The idea, I am thinking, would be to think of RSS not simply as a syndication document of content, but as a vehicle for RPC communication. In this scenario, the channel element becomes a weblog, and the item elements are new posts. The cloud element becomes a pointer to something like weblogs.com.

RSS could serve double-duty here, of course. You would still have an RSS document that provides a syndication of the current posts. What you are adding are one-shot RPC calls with (usually) single-item RSS documents that wrap new posts.

As Dave mentioned, the XML-RPC version would require a variable-sized struct with well-defined mapping of what each element means. In the SOAP version, things would be very straightforward. If you were going for the cheapest solution, you might wrap the the root <rss> element inside a SOAP envelope like this:

<SOAP:Envelope ... >
  <SOAP:Body ... >
     <rss>
        <!-- a channel corresponds to a particular blog -- >
        <channel>
           <!-- new post -->
           <item>


           </item>
        </channel>
     </rss>
  </SOAP:Body>
</SOAP:Envelope>

You can see there's a little bit of cumbersome overhead here. For the rss-rpc model, in the most cases, the only information you need to know about the channel is a unique identifier. The other elements could be specified only if they needed to be updated (i.e., you would include the <title> element of the channel itself only if you wanted to change your blog's title, likewise the link).

On the other hand, the subelements of <item> really begin to shine as blog post elements here.

It's not that RSS 0.92 is, out-of-the-box, a perfect fit for this model. It's just that it looks nice on a first approximation, and it is already well-known by many developers. The learning curve would be very nice. Also any deficiencies could be easily cleared up in the next release of spec.   12:12:35 PM    

You can tell the difference between those prefer SAX and those who prefer DOM by the XML they design. SAX afficianados usually create XML with lots of attributes. DOM fans tend to prefer only nested elements. This is because in SAX, parsing attributes and attaching them to element names is very quick and easy, whereas in DOM, attributes are sort of an unnecessary complication.

Last year I read a comment about XML usage along the lines of: "good practice dictates to use only elements and avoid attributes at all costs." Whoever wrote this must have been a DOM programmer.

I don't mind people saying this, so long as it's stated as an opinion. The workplaces drones hear stuff like this and don't realize that it's just one person's view. Show me where it says in the XML specs that attributes aren't just as much a part of XML as elements?

Imagine HTML without attributes, only nested elements. Yikes!  10:28:58 AM    

I've discovered that I don't like using web services tools that hide all the serialization from me. I like seeing the messages. I like parsing them and composing them in buffers. I think it's because I'm a control freak at heart.  10:11:31 AM    

I'd wanted to use these classes in some kind of web service, just for fun I suppose. In February I finally got around to doing this. Of course, when I sat down to code the web service, I realized that there was a lot of things about the underlying astrodynamical engine that I wanted to fix first. So I spent a week just doing that.

That's the way I usually work. I'm too comprehensive to get things done, I think. I start out to do one thing, and I immediately put myself three steps backwards, in the hole, having to dig myself out before I even get to what I considered to be the "starting point."

  12:57:12 PM    

It's quite a blast. Very rewarding. Developing software is like working on a version of the Sunday Times crossword puzzle that just keeps growing the more answers you fill in. It's sounds Sisyphusian, but in most cases, so long as you feel like you're doing something useful, it's it's own reward.

But one thing about how I work is that I don't get to talk to very many people about my projects. My wife likes to listen, but she is not a developer. Since I know that a higher-than-usual proprotion of those read Radio weblogs are developers, it gives me the feeling that I can share things with those who might appreciate it from a common perspective.

  11:36:41 PM    

Eventually, I seem to zero in on the part of my psyche that actually needs an outlet, and that I can tap into, at least on semi-regular basis, enough to sustain a blog.

  11:14:15 PM    

  11:10:26 PM    

I think Doc is really onto something with this investigation of Real Cities Network. There is something so utterly insidious about the degeneration of newspaper websites, and I hope he keeps up the exposé.

One of the Real Cities sites is the Denver's Rocky Mountain News, another daily I used to help deliver, with my father. We stocked the vending machines around Fort Collins out of the side of an old VW van in 1980-81. Most nights he did it by himself, but I gave him a hand in the summer, espeically on Sundays. It meant getting up at 2 in the morning for a six-seven shift around town, ending up at the Ever Open café up on North College.

The Rocky Mountain News is a tabloid. It still holds its own as a morning daily in a medium-sized city with two morning papers.

But the web site sucks. I occaisonally visit it when I want some Colorado news, but the pickings are always slim. You'd expect this would be the first place I would turn to find out what's going in back in Colorado, but unless it's a really big story, they probably won't even have even a blurb about. It's so far away from being the "web version of the print edition" that the two enterprises have almost nothing to do with each other.

Much of what they carry is actually national newsfeed stories. My question is who cares about this crap?. If I want national news, even if I live in Colorado, I wouldn't go to the Rocky Mountain News site. I'd go to CNN.com, or the New York Times site.

The Rocky Mountain News web site should focus exclusively on Colorado-only news, for which there is no "guaranteed-to-have-it" web source. It's amazing how the dot-com era came and went and we are left with worse media news sites than we had when they started. They have nicer graphics, to be sure, but the content is lousier than it was back in 1996.

The future does not look very good at all.  1:20:04 AM    

We lived in Ames, a college town about thirty miles north of Des Moines. The Tribune was Des Moines' evening rag. It was published together with the Des Moines Register, the more popular morning paper.

Following a familiar plot that has played out across America, the Tribune ceased publication in 1982, leaving only the Register. Funny how the web is. I can't find a single page on Google that talks in detail about the Tribune's historical demise. This page is the best I could do. You'd think everything like that would be on the web by now.

As a newspaper boy for the Tribune, I was subcontractor. I didn't get paid by the Register & Tribune Company directly, but by the husband and wife couple who had the routes for both papers in the student housing complex in which we lived. They delivered the morning paper and hired me for the evening one. I used to get paid with checks that had the Iowa State mascot on them.

I have a lot of memories about marching around in the evening after dark in the snow of central Iowa with those big ink-stained canvas bags. In retrospect, it was fun.

My only regret is that I didn't save those Tribune editions with the evolving map of South Vietnam in 1974 and 1975. What a bleak time it was. There was a cancerous pessimism of defeat that kids today couldn't possibly understand, given what has happened in Kosovo and Afghanistan. There was a feeling that the tide of history was completely against the U.S. There was nothing anyone could do but watch the provinces turn dark as they fell to the Viet Cong. The Central Highlands went one by one, until only Saigon remained.

I kept the papers for a couple years, but threw them out during a move when my dad pestered me "You really want to save these?" I told him to throw them out, not because I didn't want them, but because I thought that's what a grown-up would do.

But I would have loved to have saved those maps I delivered. I would scan them and make a flip-through web site. It's not that you couldn't do that now, by making your own maps. It's just that I believe the art of map drawing for newspapers is one that has largely been degraded by the use of digital tools. Subtleties in typography and line width are lost by relying too much on digital circuits to make the decisions for you. It's the way I feel about physics articles in my field too, that the old way of hand drawings actually went a long way to helping you understand a concept, the same way using a slide rule taught you more math than a calculator does.

  5:48:33 PM    

  5:16:46 PM    

Dave Winer is often accused of overhyping his projects. I understand evangelizing. I would probably be that way too.

It's strange therefore that I find him underhyping what is probably the most significant recent development in the Userland platform, namely the adoption of the Blogger API.

Yes, you've seen a lot about it, but given the magnitude of this development, it has actually been underplayed, in my opinion.

Why is this so significant? Because it established the Blogger API as the de facto standard for blogging web services.

For the first year or two after XML 1.0 was finalized, there was tons of criticism about how "no one really uses it." With the development of RSS 0.92, a "real XML application" was developed, one not handed down from on high, like RDF, but one homegrown, with the potential for changing the way the Internet is used.

Don't you find yourself rooting for these homegrown XML applications? Isn't it somehow a triumph of the spirit of the Internet when they become widespread standards without having been previously certified by some governing standards body?

You can perhaps track the rise of the "real Internet", this "Homebrew XML suite", by the emergence of these new XML applications. Some are document formats like RSS 0.92, OPML, and the Weblogs.com changes.xml format. Others like Blogger API and the Weblogs.com pinging API are message protocols.

The exciting part is that there will be more and more, filling in gaps and extending functionality. It's going to be fun to follow.

My obvious point is that the real platform that is emerging in content management is not any particular software application, but these expanding "Homebrew Suite" of XML protocols.

This begs the question: when Mircosoft finally barges into this game, what will they do about living with all these pre-existing APIs that have sprung like weeds in their backyard?

  6:03:37 PM    

The change in the name of my weblog indicates that I have decided to purchase Radio. The debate is over. The stated propositon that "Radio really sucks" is false.

That doesn't mean I'm completely happy with it as software. It still runs really slowly on my Mac, and I shut it down unless I'm explicitly updating this weblog. I will continue to update my Blogger weblogs using the web interface for now.

So why did I decide to buy Radio? Three major reasons:

1. I like being part of the Radio community. 2. I can imagine using Radio more frequently with a faster machine. 3. I think no matter how I use it now or in the future, Radio is too significant of a platform not to keep abreast of.

Like I said, this doesn't mean I'm completely happy with Radio. It just means that I've wrapped my head around it's functionality enough to shell out the money for it.

  5:28:08 PM    

The prices are aimed at large corporate accounts. For the casual writer like me, I will simply forgo paying altogether, and use the links for as long as they last. Thus they get no money out of me at all. By lowering the price by say, an order of magnitude, from several dollars per article down to, say, twenty-five cents, I believe they would receive much more volume, and more revenue. They would also provide a much better public service, assuming that this is still part of their mission as a business.

In short: they want 2.50 for an article, last time I checked. Isn't this absurd considering they charge only seventy-five cents for the entire printed addition?

Today, after all this sturm und drang in my last entry, I came upon the very rare instance when I would want to actually pay for an article. It happens that this week, the Times Sunday Real Estate section's "If You're Thinking of Living In..." column featured my neighborhood on Staten Island.

I put a link to this article on my weblog about New York, since I write about my neighborhood a lot. But since I knew the link would expire eventually, and I like to have people be able to read my past entries, I thought this might be an instance where I would pay money for some way to permanently link to the artcile.

On the Times site, next to the article, there is a link entitled:

Click here to order reprints or permissions for this article

I thought: maybe I can buy a permanent link. Isn't that what the link implies?

Of course, need I tell you that the link doesn't work at all?

  2:20:59 PM    

But Dave Winer is right: newspapers, by and large, just don't get the web. It's maddening to watch them come up short time and again.

Like many people, I have a habit of accumulating hundreds of browser bookmarks before I get around to cleaning them out, as I did today. Since the page titles of the bookmarks are often little help for remembering what the page contains, I click on each bookmark to verify what is is. A third to a half of them are news articles from Yahoo or from a newspaper site. Many of the bookmarks are a month or two old by the time I go through them.

You can imagine what happened today when I clicked on most of them. I got the "page no longer available" link, with a suggestion that if I wanted to retrieve the article, I would need to pay for it, usually a fee of about two dollars or more.

Users of the Times web site are used to these shenanigans. What I discovered was how universal this policy has become among newspaper sites--Chicago Tribune, Dallas Morning News, the newspaper in South Bend, Indiana, etc.

To me, this displays a very large disconnect between "what the web wants to be" and what newspaper sites are providing. Newspapers are by-and-large abdicating their potential roles as information archives.

I do a lot of research on town planning for a book I'm writing with a friend. I clip articles, mainly from the Times. Typically I build up a stack of several months before going through with my scissors. Last summer, I decided to "go digital" and store articles from the web site. My wife was very happy about this, since it meant the end of the stacks.

What a mistake. My links started expiring and I was screwed. Back to stacking the hard copies. I could copy the stories to my hard drive, but I find that I lose track of the file names easily. The old way of cutting newsprint is still the best for my purposes. It's a lot less work.

It is inevitable that all the archives of newspaper sites will become publicly searchable again. The model they are pursuing will fail. Right now they are justifiably desperate to make their sites self-supporting, and this "pay-for-articles-older-than-a-week" philosophy is the current fad. Thankfully it won't last, but I suspect we will have to endure it for another year or two. In the meantime, I avoid links to newspaper articles in situations where I want to the link to be useful at any time in the future. This cuts out a lot of good stuff on the web for research projects.

But I have more gripes than that about newspaper sites. One thing I learned from my stacking papers for clipping is how incredibly insightful it is to review the headlines from several months all-at-once in sequence. It's as if history suddenly has a definite plot, one you didn't recognize as it was unfolding.

This would be a perfect function of the web. In my fantasy of the perfect web, I would like to be able to click through a view of the Times or any other newspaper web site day-by-day for every day since they began their site, in the format that the site used at that time. What better way to re-experience history from ground level, to glimpse the elusive themes that drive people and nations?

(for sites that change throughout the day, you could have a midnight snapshot, perhaps)

Yet I know of no newspaper site that provides this kind of "true archiving". How easy it would have been to engineer this from the beginning. How hard it would be now.

  12:42:35 AM