (not dissimilar to the recent Bayesian spam filters). Topics would form spontaneously and be identified externally to the posts, by reader software and, perhaps, an assist from the server. If you decided you wanted to read about lemurs (*old Usenet in-joke*), entering lemurs as a topic note would show you *all* the posts *in any group* about lemurs (except there would be no groups to begin with). It sounded wonderful: it would end-run all of the nasty social problems that managing the newsgroup namespace and the newsgroup content itself had sprung into being. No person would be able to assert authority, no matter how legitimate, because the poster and the reader would be communicating independently with no needless intermediaries. The poster would drop his "microcontent" into a sea, and the reader would pick up a stream of memes through mechanisms peculiar to his own selection methodology.

Needless to say, this vision never really came about. Some might saw it was because the web grew up right just then, and made a lot of the changes unnecessary. Want to say something to a gazillion people? Put up a web page. Who's to stop you?

But then it was probably unworkable for many reasons, some of which correlated nicely with anecdotes from the Shirky article.

Ultimately his insight that reputation is a rather subjective entity is probably quite important. I don't start my weblog reading day by going to daypop and entering "iraq", "windows 2000", and "science fiction". I read people who I know and find interesting. In fact, I rather like the idea that I *don't* know what someone is going to post about.

This isn't specifically a refutation of the idea that weblog posts may be arranged in ways other than chronological: more it's a cautionary note that an engineering point of view as regards what is "better" is not always socially workable. Just as turning the roughly-grained but consistent conversations of Usenet into meme stew would have destroyed the medium from a social standpoint, doing different things with weblogs, other than as the odd experiment, may eliminate exactly the social strings that make weblogs successful.

Metaphor hell...

Posts are atoms.

Atoms come in various shapes and weights and have varied components. For a periodic table of elements: The Component Blog. Microcontent in many forms.

Molecules are atoms combined. Their properties vary with their elements, organization, and how everything is glued together. Think of this as a subset of a weblog. Maybe a channel, or posts with inferred mutual relevance.

Throw enough atoms or molecules together, and you get complex things. Organisms, crystals, buildings. Newsreaders, journals, blogs, photoblogs, wikiblogs, email clients, calendars, collaborative blogs, vending machine blogs.

That's why the not-Echo project must provide a framework for

1. a rapidly growing table of periodic elements [not just the RSS .92 post structure],
2. attributes and protocols that permit more kinds of interconnection among various forms of microcontent, and
3. a basis for creating new systems and structures from the many microntent flavors and structures.

This is the path to the Adaptive Blogosphere.

Info value!!! vs. info-tainment. Most blogs are infotainment: interesting, like a magazine (I put down the Economist for this), but with limited "value" of the information.

In decision analysis, using Bayesian probabilities to derive expected values, the info that has value is that info which changes decisions, and the value is the increased value of the decision. Few blogs create much value, in and of themselves. For infotainment, chron blogs by date is GREAT. This comment thread�s issue is, I think, trying to save the historical/ research value in (some) blogs, while maintaining the conversational immediate value.

For future research, some sort of author based categorization/ topic selection would be good. Consider forums with threads -- a few folk may be very active on many subjects, and each thread develops mostly "on-topic" until interest is reduced. Or a new, similar topic is launched and the forum moves on to it. Thread histories are likely to be more valuable than most blog archives�but there is less effort devoted to that, so far.

A mailing list discussion explicitly uses the subject line, to keep the posts on-topic. Archives of such posts are also likely to have some research value.

While I like keywords, in theory, I think meta-data focus on the subject line is more feasible. Key words in subjects and topic auto aggregators will prolly be needed for that, in the future.

Besides value of info, there is the filtering issue -- how to get important / authoritative info? kuro5hin tries to use ratings, of articles & comments, but I don't think it quite succeeds. Link counting is a reasonable surrogate, but still a little too undifferentiated. Like a google search returning multi-thousands of hits, a searcher doesn�t want huge numbers of hits, just a few (

Info-architects and Movable Type gurus: I need your help. I work at a public television station. We have lots of shows in different genres. I'm starting to think we should use Movable Type to organize the programming highlights. (more here.)

I'm thinking each programming highlight is a genre-categorized blog post. When you go to the History & Society blog, you see all the shows in those broad categories in chronological order. But, you can click on a subgenre - public affairs - and see just those shows.

Any thoughts on (a) is that a good idea, (b) what's the best way to go about it? Sorry for the interruption... all the recent blog topic chat has got me seeing everything as a categorizable blog.

Hi Anil,
I've been working on pretty much the same principle, with my meme-sized chunks being things like a foaf:Person or rss:item. Yep, it's all RDF based. I think the human-friendly approach to the Big Graph model is in such microcontent chunks. But I don't think it's essential to consider them at a micro- level, being able to scale in a kind of fractal fashion opens up a lot of potential for data management.

But anyway, I've found that if you drop out of the idea of specific kinds of chunks, so any meme-sized cluster can be used and this really open things up, especially your mind (man;-).

I've only been scratching the surface codewise (with my IdeaGraph project http://ideagraph.net) but rarely a day goes by without a new idea occurring to me for something useful in this environment.

Beta, give me the beta you thrifty bastard.
PS: See the next post.