Twitter will soon be rolling out ‘Annotations,’ which will allow developers to add metadata to any tweet that they want.
Twitter asked itself what type of metadata it could add to the service; past location and so forth, but has instead decided to let developers make that decision. The company has controlled how tweets can be tagged with metadata, due to developer man-power concerns, scaling worries, and other road blocks.
From next Quarter, Twitter will let you attach metadata to tweets, and then pull that data back out of Twitter later. This is going to allow for an explosion of uses for Twitter. This is absolutely huge.
Given how gigantic Twitter now is, with over 100 million users on the service, what everyone needs is better information to find the tweets that they need. More metadata, better sorting. Or as Twitter said it, the power of metadata is relevance.
Sadly, we have to wait for this to come out, but it is going to make different Twitter applications even more distinct than ever before. Also, expect to see wilder mashups than you have ever imagined before.
It has been pointed out that this has actually been attempted before, but not by Twitter. Back in December of 2009 Aral Balkan worked on ‘Twitterformats,’ which sound a lot like what Twitter is discussing today. A Twitterformat was a “machine-parseable, lightweight client-side API that extend[ed] Twitter.” Sure, Twitter may not have thought of it first, but that doesn’t meant that we are not glad to see it.
My initial reaction: this will kill hashtags.
100 million users?! Seriously, it's about 10 million active users, if that.
And well done Twitter for finally significantly upgrading the tweet specification – it's only taken 4 years!
However, without any standards for the annotations, expect to see a nightmare for developers. Remember it took months or even years for things like @username notation and #hashtags to become established throughout all the apps. Twitter's going to get even messier…
It'd be pretty nice if it did! I occasionally use hashtags, but I don't have to like them.
I don't quite get it. What kind of data would be included?
What problem does this solve that hashtags don't? I mean, obviously hashtags gobble up part of the 140 character limit. But, I don't see what this solves that hashtags don't.
What I'm most exited about is the potential to add links as metadata and finally rid Twitter of aweful short urls.
See also blogs.fluidinfo.com/fluidDB/2009/12/01/putting-metadata-onto-tweets-with-fluiddb/
The tags will be more formalised, assuming standards come to be, which won't happen overnight.
The funniest thing about this is that the annotations will become far longer than the tweets themselves, which ruins the so-called plus point of the 140 character limit, and show's how the tweet has always been badly designed.
But apps can choose what metadata to support. For instance, some will support languages (declared by the user or automatically detected), links but not “mood” or “listening to”.
It'll be fun seeing how they actually implement this, how apps will embrace it and if & how they'll limit this.
Yes, but there won't be any standards for the data, so every client will have to be constantly updated to try to correctly interpret whatever is being 'annotated', which makes hell for the developers and hell for the users all trying to keep up.
The way around this is to centralise everything through standardised APIs, á la Facebook, which funnily enough has 40 times as many active users and is growing far faster.
Has that been announced? No standards? No way to specify “this is a tag” or “this is the language”? Just because two tweets have metadata in “tag” doesn't mean they both mean the same. That's what semantic web has been using URIs to differentiate meanings… They should do something similar.
That won't solve the “apps need to catch up all the time”… but apps can be written to allow filtering without knowing what “lang” means beforehand. They'll figure it out somehow.
As for facebook, well, it's an option. And remember, orkut is huge in Brazil. So it really depends on your target audience, right? Rule of thumb: cover the most ground. Twitter *and* facebook. ;)
It doesn't appear to have been announced, no. It needs standardised fields, like in an e-mail header – that's why e-mail works across millions or even billions of e-mail clients and servers across the world… and is why techie got excited over the potential of Wave and its protocols because it's messaging done right, not Twitter's messaging done wrong.
Sure, apps can let you filter generically, but it's understanding what the data represents that is key. Knowing that 'lang', 'language', 'ln' or any other way that might be used all mean that the data is describing the language that the tweet author has said the language of the main message is in. I don't want to have to use hundreds or thousands of filters when just a few ten or twenty standard ones would do.
Orkut is huge in Brazil, but as far as I know it's not got the same structured platform API that Facebook has. And RE: your rule, I doubt there is anyone (of significance) on Twitter who is not already on Facebook (whether active or not). I'm a developer getting lots of corporate/enterprise level Facebook app work at the moment. There's been some Twitter stuff too but there's so little you can currently do with it beyond a filtered search of tweets, which annotations will only enhance but not evolve.
From Ryan Sarver's announcement it looked like annotations are limited to clients adding metadata to tweets they publish: http://bit.ly/aAoNsW
I agree. Death to hashtags.
This is going to be insane. My only worry is that each developer is going to have their own way of describing something that may be similar, but not –quite– the same as another developer.
Maybe, but if a developers use ontologies then it wont be so bad.
At least that is how I see all this – a big potential for Linked Data and Semantic Web!