Some months ago, Twitter fixed, or broke, their @replies feature. They made it so that all the ‘@username’ messages you sent to your Twitter stream were only visible to people who also followed ‘http://twitter.com/username’.

After initial irritation and (ongoing) fruitless protest, the solution has to be accepting Twitter’s decision and instead work out a new pattern for the use-case they’ve left unfulfilled.

The use case is this: The act of responding to a tweet (perhaps, but not exclusively conversationally) whilst still posting content of standalone relevance to your public feed.

The syntax I have adopted, and advocate for this purpose is just a simple enhancement of the existing @ mechanism: @@.

It works because prefixing a @username with any other character breaks out of the existing replies mechanism, and therefore keeps the post in regular public streams. I’m advocating @@ because it’s the most obvious evolution of the existing form. More obvious than -@ or _@ or !@ or (worst of all)  @, which are also being used at time of writing. I posit that @@ is clearer as it does not introduce anything new, no-one has to ask “What does * mean?”, it’s like emphasis. Furthermore, repeating the same character is easier and faster to type.

“But I don’t want to see your @@replies!”, will say some people. They’re missing the point. This emerging practice is about the author indicating that a tweet—despite being a reply—absolutely is a 1st class post in their stream. Their stream, that they alone curate. If they explicitly indicate that a reply should be public, and you don’t like it, then it’s up to you to unfollow them just the same as you would if they posted dull or irrelevant posts of any other kind.

There’s a supplementary, important benefit as well. By embracing a new convention to explicitly promote some replies, users should embrace both ‘kinds’ of reply. You’re unlikely to want all your replies public, so you can ‘whisper’ replies back and forth using the existing single-@ syntax too. Prefixing all replies as a protest is pointless, so embrace the new way to control how your replies are published.

Existing conversation trackers don’t work with this. The @@ replies get left out of threading tools. That’s a shame, but that’s the reality of new behaviour.Existing conversation trackers depend on the in_reply_to header, which in some third party apps (such as Twitterrific on Mac OSX), does not get pushed to Twitter unless you explicitly include @user at the start of a message. In effect, @@ breaks this, as does every other prefix, or placing a @mention elsewhere in the Tweet. The Twitter documentation states that the username need only be included somewhere It takes time for any syntax to be recognised and supported properly, that as much as anything is why I’ve documented @@ here, as a reference and explanation for you to use as you advocate the practice, and to encourage everyone to settle on the @@ syntax in preference to the many that currently circulate.

Further background follows:

Twitter has a tradition of evolving from conventions. The ‘@username’ replies feature became a native part of the service after users developed the syntax for themselves (and after some external tools added behaviour to it; hyperlinked names, and so forth). It later become ‘@mentions’, after users extended it to prefix all references to fellow users. Similarly, the ‘#hashtag’ syntax was introduced and was natively adopted in the same way.

Some months ago, Twitter changed (broke) the way some people use replies. No longer could you see replies your followees made to other users who you were not also following. The reason behind this is that a large proportion of Twitter users write ‘@friend LOL mate!!1’-type replies. A lot. If that kind of user is in your twitter feed, seeing their inane replies with no original context is irritating at best. Twitter changed the service to support that use case, rather than support both use cases (as they previously had a user preference to toggle between; an implementation that was processor intensive on Twitter’s side.)

The expense was twofold: First, users who post valuable, useful content within their replies. Some are engaged in a discussion that’s relevant to their public Twitter stream, thus relevant to their followers. In less cynical terms, they simply try to author replies so that each one is also a standalone, useful piece of content. The philosophy is that whilst you’re replying to something, you’re also posting something. If anything, it was an embrace of Twitter’s limitation; precisely because your replies were visible to all, there’s a reason to write worthwhile replies.

Second, because those richer, fuller replies piqued the interest of people outside the conversation, replies were a viral means of discovering new people, and interesting new threads of conversation. The cost of ignoring a reply that doesn’t interest you is extremely low (just as ignoring any other message), so the system worked well. Twitter claimed at the time that new ways to discover people would be introduced, but no news is forthcoming.

Evolution of new syntax is a pragmatic, logical response. The simplification of original @replies left this open use case, and using explicit @@ syntax means that authors are in control of what they promote into their public feed.