18 comments

• 

  AndrewBurton

  2 months ago

  Wouldn't something like the OPML Editor + DynDns work as a decentralized Twitter? As long and your server is running in the background, people can grab updates from you every few minutes; and because your server is always running, you can push updates to it via different channels. I would guess...and it's only a guess...that might be why you reworked the Frontier timer a while back, to make reaching out and pulling feeds to update a list take fewer resources.
  reply
• 

  cshotton

  2 months ago

  Yep, as my note above. You just need a directory service to find your buddies. No need for a third party to touch your messages unless you allow your friends to store and forward to others in your shared community.
  reply
• 

  Davin

  2 months ago

  Twitter is boring and has never been more exciting than an IRC channel.
  reply
• 

  Cole

  2 months ago

  Ah, but IRC won't send SMS out to you. That makes Twitter a neat kind of web service middleware.
  However I can do without the list serv style chatter.
  reply
• 

  cshotton

  2 months ago

  Twitter would best as a "rendezvous" service for a real peer-to-peer messaging service. A client-side server that has friend lists and broadcasts message updates (a la real-time NNTP) is all that is really needed to replace the current incarnation of Twitter. The "rendezvous" service is just needed to coordinate discovery of peers and friends who don't have static addresses. Register your server when you come on-line, friends can look you up, then all comms happen out on the edge of the network. We prototyped this back in 1999, but the market wasn't ready. Sounds like it is now...
  reply
• 

  AndrewBurton

  2 months ago

  Is NNTP not real-time?
  reply
• 

  cshotton

  2 months ago

  Well most NNTP implementations batch up message exchanges for inter-server comms. I was thinking more of a system that just pushes individual messages as received, or batches them when friends/peers are off-line.
  reply
• 

  AndrewBurton

  2 months ago

  I gotcha, although I keep thinking that a "pull" system is better than a push. That way if you're off-line, there's no wasted cycles for your friends.
  reply
• 

  paul at softomic

  2 months ago

  Goes to show that to play with the big boys, you have to be one of the big boys. I read somewhere that it costs Google around $1 million per day to operate YouTube. I'm not sure what kind of funding Twitter has, but at the rate people are throwing their money at GOOG, few start-ups have the ability to set up a truly enterprise infrastructure. Either way, I wouldn't be so quick to count out Twitter especially with complementary services emerging like qik.com.
  reply
• 

  Den Stockman

  2 months ago

  Plan B is good, I think. But I don't know...
  reply
• 

  3KWA

  2 months ago

  Could Jaiku be plan B for (ex) Twitter users?
  reply
• 

  Scabr

  2 months ago

  Yep,there are Twitter problems,but see stats http://www.alexa.com/data/details/traffic_detai...
  reply
• 

  Dito

  2 months ago

  I was not a die hard twitter user, but I used to a few times a day. I am most certainly over it, as my last tweet indicates: http://twitter.com/ditojim/statuses/826776994
  
  I feel like the internet has been a bit empty the last few weeks, but I was unable to put my finger on it. I felt like it was running slow. I now realize that this could have been due to my not checking twitter as often, and thus not getting the same stream of noise that carries most of the tech news over the internets.
  reply
• 

  James

  2 months ago

  All the Twitter guys need to do it take a step back, re-assess what service they're providing and make some bold tech decisions. All their whining about getting databases to scale with RoR drives me insane. Why is Twitter activity even touching the database?
  
  Do the maths: daily twitter activity amounts to a few hundred megabyes of data. They can EASILY store a week's worth of stuff in memory without needing the database (apart from asynchonously backing up the in memory graph of course). All these weasel words about the N-N problem are symptoms of them solving the wrong problem and/or using a fundamentally wrong approach. It's up to them what problem they want to solve, and I can guarantee that if they let that guide their tech decisions there'd be no problems.
  
  If they're too lazy / stupid to do it themselves, just buy IBM MQ, it would eat puny workloads like their's for breakfast.
  reply
• 

  Mariva

  2 months ago

  If this is true, it makes me sad. Twitter is so much fun and useful. But you're correct, an open-source, peer-to-peer version of Twitter (if that's even possible) would be far preferable.
  reply
• 

  w.d.watson aka megalar

  2 months ago

  First off, I must say that I think Twitter is far from dead. Certainly not a ghost town. I follow about 180 people and receive ~80 tweets an hour. That being said, I do agree that it should decentralize. Many people have an Idea for a XMPP based system where people would register with Twitter (to avoid name collision) and then subscribe to the nodes they want. The Twitter public timeline would then be handled by rss. This would take the strain off of Twitter and put it where it belongs, the heavy users. I think this is the only way the service can continue to function.
  reply
• 

  Joubin Houshyar

  2 months ago

  Solution is clearly in a distributed name server, but it doesn't make sense to limit that to twitter functionality given that we can use that to cover not just communication but also finance and even personal currencies in a social networking context.
  
  There are 3 fundamental elements here and it does not pay to conflate them:
  
  1 - Identity
  2 - Presence
  3 - Function
  
  Twitted bits of information between parties (established by 1 and 2) is such a tiny tiny part of the functional dimension that it does not seem sensible to build a distributed identity and presence service (1 and 2) merely to service an enhanced twitter like functional service.
  
  Think big.
  reply
• 

  Berwyn

  2 months ago

  Hi Dave,
  
  Crosbie Fitch has suggested to us that microPledge.com might be quite useful to fund a community-written version of twitter. His main concern is to make it a distributed service to avoid the slow-downs that are currently crippling it, and to make it a publicly owned (GPL) system. Whether you do a new GPL system or take the easy route and modify the current MS-PL one is entirely up to you.
  
  We are not Twitter users, and so we would have a poor shot at championing and posting specs for a project to do something like this. Crosbie suggested that you might be interested in championing this.
  
  Our experience is that a better-spec'ed project gets more traction. It's also important to have a project linked from popular forums, so if you could get third-party twitter forums on board, you'd have a winning ticket. What's your take on these issues?
  
  It looks like some of your subscribers on this entry may also lend a hand in the spec.
  
  Cheers,
  Berwyn
  reply