-
Website
http://blog.feedego.com -
Original page
http://blog.feedego.com/2008/11/why-i-dont-believe-in-openmicroblogging.html -
Subscribe
All Comments -
Community
-
-
Top Commenters
-
Popular Threads
All Comments
What was wrong (and still is?) with twitter is the centralized form. Blogs, thanksfully aren't centralized and we just follow each one other's content easily. Picture this with me: I create a little blog where I'll only post titles without "post bodies". To be informed, all you'll have to do is subscribe to my feed. The people you follow (like me) are actually your OPML file. You want to comment on one of my updates? Send a trackback from your blog.
What's nice here is that I don't have to push my updates to my followers, they pull them. If I have zillions of subscribers, it will be up to me to host my micro-blog on a bigger server, I won't penalize a centralized and shared thing.
Actually there's a lot of benefits in this design, and I humbly think that we should go this way.
little silly in the first place. The thing is it's caught on whatever
it is so it's still a great opportunity to meet others, build
reputation, and more. I think more than anything, approaching it as a
"new" technology has at least unified people and initiated more
conversation. If anything, we now know more about this stuff and it
is more mainstream than it was before. I totally understand where
you're coming from. For now I guess we'll go with the masses.
However, maybe it's not necessary when "opening" all this up to create
a new standard. I'd love to see if Evan P. sees any limitations in
existing standards around this stuff - I don't know if I quite do.
My short answer? Fitness for purpose. Yes, one could take a gazillion different standards and glue them together into a single application. But that's glue. I hate gluing things together, personally. RSS/ATOM isn't exactly right for sharing microblogging posts between servers for cross-server following. It can do the job, but that's not what it's for. It's a five pound sledge when you need a ball peen hammer. And trackbacks? Now we're gluing a non-XML spec to XML based specs. Again, it works, but it's messy.
OMB solves a general problem: sharing small snippets of text across heterogeneous systems in a scalable way. That's a very different problem than what RSS is for (pub/sub rich content sharing).
Sharing (small snippets of text) can me said in another way: (Sharing text) that is small. The size is not the question. The process is.
And oh, yes, thanks for reminding of FOAF - this could help too :)
To put it another way, RSS is a publishing format. OMB is an asynchronous messaging format. It has more in common with Jabber than it does with RSS.
Messaging != publishing. Twitter's architecture mistake was to treat their system as a publishing system when it clearly isn't- and that created a lot of failwhales.
RSS = publishing
OPML = subscription management (the "client" of publishing)
FOAF = relationship/social connections
Trackbacks = relationships between published works
OMB = HTTP-based asynchronous messaging
If we were to draw a Venn diagram, all of these circles would overlap a little bit. And Laconi.ca uses all of them in some way (maybe not OPML, and it's kinda got its own trackback system in @replies), but only one of them is a messaging format.
And IMHO microblogging is more about publishing than messaging. If it was the case than XMPP and other proprietray IM stuff will do it better. When I post something on twitter, I really "blog" a little bit, it's open for everyone of my followers to read. I'm not messaging, I'm publishing. When I post a reply, I comment, and here too it's not actually a message (even if it's destinated to the @person) it's publishing because it's open for my followers/subscribers.
I'm going to stand by this fitness for purpose claim. RSS is massive and top heavy and generic. This makes it useful in many situations, but it makes it overkill for lots of other situations.
XMPP is point to point messaging. OMB is broadcast messaging. It's still messaging. Compare a tweet/dent to a blog post. A blog post is itself a discussion forum (in most blogs). The comments are data attached to the blog post. Parent-child. Bloggers often organize their blog posts into taxonomies, because each post is really contained in a site (parent/child again), and the blogger's identity is the site itself, not any individual blog post. Blog posts contain rich content. Parent/child, pub/sub organization is inherent in blogging formats.
Tweets/dents do not have any child data. Comments are connected to tweets in a steamed fashion- @replies aren't really replies to dents, but to the stream the dent came from. There is no way to organize them into a taxonomy (hashtags are not part of OMB itself, and they're really a little-language parasite, not a true taxonomy). The identity is a stream- the stream is the parent of the individual dents, but there's a big difference between a stream and a tree, which is what blog organization looks like.
IM is a streamed format. Replies have to be connected semantically by the readers. There's no way to arrange messages in a taxonomy. While most IM systems support rich content, it generally sees limited use compared to text content, and it's certainly not the primary purpose.
Microblogging is closer to IM/chat than it is to publishing. It's similar to publishing in that: it's publicly transcripted, it's stateless and asynchronous. It's similar to IM because it's streamed, semantically linked, emphasizes plaintext content.
But where is the equivalent of a private message?
The conversation also is different: If following means feed reading, and trackback responding, you would have to use two different applications for the conversation to take place: blog for posting, feed reader for reading, and then again blog for responding.
Private messages are those old emails, you know, they're still efficient. And there's proven protocols that works pretty pretty fine. Why DO I NEED another box on every microblogging system I'm subscribed to?
For the reading/commenting why should it be done in two different apps? Can't we build a blogging software that reads feeds too? Oh wait, there's tons of "even javascript' widgets that we can insert in our blogs to show what's new in the blogs we read :)
SMS and IM are for text what phone is for voice. Nothing to have with emails. Private messages (or Direct ones in the twitter's termlinology) are really emails without the protocols. Since you have to check a box.
Yes, the privacy/audience is the question, and you're right a DM is a tweet between a sender and a recipient and this exactly the same thing for emails. Better yet, you can do a CC, a BCC with emails...
DMs are even closer to a traditional IM, and even more torpedo the argument that microblogging is about publishing and not about messaging.
I won't say anything more 'cause this is your blog and I feel I am overdoing it.
This can't depend on periodic polling from the subscriber; the number of actual messages posted (a handful per day) doesn't justify the kind of polling frequency needed to have an acceptable latency (<1 minute, per most microblogging users). If we ran it on polling, only ~0.3% of polling calls would result in a positive result.
I looked into trackback for doing this kind of "push", but the semantics of trackback are that the post they reference are a *response* to something the recipient posted -- which isn't normally the case with a notice. In microblogging, notices are pushed to all subscribers, regardless of whether the notice is a response to something they've said before.
Finally, microblogging depends on *subscription*: that is, that I authorize your server to send me your update notices. That's why we use OAuth -- it's perfect for distributed authorization.
I'm the last person to want to re-invent the wheel. That's why I used existing standards like OAuth to make OpenMicroBlogging work. If you'd like to help out and taken OpenMicroBlogging to the next level, I'd love the help -- there are a lot of people who want to see _some_ standard, based on easy-to-implement existing technologies, go forward.
You've done well by starting by the very thing that makes our point of views differ. And this is exactly why I started this post by giving a definition I think that is valid. And depending on what someone chooses as a "definition" of MicroBlogging, one of us will certainly be right. Let me explain:
I personally don't think that microblogging is that mission-critical. For if I really was in any emergency situation I'd better use "classical" means of communication, such as phone or... just my voice to call for "help" :)
It's the way we see this communication that defines what "is" microblogging and "how" it should be designed.
My humble definition is exactly that one given by Wikipedia -not that I believe that what Wikipedia says is for sure TRUE- but because of my idea that microblogging, no matter how "serious" it can be, is not a mission-critical thing and is not "real time" or even close to it.
When viewed as a real-time thing, it sure requires the sender to "push" information toward the recipients, and this will make all your arguments more than valid. But when viewed as a micro-blogging or as a blogging of micro-content, then it's easier and more adequate IMHO to use the "polling" method.
In simpler words, what separates our points of views is that I see microblogging as if "I" was asking how "You" are doing. While you're seeing it as "I" am telling "YOU" how I'm doing.
This is what makes microblogging a messaging question for you, while it's just a "content publishing" for me.
We DO agree on *subscriptions*. I see it as OPML, because the subscriber checks how his friend is doing, and you're seeing it as OAuth because you're seeing microblogging as "I'm telling the world how I am and want this to be known almost in real-time".
I feel like I've repeated my argument many times, but it's really the way we "define" microblogging that direct us to a given design.
Again thanks for your answer and the smart exchange!