DISQUS

Harold Gilchrist · 1 year ago

I don't believe the answer to decentralized Twitter is RSS. I believe the answer is for anyone to be to run a standardized microblogging protcol server (mbps). Wasn't that the answer with the web? ie. http. If the Web was designed like Twitter (pne centralized server) is now we all would have all of our web pages on the same web server and have the same problems that are being experienced by the twitter net. Decentralization needs to start at the server. Someone or group needs to start standardizing a microblogging protocol that would be lead to an open server that can be run by anyone and supported my all present and future micro blogging clients/catchers like Twirl. Yes RSS is in there somehere but is not the core architecture that is needed to solve the decentralization problem of Twitter.

Sean · 1 year ago

Isn't twitter based on XMPP/Jabber?

Bob Lee · 1 year ago

We don't need support from Twitter. We can put <link rel="alternate" type="application/twitter" .../> tags on your home pages (which are linked from Twitter already). It requires one more level of indirection, but it utilizes an existing abstraction.

Jim · 1 year ago

Thats exactly what i was thinking as i read this article.

Its not necessary for Dave to host our feeds when we all have systems in place on our blogs that already do such a thing. Plus we could serve it through feed burner to take any real heat off of our own hosts.

smbeebe · 1 year ago

I already have RSS feed of my entire twitter stream, however, your idea is better because it differentiates each users' stream with an "alternate stream" hosted elsewhere which segments your twitter traffic and keeps a local copy ...smart since we already know that twitter sometimes goes down.

Rex Pechler · 1 year ago

Seems like a good idea, at least in principle. But this backup system would work best if everyone had it set up... but then if everyone did, your Amazon S3 bill would rack up pretty quickly. So, how to decentralize that even more? maybe ad-hoc hosting from our devices = p2p? how?

dshaw · 1 year ago

You could always take the original S3 hosted feed and then feedburn it or something like that to minimize costs.

dave · 1 year ago

First -- it wouldn't cost that much. Feeds are tiny. You want to see the costs rack up? Try hosting Obama's race speech MP3 and having the top hit for Obama Race Speech on Google. That's some real money. :-)

Aside from that, a required disclosure -- Switch-A-Bit -- the company I am co-founding, would almost certainly pick up the tab for hosting the fees, so you don't have to worry about it coming out of my pocket, it's very likely it wouldn't.

dshaw · 1 year ago

switchabit.com ? Sounds interesting.

Rex Pechler · 1 year ago

oh! just read that article.... hosting on our blogs of course.... i see. so, how do we build the same connectivity/API's to interact through our own systems?

I wouldn't mind using my blog as a proxy for Twitter, where it would act as a backup and archive for my messages and then post to twitter... but I really like the Twitter IM and SMS interfaces, so how does that play together?

Rex Pechler · 1 year ago

I think I've got it:

(I'm just thinking out loud here, hope you don't mind... also, I use WordPress, and PHP so I'll use that as my example system here)

A php script that follows the Twitter API format for posting messages. This way I could post easily, and maybe clients could be adapted easily to work through my server instead of Twitter's.

Next, a php script to archive any messages that I post directly to twitter, so I can use the current Twitter interfaces and have a backup.

I would then have a reliable microblog feed (through a third script) for when Twitter goes down... But what about other people? Assuming other people would set up the same or similar systems, we could create a directory of their backup feeds. This could be promoted and collected through a twitter bot, and hosted somewhere publicly.

... and then finally a fourth php script, this one to get data from my twitter friends' backup feeds. It would keep a list of the people I follow, and then ask the directory for (and store) the location of their backup feeds.

Thoughts?

Derek Gathright · 1 year ago

Well the question I have is do you post to Twitter, then have a script that scrapes your Twitter feed, or... post to your own feed, then have a script that posts to Twitter (and Pownce, Jaiku, etc...)? This decentralized idea appears to be about owning & hosting your own feeds, so I think it would be better served to have your own feed which auto-posts to Twitter and whatever other services you like.

If we are going this route, we might as well incorporate it into a self-hosted social network. In reality, your Facebook/MySpace status updates and your Twitter updates serve the same purpose, start with that, but plan to later incorporate standards for a personal profile & XFN, then who knows beyond that.

Sure all the tools are available nowdays, hell, we're pretty much talking about microformats and the semantic web, but we still lack leadership to provide everyone with a vision for exactly how everything will work together.

Jesus, now I'm gonna be thinking about this all night. Thanks DAVE! =P

Rex Pechler · 1 year ago

I'm proposing to build a WordPress plugin that will both allow you to post straight to it, and to scrape Twitter. We can call it the WP MicroBlogging Plugin... and it will be open source and extensible. It could also run as a standalone, without WordPress, and just provide a feed/page.

Derek Gathright · 1 year ago

The issue I see with piggy backing off Wordpress is that yeah, the capability is there, but I don't want my tweets in my blog feed. I initially used Twitter Tools to do it, but it diluted my blog so much that I eventually had to turn it off. I think there is a clear distinction & purpose for microblogging vs. normal blogging.

I think there is a real problem with decentralizing twitter and throwing it into a microblogging genre, then trying to mix it with different networks while still being able to converse with users that still has to be resolved. See the comment I just posted at the bottom of this discussion. If we don't mix it with Pownce, Jaiku, etc.. then there's really no point to this whole thing. Ok, we get a "backup" feed for the downtime Twitter has, but that will eventually be resolved and twitter won't have downtime.

Rex Pechler · 1 year ago

I want to add features on top of Twitter. Like, add custom commands to the
API, like delete_last (to delete your most recent post, if you made a
mistake)... This could be part of a wordpress-based distributed social
network. (isn't there something in the works?)

Also, we can separate microblog-posts and normal blog-posts within
WordPress. With options like "Daily summaries" or "Summarize all tweets
between each blog post" or "don't create posts, but show tweets in my theme"

Bill · 1 year ago

I'm confused as to what Twitter's motivation to add the feature would be.

Derek Gathright · 1 year ago

Because it benefits the user.

This would be to Twitter what Wordpress.org is to Blogger or Wordpress.com, and last I checked the latter two were still worth quite a bit of money. Sure Twitter would lose hosting for 1% of us geeks that want to do it ourselves, but the explosion of it as a standard would more than make up for it.

Erik Giberti · 1 year ago

What about a torrent style system for hosting of the feed? That would let you replicate the feed more broadly so high traffic feeds could be digested more easily and it doesn't rely on the user to set up machine instances etc

ka9dgx · 1 year ago

I did some back of the envelope calculations about twitter... and put them in a blog post. - http://mikewarot.blogspot.com/2008/04/guestimat...
I also put the spreadsheet online (thanks to google docs) -
http://spreadsheets.google.com/pub?key=phKIfY2Y...

I estimate that the aggregate of ALL tweets is somewhere around 3k second, or 11 Mb/hour, or 264 Mb/day. I assume 332 bytes/tweet, which is very conservative.. so actual bandwidth would be less.

The trick is of course, that the 3k/second has to be tapped for the appropriate output streams. It could be done in a distributed framework.

--Mike--

kleevr · 1 year ago

OVFO, one vote for open

Richard · 1 year ago

I've recently written a Twitter client and would definitely be interested in something like this. If you get a hosted service set up let me know.

dave · 1 year ago

I'll put up a hosting service today. I'll post a note when it's ready.

ka9dgx · 1 year ago

While I look forward to a distributed twitter replacement, I think that there is more value in the silo model than we might have all suspected. It's important to make sure we deliver ALL of the social value, or it'll never work, no matter how clever the programming.

Here's why silos work: http://mikewarot.blogspot.com/2008/05/why-silos...

Arik Fraimovich · 1 year ago

I gave a lot of thought to this issue when you first started posting about it two weeks or so ago. The thing is that twitter is more than just a platform to publish your tweets on. It's a communication tool, and therefore it's not that easy to decentralize it. If twitter is down, and all I can do is post my tweets it has all most no value to me. The value of twitter is from the conversation with other users.

Unless we come up with a solution to decentralized twitter streams and conversations, we won't have something worth. Maybe creating something over the jabber protocol (or at least very similar to it) will make it work. Because as far as I know, different jabber servers can talk one to each other - and that's exactly what we need with twitter.

And regarding the support of twitter for location of the alternate feed - we can just put it's address in the location field.

dave · 1 year ago

In my experience, bootstraps work one step at a time. Start with a piece of the functionality, get it working, and use it to get the next level operational. I've had good results with this. Watch the FriendFeed guys -- they're doing it. I want to do it outside the confines of a company cause they can go out in a lot of different ways (not all bad for them, but many of them bad for users). The problem is their interests and ours diverge as the network gets bigger. So if they were to go away we'd be left with NOTHING. I want to be sure I have something. What I have might not work for you, but I think I see a way for it to work for what I do. To me the conversation is not what it's about. I've always had this differnce with a number of bloggers too. I like small pieces loosely joined, not just in principal but in practice. In the meantime where's the harm in tool developers backing up the users stream in a feed on a serve outside of twitter.com.

Arik Fraimovich · 1 year ago

I guess you have a point there. As the famous proverb says - they didn't build the Great Chinese Wall in one day :-)

Keep up the good work, and I hope to help when I'll have some time.

xlogik · 1 year ago

One of the guys from process-one blogged about a possible Twitter like platform built on the eJabberd server. http://www.process-one.net/en/blogs/article/int...

Just thought it might be of interest and worth a look!

Derek Gathright · 1 year ago

Here' s one problem I just thought of... One of the main attractions with Twitter is the @ replies and discussion you have directed at twitter users. On a decentralized network, how do you address this issue? Namespace it with @twitter:davewiner? Seems a bit far fetched that would catch on, especially with the extra characters it consumes. Sure, the @reply is essentially a trackback, but that requires that you pick a specific status update to reply to, but what if I just want to direct a general comment towards someone?

I think that is the major stumbling block to decentralizing Twitter. But, if all we want to interact with is Twitter and not involve the other projects (Pownce, Jaiku, etc...) then problem solved. (Not really the best "open" approach though.)

P.S. Looks like opentweet.com was registered yesterday (by someone at angelsoft.net), I'm gonna guess it was after they read one of these posts.

dave · 1 year ago

...or you could simply use yonkly.com when emad finishes the open source code -it's a clone of twitter, and uses twitter's api to call in and out...that way you can have your very own twitter for you and your close friends but still dump into the pool...