Twitter replies are broken, and Twitter has a great plan to fix them

Anytime a major social media platform makes even minor changes, it generates consternation. So we can expect a lot of griping in the coming hours about Twitter’s announcement that it’s going to change how it deals with replies and when usernames count against the 140-character limit.

And Twitter didn’t do itself any favors by writing a confusing blog post on the subject. I misunderstood Twitter’s plans the first time I read the post. Several of my Vox colleagues did too, and their initial reactions weren’t positive.

But now that I’ve read the post more carefully — and also read the less confusing version of the announcement Twitter wrote for developers — I’m convinced that the vast majority of Twitter users will like the new system for handling replies. In fact, the change is long overdue.

Twitter’s handling of replies might be the platform’s most confusing feature. Right now it works like this: If you start a tweet with a username, the tweet will only show up in the timelines of people who follow both you and the other person you’re mentioning.

This seems natural for those of us who have been using Twitter for years. But I’m constantly seeing newer and less prolific users do this wrong, starting a tweet with a username and not realizing most of their followers won’t see it. It’s one of the biggest ways in which the platform seems hostile to new users.

And the reason Twitter works this way is essentially a historical accident. In the beginning, Twitter didn’t have a concept of replies. Every tweet showed up in the timelines of everyone you followed.

But over time, a social convention developed that starting a tweet with a username signaled that you were replying to someone. This created clutter in people’s timelines — following only one person in a conversation was like listening to one side of a telephone call. So in 2009 Twitter tried to help out users by hiding tweets from your timeline if they started with a username you weren’t following.

Even after Twitter did this, however, the conversational experience was still far from ideal. The problem was that as you followed more people, multiple conversations got jumbled up in your timeline and it became harder to follow any one discussion.

So Twitter made replies a native feature of tweets. It started keeping track of which tweets were replies to which other tweets, allowing the platform to start showing users entire conversation threads instead of just individual tweets.

So for the past three years, Twitter has used two different systems for identifying conversations. It still hides tweets that start with a username you don’t follow. But whether you start a tweet with a username is actually irrelevant for Twitter’s threaded conversation feature — Twitter keeps track of that information directly when you click the “reply” button on a tweet.

Twitter realized that this first method for identifying conversations had become anachronistic. If the platform is already directly tracking which tweets are replies to which other tweets, it no longer needs to rely on the earlier hack of looking at whether a tweet starts with a username.

And so this week Twitter will announce that it’s going to continue providing users with the same useful feature — only showing conversations when you follow both people involved — without forcing users to worry about whether a tweet starts with someone’s username.

This will mean the end of one of Twitter’s strangest conventions: putting a period before someone’s username to make a tweet visible to everyone. In the future, you’ll be able to just start a tweet with someone’s username — “@ezraklein is the founder of @voxdotcom” — without having to put a period at the beginning to signal to Twitter that you want everyone to see it. And if you want everyone to see a tweet that’s a reply to another tweet, you can do this by clicking the retweet button, a more natural strategy than putting a period at the beginning.

Once you start thinking about replies as a native feature of a tweet, you realize that it’s not necessary for reply tweets to include people’s usernames at all. If Twitter knows which tweet your tweet was replying to, then it’s easy to also figure out whom the tweet is replying to. So rather than forcing you to include the other person’s username in the tweet (eating up some of your scarce 140 characters), Twitter is going to start treating that as a separate bit of metadata, like the tweet’s date and location, and display it outside of the tweet.

But note that this is only true of people you’re replying to, not mentions in general. If you write spammy tweets that mention random people, those extra usernames are still going to count against your character count — preventing people from overusing mentions.

One big question that doesn’t seem to be answered by Twitter’s announcement is how the company will deal with “Twitter canoes” — situations where two or more people hijack one of your tweets to start a long Twitter argument, flooding your mentions tab with tweets you don’t care about.

Right now this kind of thing is constrained by the fact that usernames are counted against character counts, so people have an incentive to remove nonessential people from the thread to give them more characters to argue with. But in the future, you won’t have to list the people you’re replying to, creating the possibility that dozens of people (Twitter documentation suggests the number could be capped at 50) could get dragged into long-running arguments.

But this shouldn’t be a terribly difficult problem to solve. Perhaps Twitter will include a thread-specific mute button that lets you bail out of conversations that have become too tedious. Or maybe Twitter will come up with smarter ways to figure out which conversations you’re likely to actually care about.

Either way, cluttering up tweets with the names of the person you’re replying to was always kind of a hack, and most users will be happy to see it go away.