As you may know, there's been a major redesign of the Scripting News home page. But, you may not be able to see that this is part of a much bigger change in the way I do content management, one which is being modernized to fit into the new realtime flow of links, pictures, podcasts, etc facilitated by Twitter, and what I hope will be a thriving ecosystem of feed-like apps that run outside of Twitter.
Things are changing, not just here but everywhere. And I'm moving quickly, and as this happens things break.
Yesterday I posted three items to my feed that were just titles. This was a bug, in some provisional software that patched the new system to the old system. I'm not going to fix the bug because there's going to be a new way to do the feed, and I'm just one person programming here, and I can't afford the time to maintain provisional code. I have to be working on the main line.
And here we have a choice -- continue to compromise for people who use Google Reader, which is a very fussy piece of software that doesn't play by the rules of RSS 2.0, or break Google Reader, just for this feed -- and do things the way the Twitterverse works, which imho is also the right way to do things. Obviously, given the way I phrased that, I'm going the way I feel is right.
The question is this -- must feed items have titles?
The answer is clearly no. No, because the RSS 2.0 spec says so and no because Twitter has built a wonderful ecosystem of communication around the idea that posts can be short enough to be titles. How could you require a title to have a title? It would be ridiculous.
I already have a great linkblogging tool called Radio2 that takes blogpost-like items and turns them into an RSS 2.0 feed of Twitter-able items. It seems perfectly reasonable that Google Reader could adapt to make sense out of these feeds. It's just one piece of software. But they don't move, and that gives others a tough choice. To move anyway, knowing people who use Google Reader will be left behind, or continue to provide content as if Google Reader was the only way people were consuming it, when that isn't even close to being true anymore.
I hate it when users are caught in the middle like this. You guys did nothing wrong other than use a piece of software that works for you. And of course since you want to read my site, I think that's fantastic. Maybe there will be some compromise possible, a way to accomodate everyone without making me, just one person, do the work that a big company like Google should be doing to support you, people who use their software. But right now that just can't happen.
You will continue to get links to my stories in your input stream, I hope. But the full text will not be there. You can click the link to read them on my site, or skip it, or unsubscribe. And I'm sorry to be parting in this way, but we have to move forward, and it looks to me very much like Google Reader is behind us now.
1. I created a new Radio2 category called Scripting News.
2. When I want an item to appear in the Scripting News feed, I just check the box in the category list.
3. I click Post. When I do that, a bunch of processes are set in motion, including a callback script running on my server that copies the category feed to the location on S3 of the Scripting News feed, and pings our cloud server to tell it that we updated. It in turn notifies all services that have requested updates, and they read the feed. All this happens in a couple of seconds, at most.
An aside: One of the benefits of this switch is that all the effort that's been poured into feeds on Radio2 in the last couple of years now accrues to Scripting News. The feed not only has archives, but the archive is described in the feed itself, so apps can easily find it. Podcasting isn't an afterthought, it's fully supported, and not hacked in. And there's so much more.
4. One of those agents cross-posts to Twitter. That's how you see the link on Twitter, and it's why my content, that lives outside Twitter, is archived separately. I'm not waiting for Twitter to give me access to my links, since I never gave it up to them.
This is also consistent with the philosophy of distributing pointers to the content instead of the content itself. That way there is a single archive for it, one authoritative copy of each bit, even though it can be referenced in many places.
I've never been a fan of full content in feeds. I know some people feel it's the only way to go, but I don't agree, and never have.