I have been working on a new startup called Edglings (see www.edglings.com). It’s a product company, building applications that dovetail with Twitter, trying to make the experience of Twitter richer, deeper, and more social.
How do I and my partners intend to do that?
First of all, by exploiting the microstructure that is starting to spontaneously emerge in Twitter.
Innovations like the use of ’@username’ have moved from convention to core element of the operational Twitter platform. Other innovations, like tags (’#foodporn’) and retweeting (‘RT’) are still floating around as general conventions that are not specifically embodied in the Twitter machinery at a deep level.
I anticipate a large number of other conventions will emerge to create more structure where there is little, or at least not enough. For example, the Cotweet folks – who are building a solution for businesses using Twitter – have suggested the use of the caret (’^’) as a way to indicate the identity of an employee twittering on behalf of a company via a corporate Twitter account, like this:
We are looking forward to the open house this weekend ^EB
I have suggested one large-ish bit of microstructure, the subtag. This grew from the failure of Twitter tags (or ‘hashtags’) at the recent South By Southwest event, where so many ’#sxsw’ tags were in Twitter that searching for that tag didn’t perceptibly narrow the stream of tweets.
Subtags (perhaps better thought of as tag intersections or tag relations) is the idea of making the tag more distinctive. For example, at the recent Web 2.0 Expo, I could have used different subtags for different sorts of topics related to Web 2.0 Expo:
- #w2e – for the conference as a whole
- #w2e.oe09 – for the session on Open Enterprise 2009
- #w2e.kathysierra – for the Kathy Sierra session
- #w2e.parties – for the social side of things
You get the idea. But for subtags to catch on, at least outside of a small geeky minority, tools will have to support them. And you can be sure that one of the projects going on at Edglings is pursuing that, and related topics. Subtags are sort of supported by Twitter, already, since searches for them work as you’d expect. However, there is no repository tool that supports them at this time.
One observation: this sort of tinkering with microstructure is a stepwise refinement, moving from an existing way of doing things to a slightly different, possibly better way of doing things.
Occasionally, however, like the development of the idea of Twitter tags (by Chris Messina), a real jump forward has to happen. When Chris and a few others (including yours truly) started using Twitter tags, there wasn’t any support aside from search for them. [Note that Chris’ original idea for tags in Twitter was something significantly different than what has emerged, which says something about community-based innovation like this, too.]
I see that Robert Scoble is kicking around the idea of a huge, search-based system for attaching metadata to a new, alternative microstreaming service like Twitter that he suggests the flailing newspaper industry should build, coming up with something like Friendfeed, but incorporating capabilities meant to replace classified ads, and pull away that revenue from services like Craig’s List.
It’s a strange post, one that balances one improbable idea on top of another and then another and so on. But the point of my comments here are to suggest that this sort of large-scale, all-at-once, completely-designed-from-the-top-down sort of innovation doesn’t happen on the Web very much. Ideas like blogging, wikis, instant messaging, or ecommerce started from small ideas that worked in the small, and then evolved, in a messy chaotic way, into larger systems that work in the large.
Scoble seems to want it the other way, at least for the sake of the newspapers:
Here’s what I meant by “invisible comments.” Look at friendfeed. Each comment is a piece of metadata that can be used by the search engine. For instance, I can put “greattoasteroven” into a comment, like I did underneath Dave Winer’s toaster oven post, and now you can use friendfeed’s search engine to find it. But why does that “metatag” have to be visible? Here, let’s design an invisible commenting engine.
First, let’s invoke the invisible commenting engine. We could pick something like %%0, which is a string that would never get used. So, let’s design an invisble comment:
%%0 metatag=”greattoasteroven” This one could do metatags.
%%0 location=”25 Pinehurst Lane, Half Moon Bay, CA 94019″ This one could do location.
%%0 price=”$1,499″ This one could do price.
%%0 language=”german” this could tell the search engine to pull this entry up only if the speaker wanted german results
%%0 invisible=”Dave Winer is a nice guy.” This could be used to leave a comment for you, or for the search engine, but that wouldn’t be visible for the outside world. You could even use that to make things only visible to certain people, so you could have private conversations INSIDE the comment thread.
%%0 administrator=davew This one could make Dave Winer an administrator of the item, which would give his comments a different color, and would enable him to delete or edit comments on that thread.
%%0 post comment at 12:01 p.m. Pacific Time April 29, 2009: “this comment will appear on April 29th at 12:01″
%%0 credibility=10:davew This would mark Dave Winer as “highly credible” on this topic, and because everyone else has a default credibility of 5, would give him a different color and a star icon. You could then do the next invisible comment:
%%0 display credibility of >8 only (this would make only comments of people with credibility of 8 or above visible, once the user had invoked this).
Anyway, I could keep going all night long. In the real-time web having such a console would be important to be able to talk to the search engine. We could brainstorm such things all night long.
Now, isn’t that “geeky?” Yes. Would the celebrities that now are moving into Twitter get it? No AND yes!
See, while the geeks might be stuck with such a command-line interface, I expect that very quickly developers of applications like Tweetie, Twhirl, and TweetDeck would add on UIs to take advantage of the invisible functionality. You could click a “price” button in TweetDeck, for instance, which would show you the price of the bike you are now looking at (or allow you to input it as a Tweet-like message, if you knew the price).
This would build a very rich microblogging service and I think everyone else would want to build invisible comments into their system and interroperate with the ones we designed.
My bet is exactly the opposite.
- It’s too late to build a different, private service. Twitter is the dominant platform in the rapidly expanding microstreaming market.
- Media players – like newspapers – will view microstreaming as ‘just another channel’ and as a result will not do much innovation in the new medium.
- Microstructure will emerge one tiny fragment at a time, in a messy way. Subtags may catch on or not, but in either case they won’t rely on a grand meta-schema of which they are one instance, like Robert’s ‘invisible comments’ formulation. Much more like that we will see tiny conventions appear: like ’@’, ‘RT’, and ’#’.
Microstructure is big, because its adoption doesn’t require any major investment of time or effort, and even the first applications to support microstucture can be small and simple. Microstructure works because people will be able to understand it at first glance, without an operating manual.
So, as we, at Edglings, start to roll out our new Twitter appliances the first indications will be almost below the level of awareness, because it will be only one babystep away from what you are already doing and seeing.