- September
Posted By : Adrian Chan
Structured Tweeting?

Adina Levin of Socialtext posted recently about Tags for ActivityStrea.ms. I’ve been enjoying online conversations with Adina quite a lot of late; there’s a constructive Venn overlap between our approaches to design for social media and social interactions.

Given the amount of time we all spend skimming through Facebook status updates, twitter, and blog posts and comments, the idea of tagged activity streams has a strong appeal for me. But without going into detail into the proposed architecture for activity tagging (which I would like to do separately), I want to just conjecture for a moment on the question this idea begs. To wit, structured conversation.

A while back, Stowe Boyd and Chris Messina launched a related effort around microsyntax. I don’t know where it currently stands, but the idea — to codify and back inline (in the tweet) syntax that would permit tweet parsing for some common types of tweets made a lot of sense to me. Besides location, consistent syntax could be used to identify any number of things, from reviews and recommendations to requests, offers, invitations, and so on.

Where microsyntax uses structure embedded in the tweet, activity tagging would place rely on structure outside the message (as I understand it). External structure has some advantages, not the least of which is a certain robustness (microsyntax depends on the user’s written compliance with syntax).

However, and this is what I wanted to conjecture about, both are signs of a need for structure. Or if not need, then at least a nascent interest in structure. Years back, the structured blogging effort made a go at wrapping classification around the blog post format. It was abandoned for several reasons. Its taxonomies were probably excessively detailed. Its success would have required participation by search engines and blog indexing services. But most of all, it required a lot of extra work on the part of the user. The structured blog post was not a replacement for the standard post (subject, body, datestamp, tags) but a set of supplemental formats that could capture review, music, product, and other types of posts. (Fields were added for product name, manufacturer, rating, etc).

One can easily imagine the potential benefits of structured activity updates, as well as tweets, and other status updates. In fact one can imagine structure scaffolded around individual posts rich and connected enough to provide a back door into social networking and profile-based groups and communities. Theoretically, there’s a slippery slope from structured conversation to the navigation and page-based organization we enjoy on the web today. In other words, messages could form the basis of browsing and finding just as pages do today. In theory.

But in theory, the idea is quite compelling. If “information overload” stands today as the single-most unwanted byproduct of the conversational turn in social media, then structure could help to solve some of its problems. One might imagine a set of codified values and attributes that authors and readers might use on messages. Messages could be classified by linguistic expression (or linguistic action), such as request, question, answer, comment, invitation, offer, forward, and so on. Twitter’s codification of replies is one such example. Structured updates might then also be sucked into sites that organize them by their attributes, thus enabling group and community-like activity. Updates and tweets might ultimately form the basis of a kind of “message board” system (sound familiar?) built around dynamic aggregation of individual messages. Pages would not be content containers, but would be “written’ by aggregation of distributed content.

None of this seems likely, however, for several reasons. Search engines and readers would have to participate around shared standards. And users would have to make the effort to classify their updates. Neither seem very realistic. But it’s a compelling idea!


  • I think to codify ideas to avoid 'information overload' would be helpful! I get tired of sifting through so much useless information.


  • Certainly would be helpful to the consumer! Would you be willing to add meta data to your tweets when sending them tho? That's the stickler 😉

  • Very true, it is extra work!

  • I agree this is probably unlikely to happen for human-generated tweets. But what about for machine-generated tweets (such as when Foursquare broadcasts my location)?

  • That would be more do-able. Codification of system messages is easier. Opt-in codification by users of their own posts, which are conversational, though, would be so great… and an improvement to the social mess that is the realtime web of social flows and updates!

    System messages would be a pull, I presume, w/ users setting filters on their subscriptions, and these system messages would then function as notifications. User opt-in meta data would be a push. Together they would provide a lot of additional utility to users interested in telling and following social news.

  • I bet some percentage of users WOULD be willing to tag their Tweets simply because doing so increases the likeliness that your Tweet and related content would be seen by a larger audience .

    Isn't that what happened with Flickr? Delicious etc. The individual does something for themselves that helps everyone else.


  • I agree — simplicity in the UI, and clarity in the tagging options are key, but there are good folks working on this at http://activitystrea.ms/ . Some structure is inevitable.

    Whether it's just declarative (eg, tagging object and activity) or whether it might also involve expressions (i”m asking a question, making a request, extending an invitation), and whether there's a a way of classifying what response the user wants (no reply needed, please forward, please comment?) would shape the possibilities for extending tagging into real conversation structure.

Leave a Reply