Computer says no.

Editing of Tweets is a bad idea. Here’s why.

Twitter is the always-on, always-live firehose. There’s a ton of practical and technical reasons why enabling editing is a bad idea.

Twitter doesn’t currently enable its users to edit tweets after you’ve hit the ‘submit’ button. A quick search on Twitter confirms that it’s a hot topic, and I wouldn’t be surprised if this was one of the most frequently requested features.

But it’s a terrible idea.

The main problem? Re-publishing

Twitter is a platform of immediacy: As soon as you hit ‘Tweet’, all your followers see what you just wrote, within seconds. That sets Twitter apart from every other social media. On Facebook, for example — which does allow* editing — posts are shown to your followers with a bit of a delay (anything from a few seconds to an hour). But also — Facebook is a tool for communicating with friends, family, and people you trust. Editing a Facebook post, then, has a very different impact than editing a Tweet.

Technically — as in, measured in lines of code that would need to be changed — the editing of tweets would be trivial (albeit with some nasty edge cases; I’ll discuss a couple of them slightly later in this post).

From a policy point of view, however, it would be really, really complicated.

The concept of instantly being able to re-publish content is a powerful feature of Twitter, and while opinions vary on whether a retweet constitutes an endorsement of a sentiment or merely an amplification, an idea can be retweeted within seconds of being shared. This makes the concept of editing a tweet complicated, because it is very difficult to know what the edit actually means.

What are the problems?

If you start thinking a little bit about what the implications of edited tweets are, you open up a whole series of questions, all of which have two or more perfectly sensible solutions, but each solution bring their own challenges.

To wit:

Retweets and Favourites

When there’s an edit, should the post keep its retweets and favourites, or lose them?

What if British Airways retweeted a tweet that used to say “British Airways are so awesome”, but now says “I advise everyone to smuggle bombs onto airplanes”?

Keeping them makes sense, maybe — but what if the content of the tweet is now materially different? What if British Airways retweeted a tweet that used to say “British Airways are so awesome”, but now says “I advise everyone to smuggle bombs onto airplanes”?

If this were possible, then that would basically mean that nobody would retweet anything ever again — the risk for trolling and abuse would be too high. This point alone should be the #1 reason for not allowing editing of tweets — it would be the death of the whole Twitter platform.

Stripping all retweets and favourites from a tweet also makes sense — it certainly takes care of the problem mentioned above — but at that point, if you lose all the social amplification of your post at the point of an edit, is there even a point to editing? It would be easier to copy the original tweet, then delete the old one (or just leave it), fix the typo, and re-post. If that sounds familiar, it’s because that’s how we do things now.

Versioning

When there’s an edit, should it be possible to see what has changed from the previous version?

Some would argue no — you edited the post, right, so why would the original stay around?

Others would argue that yes, to ensure transparency, keeping the old tweets accessible somehow would be a good idea. That in itself brings up a lot of user experience challenges — How would a user get to this version history? How many edits would there potentially be for each tweet? How would an API user / external app deal with versioning? How does Twitter expire cached content?

It would also be possible to just make the old tweet disappear, but you’d probably need to keep a record of the old tweets for law enforcement / policy breach reasons, and so it would almost certainly require Twitter to keep copies of edited tweets anyway, so even if this isn’t exposed to you as a user, Twitter would still have to find a solution.

Cached Content and non-synchronicity

A service as intensely used as Twitter lives and dies by its caching and its ability to keep content current. This is already one of the more complicated things about Twitter as it is, but by adding versioning and editing, you’re adding an order of magnitude worth of complication.

Caching is done at many different levels — your browser may cache tweets, your client might do the same, and Twitter itself caches things in many different ways in its own back-ends, both for public consumption, for the API endpoints, the firehose, and so on.

More worryingly, a lot of the systems in use by power-users (such as British Airways, as used in an example earlier), cache the tweets.

Finding mechanisms of clearing / resetting caches downstream (to clients, tools, bots, and API endpoints) is an incredibly complicated process.

For example:

1. What if you load your browser window, you see a tweet you like, and you hit ‘retweet’, but between the browser loading the tweet and you hitting ‘retweet’, the author of the tweet edited it? What should happen?
2. What if British Airways has a tool showing all the tweets tweeted at them, and replies to one of them, but the original tweet has been edited, so the reply no longer makes sense or — even worse — makes BA look bad?
3. What if I use a tool like Buffer to schedule a particular tweet for re-tweeting at a later date, but the original tweet is edited between me scheduling it, and the date it was scheduled for? What should happen?

Of course, you can try to design around all of these issues, but it adds vast layers on complexity on top of Twitter’s platform.

Notifications

Should a user be notified when there’s an edit of a tweet in your timeline? How would this be done? Should the edited tweet be re-inserted into the timeline at the time of the edit?

You could argue either way, but if it is re-highlighted, you could see this being abused for marketing reasons (at best) or by spammers (at worst).

I could ‘predict’ the result of the Superbowl today, for example, then go back and edit it to show the correct result. It would certainly cause a lot of confusion further down the line — how can that be prevented?

Timestamps

Come to think of it, should the edited tweet get a new timestamp?

If it doesn’t, then how would an edited tweet inserted into the timeline be differentiated from ‘live’ tweets? That would get confusing because suddenly tweets are no longer chronological…

But if edited tweets do get a new timestamp, we’re back to the original issue; you may as well just delete and re-post the tweet.

Extent of an edit

You might argue that you could reduce the abuse issues by reducing the extent of the edits to a certain number of characters to prevent abuse — but that wouldn’t work: A single character can make a huge difference in the meaning of a sentence (‘Let’s eat, Grandma’ versus ‘Let’s eat Grandma’), and a complete rewriting of a post could be done without any meaningful change in the underlying concepts (‘Let’s eat, Grandma’, versus ‘Grandma, it’s time for dinner’).

In conclusion: Editing creates more problems than it solves.

One of the beautiful things about Twitter is its brevity. Tweetstorms and photos-of-text aside, being able to share only 140 characters at the time means it’s easy to skim-read large quantities of tweets from a relatively large number of people you follow.

We’ve all been guilty of posting tweets that make no sense because autocorrect gobbled up any meaning, typing faster than you can think, hitting publish by accident, or because pudgy fingers managed to mangle the words, but the solution isn’t to enable people to edit an individual tweet.

What we have currently does trip people up on a regular basis, and yes, it’s annoying and embarrassing to have no way of editing a tweet after it’s been sent… But think of it this way: At least it’s not like SMS messages or e-mail, where you can’t edit and can’t recall (delete) the message.

Seen from Twitter’s point of view, I completely understand why they haven’t implemented editing of tweets yet. Sure, it would be lovely to be able to edit a tweet — but it creates a lot more problems than it solves. And realistically, if you do spot a typo, it’s a lot easier to delete and repost. It’s only a 140 characters, after all; you won’t wear out your thumbs re-typing it…