Twitter, Imagined Identity and Flux
Posted 5/05/2008 08:17:00 AM |
Part I: Economies of Cooperation
As someone who spent almost seven years working in open source, I'm the last person who will argue against openness, but there are a few things wrong with TC's notion. Most important, "open standards" have to work in cooperation with business, not in fierce competition. As anyone who has worked in the industry knows, free software is anything but free. Both the cathedral and the bazaar have extensive present and built-in costs.
So what does this have to do with Twitter? The idea of the "open Twitter killer" is built on an open service model. Open services are the Web 2.0 version of open source software. Just like open source software, open services have built-in and present costs: hardware, data centers, staff and developers. Because open services are standards-based, there is implied future-protection justifying the provisioning costs. Putting it more bluntly, open services can't exist without business, especially at Twitter-scale.
When the founder of DataPortability.org publicly brainstorms ways to kill Twitter on Techcrunch, he's taking the movement backwards. Anyone with engineering skill can dream up a way to "open" a service; the real challenge is bringing companies in to the fold to support open services. Talking about how to kill them is not a good way to do this.
Part II: Interaction and Next-Gen Social Networks
This brings us to our second point: does an open Twitter work? Even if some large companies stepped up to support the cost of an open Twitter, one that never suffered downtime, would we migrate to it? Barring a complete failure by Twitter, the answer is an obvious no. Why? While the DataPortability folks and TechCrunch think Twitter is just a messaging service, the other 99.9 percent of us see Twitter for what it is - a social network service.
I've always had a big-tent approach when it comes to social networks; a social network is something you feel, rather than something born from a set of features. Twitter only marginally stands up to the boyd/Ellison definition: Is Twitter bound? Does a 140-character bio really count as a profile? I believe that Twitter forces us to rethink some of the assumptions around social networks. Here's how Twitter pushes thinking on the subject forward:
- Message-centricity, as opposed to Profile-centricity: The core of any social network is messaging/communication, as illustrated by Dunbar in Grooming, Gossip and the Evolution of Language. In profile-centric social networks (Myspace, Facebook), messaging has largely taken place through the profile (I talk through my profile, you respond through my wall/apps/etc.). By locating the network around the profile, we were really locating it around "communication". In Twitter, the "profile" is our communication, an always-on, interactive wall.
- Imagined Identity: In focusing on interaction and communication, Twitter has eliminated many of the social network profile elements that make people uncomfortable. In Twitter, you're not expected to list your favorite movies and upload lots of pictures (which, to many late 20's and 30's users feels like online dating). Rather, your identity is imagined - constructed through your communication and relation to your followers. For those who aren't in social network expansion, the imagined identity (as exposed to the explicit identity of dating sites/Facebook) is much more comfortable.
- Close Community: I've often described Twitter as a "close" community. While outlying bloggers have thousands of followers, most of us are followed by far less than 100 people (not including spambots). The knowledge of one's disclosure community in a social network makes interaction more personal. A close community prevents some of the context leakages of monolith social networks; of course, Twitter needs a better approach to scale close community forward.
- Constant Flux: The previous three elements - message centricity, imagined identity, and close community - interact to create a constant state of flux in Twitter. This is Twitter's killer feature. For those who use Twitter in a close network, you constantly renegotiate your friends' "profiles" throughout the day. As your concept of a "profile" is your friend's last few posts, each new post is new information. This is why you keep checking Twitter throughout the day - people you care about are updating, communicating, and sharing.
Twitter isn't a platform, it is a unique social network. It is a social network stripped to its most essential elements. Twitter provides social network designers a roadmap forward, a way of thinking about social networks more fundamentally. An "open" clone offers very little by way of competition. Further, an open clone that lacks the design or interaction aspects of Twitter would actually feel very different. Twitter is really about the user experience - something that simply can't be replicated via an open standard.
Permalink |
|
to this post
View blog reactions | Post to
11 Comments: (Post a Comment)
-
At May 05, 2008 12:34 PM,
Pascal Van Hecke said...
-
Hi Fred,
I disagree and I think your post is based on some misconceptions.
I'm not talking for Arrington, but here are my points:
- there's already an open source implementation of Twitter and it's called Jabber Pub/subscribe - that is basically the Twitter functionality (but without the SMS gateway - that would be up to the provider to cater for)
- The idea is not so much replace twitter, but have multiple Twitter/Jabber servers interoperate
- Before the Jaiku/Google takeover, there were plans to make Jaiku and Twitter interoperable that way: I attended a workshop in Amsterdam where Blaine Cook, David Recordon and Ralph Meijer (XMPP foundation) talked about standards to bridge multiple social networks over XMPP and http: http://www.mediamatic.net/artefact-26258-en.html
- Messages sent across twitter servers are _not necessarily_ public (unlike blog posts). Jabber has authentication built in (unlike RSS publishing and consuming...), so you can limit your followers list to close friends like you would do within Twitter. It _does_ require trust in the subscribing service that they _will_ keep your private message private... That trust is similar to the trust you have in other email providers when you send an email to someone with a different domain than your own!
All in all, a similar posting could have been written about sending emails between different providers when the closed messaging systems in the early nineties opened up to eachother... -
At May 05, 2008 1:29 PM,
Chris said...
-
Also disagree Fred - for all teh reasons that Pascal posted and more.
And I also I would like to clarify that I was not calling for the death of Twitter - just explaining how it could become decentralized much the same way Blogging is.
I was certainly not speaking on behalf of DataPortability on the matter either. -
At May 05, 2008 2:08 PM,
fred said...
-
What points do you disagree with? I don't disagree with the technical assessments - of course there are more established protocols Twitter could use (and probably leverage for better reliability and sanity).
My argument is that a decentralized Twitter is no longer Twitter. Twitter is not simply a messaging protocol, Twitter is a social network with its own context, norms, and place. To decentralize Twitter would radically change the service. I'm talking about social impacts here.
I've written about this previously, so I won't rehash. Users have contextual and experiential expectations in social software. Reconfiguring Twitter so that my Facebook friends could message me is detrimental in every aspect; for most users, Twitter is a curated network of close connections. You can disagree with me here, but that's my argument and I'm sticking to it.
Chris, fair enough - I definitely read a little too much into your statements. So I apologize there. What I do see here, though, is some serious stick waving in a very, very high profile place. And while I obviously disagree that is the right approach, its also likely you weren't aware of how this piece was going to turn out (I've talked to plenty of reporters in my day..). -
At May 05, 2008 2:29 PM,
Pras Sarkar said...
-
I agree with Fred here. Talks of decentralizing Twitter may make technical sense, but it makes little social sense.
What is the advantage of decentralizing? Is to make it scale (technical) or is it to make it easier for the community to use/spread (social)?
The discussions and proposed solutions so far cater to alleviating scalability issues with Twitter. Aside from that, Twitter isn't being recognized for the inherent social capital which is much more valuable as a whole rather than a sum of its parts. -
At May 05, 2008 3:35 PM,
Aditya said...
-
I will not call Twitter a social *network*, but a micro-network of sorts, since it focuses on the actual 'social' *part* of networks, and not the remaining aspects such as photos, videos, interests etc. that you mentioned. I thoroughly agree with the imagined identity. There's no better profile of you than spontaneous conversations.
That said, decentralization of Twitter will not take too long. Honestly, we already have our Tumblogs which are along the same format as Twitter. A microformat for each post, and a service to compile the data later, you have your decentralized Twitter.
But again, to make the decentralization effective, you *will* need ONE party to compile all that data in one place. Otherwise, it is no different than going to each person's weblog, and posting comments (@replies) to their posts (tweets). We have already seen what a decentralized Twitter looks like, and the centralized one became way more popular. -
At May 05, 2008 3:47 PM,
fred said...
-
Aditya and Pras, great comments. Thanks for contributing to the discussion.
-
At May 05, 2008 7:44 PM,
Pascal Van Hecke said...
-
Hello Fred,
If I read your posting as a plea against the "naive dataportability" idea that you can "transplant" a social network (= type of friends, tone of conversation, context, "atmosphere"...) from one system to another just because the protocols are interoperable, the I _do_ agree.
The point I want to make is that I think you are choosing the wrong example here.
As your posting and the comments at http://chimprawk.blogspot.com/2008/03/fixing-information-overload-in-twitter.html pointed out, there are many different usages and styles in microblogging. That usages and style depend more on the person than on the service the account is hosted on, more on the frequency, public/private distinction, audience than on which domain you'll find his archive and whether (s)he has an @twitter, @facebook or @stutzmann.com microblogging account.
Twitter, differently from other social networking systems/sites (e.g. Facebook you mentioned), imposes very little on the context people use it in.
If you choose to keep your posts private, and keep the same criteria in who you follow/interact with, then your Twitter experience will stay the same, regardless whether it is decentralised or not...
Having a decentralised microblogging system might even encourage you to use multiple personas and accounts (e.g. a private/family versus an academic one) on distinct domains...
@Prakas: the added value would be that, at least for the Geeks amongst us, you'd have more ownership of your own account, since you'd be able to selfhost it (or at least have it under your own namespace/domain). Increased ownership will increase the time and effort invested in your microblogging persona.
@Aditya: disagree. A decentralised Twitter looks more like (faster, over xmpp) email than blogging. You do not go to an @vanhecke.info page to read @vanhecke.info mails, you'll read them at your adityamukherjee.com-hosted "inbox" or dashboard. The big issue is loss of control when a message leaves the boundaries of your own system but that, as I pointed out in my previous comment, is a trust issue we take for granted with email nowadays, so why could that same level of trust not be established with microblogging? -
At May 05, 2008 7:56 PM,
pascalvanhecke said...
-
Just to further clarify my (even more) clumsy English at this late hour:
I just saw del.icio.us/meryn's description of this discussion on del.icio.us:
"An open clone that lacks the design or interaction aspects of Twitter would actually feel very different. Twitter is really about the user experience - something that simply can't be replicated via an open standard."
My point is exactly that there's no such thing as _the_ Twitter UX since people use a host of different clients other than the web interface anyway. -
At May 06, 2008 1:08 AM,
Aditya said...
-
@pascal van hecke → If you compare dT (decentralized Twitter) to a mailbox, then that means you will only get updates that are aimed at you. General updates will not reach you unless (say) you're subscribed to a person (like Google groups).
Now, using this model, how do you find more people to follow, or people who share your interests? Where is the public profile that you can see? Where is the API to get their tweets for use?
The Twitter experience is not interface based, it's content based. Content which is easily accessible, viewable and retrievable. That is why we say that a person who privatises his tweets doesn't get the point of Twitter. dT will become too spread out, in various corners of the web, which will result in the dispersing/loss of that content. -
At May 06, 2008 5:33 AM,
Pascal Van Hecke said...
-
@Aditya
Quoting and commenting from your comment:
"General updates will not reach you unless (say) you're subscribed to a person"
Yes, but that's how it works now, isn't it? You are right that you'd need a 3d party aggregators(like there are already several in the blogging world) to follow the equivalent of the "public timeline" (in as far as that is possible...)
"Now, using this model, how do you find more people to follow, or people who share your interests?"
With 3d party tracking and aggregating services (the microblogging equivalent of technorati, Google blogsearch).
"Where is the public profile that you can see?"
At SomeTwitterCloneService.com/accountname, or even better, at http://twitter.adityamukherjee.com/.
"Where is the API to get their tweets for use?"
I'd have to use the api provided http://twitter.adityamukherjee.com to get your latest tweets. And I'd have to authenticate as one of your subscribers if I you had your tweets on private. (Cfr rss and the metaweblog api in the blogging world).
"The Twitter experience is not interface based, it's content based. Content which is easily accessible, viewable and retrievable."
I agree completely. But I'm pretty sure Twitter clients like Twhirl will be able to hide the complexity of a decentralised system. The only difference would be you'd need to pass on a domain with an accountname, to subscribe to chimprawk@twitter.com or Aditya@twitter.adityamukherjee.com or twitter@adityamukherjee.com, however you want to be addressed.
"That is why we say that a person who privatises his tweets doesn't get the point of Twitter."
Like Fred at http://twitter.com/chimprawk does :-) ?
See, your conception of what Twitter is or should be, is so radically different from Fred's that I really don't think decentralising would matter...
"dT will become too spread out, in various corners of the web, which will result in the dispersing/loss of that content"
It _is_ technically harder. But that's what people thought of decentralised email (across different servers) last century as well.
It _might_ and probably will make the way people use it more diverse than it is. But Twitter usage already _is_ diverse, cfr the different ways you and Fred use it. There's no such thing as the Twitter community, there are the countless of communities every Twitterer has built around her or him with her or his followers and subscribers. And that would be similar in a decentralised Twitter... -
At May 06, 2008 12:16 PM,
fred said...
-
Ok, comments becoming a little unwieldy so instead of responding in line...
First, I'd like to know percentages of users that use Twitter via a client as opposed to the web. Yes, there are statistics that say X percent of Twitter's traffic moves through the API, but a volume-based metric will skew.
I would actually bet that there's a large middle- and long-tail sector of Twitter that use the web - basic adopters that aren't quite ready to download/install a client. I'm not sure how to get at these numbers, as public timeline samples will also skew towards heavy users.
Regardless, I disagree with the argument that there isn't a Twitter experience. My argument is that Twitter, and the cloud of applications that surround it, do constitute an experience. And within those clients, we see differing experiences. The desktop client? They use Twitter more like IM. The web updater? They tend to update less often (just observations here). These differences do constitute and evidence experience. (Pascal's diversity argument).
The question returns to location of communication. I would actually argue that as Twitter moves further away from the web and web-based clients, differing norms are established. I could see cliques emerging down the road - hyper-twittering client users vs. more conservative web users.
The point? The calculus of Twitter lies in the newsfeed; for the majority of users, that stream is special, curated. One hyper-twitterer in a newsfeed of "normal" twitterers can dominate and devalue the stream. A balanced feed is essential. Oddly, with diversity and universal access comes imbalance (see our inboxes). Twitter loses the form that we've imagined into existence.
