Comments on: On Twitter as a research problem

By: links for 2009-03-10 « Where is my towel?

links for 2009-03-10 « Where is my towel? — Tue, 10 Mar 2009 07:02:09 +0000

[...] On Twitter as a research problem (tags: research twitter communication ****) [...]

By: dsandler

dsandler — Thu, 19 Feb 2009 16:01:11 +0000

Shriram: Don’t worry! The design is quite straightforward; in many places I tried to find the simplest approach that would work in a decentralized system.

(One small example: using RESTful HTTP APIs and gossip instead of a complex, higher-maintenance P2P overlay that doesn’t play nicely with NATs and firewalls.)

By: Shriram Krishnamurthi

Shriram Krishnamurthi — Thu, 19 Feb 2009 15:57:07 +0000

Dan -- don't turn into an architecture astronaut!

By: dsandler

dsandler — Tue, 27 Jan 2009 05:47:24 +0000

Indeed! The Year of the Ox, as I was informed today.

By: orb

orb — Tue, 27 Jan 2009 04:45:59 +0000

holy christ that (your plot of followers vs. rank) is an awesome power-law. also, happy new year! (hope all’s well with you.)

By: dsandler

dsandler — Fri, 23 Jan 2009 06:14:07 +0000

A very good question. FETHR bears some resemblance to server-to-server XMPP. However, XMPP has a lot of weight I’m looking to avoid (very large specification, streaming XML stanzas, persistent server-to-server connections etc.). FETHR is something you can implement pretty much from scratch in a few hundred lines of code.

By: PJ

PJ — Mon, 19 Jan 2009 21:28:15 +0000

So FETHR:Twitter :: XMPP:AIM ?

I hope. About time. Though I’m not exactly sure why FETHER != XMPP.

By: dsandler

dsandler — Mon, 12 Jan 2009 02:08:54 +0000

Todd: That’s not quite the same thing. You haven’t given a Twitter user connectivity to other systems, you’ve made another system that carbon copies things to and from Twitter. In this case you still maintain two identities and shuttle data between them.

By analogy: a multiprotocol IM client (Trillian, Adium, iChat, etc.) doesn’t solve the problem of non-interoperability of IM services—it just makes it slightly less insane to use all of them at once. I’m trying to solve the interop problem.

By: Todd

Todd — Mon, 22 Dec 2008 18:28:48 +0000

“…there’s no way for a third party to offer Twitter users the ability to follow users of other systems or vice versa..”

Slightly inaccurate. Anyone can construct their own version of Twitter, for almost nothing, using Laconica’s Identica ( http://identi.ca ) and Zeep ( http://zeepmobile.com ). Once up, interop is fairly straight forward using the Twitter API…But that assumes Twitter will support OAuth, which they do not at the moment ( Grrr! )

By: dsandler

dsandler — Fri, 19 Dec 2008 16:54:31 +0000

Matt, I completely agree. Twitter (including its restrictions on content) should be considered one special case of the more general decentralized micropublishing problem.

By: Matt

Matt — Tue, 16 Dec 2008 22:30:23 +0000

Isn’t the question really, how best to implement a messaging or micro-blogging standard that functions something like Twitter?

Who is to say that I couldn’t have a 300 character microblog post, while unsuitable for Twitter, another service or hub in a decentralized model would gladly accept.

Twitter seems to be the most popular micro-blogging service right now but it’s not the only game in town.

By: dsandler

dsandler — Tue, 16 Dec 2008 21:50:02 +0000

@RodBegbie Fair enough. An answer in 139 chars (including the "why research Twitter?" prompt).

By: Rod Begbie

Rod Begbie — Tue, 16 Dec 2008 21:23:56 +0000

TL;DR. Can you summarize in 140 characters or less?

(Also, minor etymological nitpick: Lazyweb predates 2005. I first heard of it via Matt Jones & Ben Hammersley’s lazyweb.org site in late 2002/early 2003: http://www.boingboing.net/2003/01/04/lazyweb-ideas-have-a.html)

By: jstewart

jstewart — Fri, 12 Dec 2008 18:15:32 +0000

You should talk to amislove about the social networking aspect, if you haven’t already.

By: dsandler

dsandler — Thu, 11 Dec 2008 21:04:04 +0000

…but a mass exodus of interesting people would have to do the same.

This gets back to the network effects of Twitter. Any solution that doesn’t accommodate the existing userbase is in serious trouble.

By: Matt

Matt — Thu, 11 Dec 2008 19:51:54 +0000

The monetization of twitter may make or break its utility for some. If the level of intrusion for the average user is low I suspect most people will be okay with it. I would prefer to use a decentralized non-commercial service like Identi.ca/Laconica but a mass exodus of interesting people would have to do the same. I wrote a short post about it back in August.