Centralized versus distributed

[Josh Fraser]

Jesse, thanks for kicking this off and pulling us together to talk
about this. I whole-heartedly agree with your desire to open up the
protocols around twitter.

The biggest question I have after reading your post is what you are
thinking in regard to having a centralized system. Can you clarify
what you are thinking about decentralizing things? The post about
Tornado makes me think we may be diving too deep into implementation
details when we need to start off at a more conceptual level. It
sounds like you are thinking about building a clone of twitter to run
on app engine instead of doing something more revolutionary. The
problem with twitter goes far deeper than the fact that I can't read
their source code.

I feel a better conversation would be on how we can change the model
so that the data is stored across multiple servers that are controlled
by the content creator (or their selected provider). I disagree with
several of the points you made regarding RSS. You said "we've been
trying too long" to get adoption of an RSS pubsub model, but remember
it wasn't until this summer that the new protocols started taking
off. This stuff is still brand new! If there has been any organized
effort (besides a few blog posts from Winer) to get adoption of RSS by
twitter clients, I must have missed it.

I also don't think you're giving the various twitter clients enough
credit. When I look at the various twitter clients I see a group of
companies that move very quickly and have proven over and over that
they are willing to jump on new technologies around twitter. For
example, I just pulled up my Tweetie client and they have integrations
with 7 images services, 5 video services, 6 url shorteners, 2 read it
later services, tweet blocker, follow cost, favstar.fm, overlapr,
favored and tweeteorites! It seems reasonable that given enough
momentum they would be willing to jump on an RSS-centric version of
OpenTwitter as well.

I feel pretty strongly that the solution isn't to create another
centralized system. I'm not interested in creating a copy of twitter
that just runs on someone else's servers. If that's all we're doing
we're just moving the problem instead of solving it. If I'm totally
alone here, that's fine. I just wanted to get us talking about this
before we get too far down the road.

Again, Jesse thanks for kicking this off. I look forward to some
spirited discussion!

[Ryan Altman]

I am in agreement with this questioning, and I believe looking at a
more abstract model which can be adapted to current structures is
necessity.

"I feel a better conversation would be on how we can change the model
so that the data is stored across multiple servers that are controlled
by the content creator (or their selected provider). "

Generally, macro-blogs are currently in this model, where bloggers are
generating content across various services/domains in a distributed
manner. Unfortunately, blogs are not typically setup to be both the
publishing and subscribing software. A few big services come to mind,
like Tumblr or LiveJournal, but Tumblr is a "closed" service and
LiveJournal is not much different from any other blog.

What OpenTwitter sounds like is the laconica software, which is
"federated" for the most part, although the OpenMicroblogging protocol
is rather constricted and offers little more than a clone of twitter.

[Jesse Stay]

Ryan, the issue with OMB and Laconi.ca is it revolves around a single content source (being Laconi.ca-based setups).  A Wordpress or Tumblr can't take their existing content and output it for Twitter clients using Laconi.ca.  I'm hoping we can come up with a gateway, and perhaps even language libraries that one can just plug and play into any environment, and take data from a variety of standards-based formats, then output those formats into a Twitter-like syntax that any Twitter client can read.  The more that adopt this, the more a "standard" the Twitter syntax becomes, and the Twitter model becomes much less centralized.

Once enough are adopting it, we can begin to standardize that "Twitter-like syntax" or whatever it ends up being called.  The idea of this group right now is to make it as easy as possible for any content source to adapt their content into a Twitter client-supported format.

The idea here isn't to discourage people from using standards already there such as RSS, ActivityStrea.ms, etc.  It's to utilize the tools that are already out there and decentralize the control that Twitter has at the moment.

I should also note this isn't meant as a spurn against Twitter - my hope is that Twitter can join us in this effort.  The idea is to decentralize the microblogging space here.

Jesse

[Ryan Altman]

I fully agree, actually.

Wordpress/Tumblr => Laconi.ca => Twitter <status> is a pointless
transform.

Wordpress/Tumblr => <status> is the most logical, and I think both of
these services have built the twitter-api format into the systems
within recent weeks.

I highly agree on the "Twitter-like syntax" that you mention for the
<status> as well as the <user> structures. Those formats, along with
the necessary api methods provide a standard interface, and any
service can implement on whatever platform/language they choose.

Generally,
Service API (xml) + XSL => <status>

This is very simple process when transforming XML with XSL. many
services have XML output (even RSS) and so long as they link the
necessary style-transform, status elements should easily be generated.
I would even venture to bet that proxy-transforms are in the future
for services that may be slow to provide translation to the status
format.

I am not a twitter user, but the concept is something I do advocate.
My only wish is for being able to interact without being attached to
the twitter system (decentralize). IMO, there is no reason why the
status-update service should remain a closed channel, as it has value
similar to that of posting blog entries (which anyone can comment on)
or email (which is not restricted to domain.com).

So, getting many services to use the twitter syntax is something that
may just happen naturally, but having them all inter-operable is
another story. I for one, do not want a bunch of various accounts just
to communicate with users on each service, and I think that is where
twitter is failing atm [, but works with OMB].

Ryan

On Dec 31, 2:26 pm, Jesse Stay <jesses...@gmail.com> wrote:
> Ryan, the issue with OMB and Laconi.ca is it revolves around a single
> content source (being Laconi.ca-based setups).  A Wordpress or Tumblr can't
> take their existing content and output it for Twitter clients using
> Laconi.ca.  I'm hoping we can come up with a gateway, and perhaps even
> language libraries that one can just plug and play into any environment, and
> take data from a variety of standards-based formats, then output those
> formats into a Twitter-like syntax that any Twitter client can read.  The
> more that adopt this, the more a "standard" the Twitter syntax becomes, and
> the Twitter model becomes much less centralized.
>
> Once enough are adopting it, we can begin to standardize that "Twitter-like
> syntax" or whatever it ends up being called.  The idea of this group right
> now is to make it as easy as possible for any content source to adapt their
> content into a Twitter client-supported format.
>
> The idea here isn't to discourage people from using standards already there
> such as RSS, ActivityStrea.ms, etc.  It's to utilize the tools that are
> already out there and decentralize the control that Twitter has at the
> moment.
>
> I should also note this isn't meant as a spurn against Twitter - my hope is
> that Twitter can join us in this effort.  The idea is to decentralize the
> microblogging space here.
>
> Jesse
>