This week, many people have been given beta access to Twitter’s new Retweet feature. Unfortunately, rather than seizing the opportunity to pave the cowpaths by building a feature that reflects the way users are currently retweeting each other, Twitter have launched something which behaves quite differently.
You have to change your retweet behavior to use the new feature. This has angered many users, myself included, so I’d like to explain how I think the new feature should have been designed. To start with I’ll look at where retweeting came from, I’ll then explain some of the problems with the way it works currently, how Twitter are trying to address these problems with the new feature, and finally how I think the problems could be better addressed.
Why do we need a retweet feature anyway?
Up to now, retweeting has been done purely through syntax – by editing the original message and adding notations like RT @alexbfree or (via @acroll). Often people will add a comment too, as seen in this example:
There are a number of problems with this current approach, most of which are explained by Twitter founder Evan Williams in his explanation of the new feature. Let’s look at each in turn:
1. Attribution confusion
From my tweet, you can’t tell that this was originally a tweet by @CBCMontreal, to which @zoonini added “This is troubling”. It might appear that @zoonini was the original poster.
This is because I had to remove her “RT @CBCMontreal” to make room for my “RT @zoonini” and my own comment.
Evan believes the solution is to show the originator’s avatar in place of the retweeter, and credit the retweeter with a tiny link at the bottom:
Ultimately, this is a value judgement about whether as a retweeter you want to credit the person who passed on the information or the person who first posted it. Personally, I want to credit both, but given a choice I err towards crediting the person I know who passed the link on. I’m sure this judgement will be different for everyone, and the new system doesn’t allow for this.
2. Identity protection
There’s another issue with attribution, that anybody can claim to be retweeting you by writing RT @yourname before any kind of message, and then it looked like you said that thing, which could be something you disapprove of, such as advertising spam. There is a clear need for retweets to be attributable and for it to be harder for people to misquote others, intentionally or otherwise.
Here we see a loophole in the design of the new feature – since we can still retweet “the old way”, there is nothing to stop this behaviour continuing. If Twitter had got the redesign right, they would have the confidence to block the use of RT @name syntax in messages. I think they know that people won’t use the new feature, so they had to leave this option open – which can be used for good or bad purposes.
3. Messy
There’s no denying that my example tweet above is messy and hard to read. RT syntax is not intuitive (as I was reminded recently when explaining it to a new Twitter user). I wholeheartedly agree that there is a need to take the metadata about this being a retweet and from whom, out of the message. It’s a no-brainer that having a retweet button you can click to do it for you would also enable this metadata to be captured.
4. Untrackable
At the moment, you can’t follow the history of retweet back to its source – you can’t see how many people are retweeting a particular message. Why is this important? Because we are moving towards a social, real-time web. In the same way Google revolutionized search by looking at who was linking to a particular site (PageRank) and looking at the authority of that site, the next challenge is to build a PageRank equivalent for the social web.
Companies like Twitter need to determine what is most popular based on how many people are talking about it, and more importantly who are the people who are the most important generators of content. Looking at how many people retweet a message, and whose messages are the most retweeted, give a direct measurement for both of these. These are needed to make real-time social search a viable option.
Alistair Croll has been taking a look in more depth at the idea of a PageRank for humans over on Watching Websites.
5. Noise and Redundancy
Evan goes on to mention Noise and Redundancy as two other issues with current retweeting – that you see repeat too many retweets or too many copies of the same retweet. I don’t think these are major issues – if a tweet shows up more than once, that means more than once person thought it had value, so it is right that it should get more visibility. A “mark as read” option would be more useful than grouping all occurrences of the tweet. As for people who retweet too much, I think the solution is simple – unfollow them!
This is the point where Evan stops, and offers the new feature as the best way to combat these six issues. I think there is one other key issue he has failed to address though:
6. Freedom to comment
Possibly the most frustrating thing about retweeting today is having to tweak and edit the message so that you can fit in the RT credit and more importantly your own comment. The comment is the key part of the retweet, it’s what makes the content tailored and relevant to your audience. It’s hard at the moment – this is often done with a “<” or “<==” at the end of the tweet, followed by a comment. But there are often few characters to play with. This is completely overlooked by the new retweet feature. In the new feature, you cannot add your own commentary at all:
Clicking Yes posts the retweet. A retweet becomes a verbatim quote, with no personalization or commentary applied.
Changing the nature of retweets
And this is the most controversial thing about the new features. It changes the nature of a retweet into the granting of permission to another person to send their message out to all of your followers as well. You might even call it a new form of spamming.
In any medium, the message should be tailored to the audience. On Twitter, every person has a different audience. Untailored retweets are undesirable because they take away the context of the retweet. You don’t get to see why the person you are following retweeted – and this may be different for each retweeter. Some may retweet things they are impressed by, others may want to retweet things they want to ridicule. This is where grouping all retweeters together breaks down. The new design fails to accomodate the variety of tweeters and their audiences and makes Twitter more like an advertising platform than a forum for conversations and diverse opinions.
At the very least, it’s reducing a retweet to a version of Facebook’s “Like” feature rather than an opportunity to comment and add context to the content.
An alternative solution
This brings me to my proposed solution, which I believe solves all the problems mentioned above, while also preserving our right to make a retweet our own with commenting.
I propose that instead of retweet being a one click operation, it should include 2 steps:
Step 1: Click retweet on the tweet you want to share
This is exactly the same as in the new feature.
Step 2: Add your comment
The key difference I am proposing is that clicking Yes would then give you the opportunity to add a comment – a special kind of tweet which references an existing tweet:
In effect, the link to the existing tweet is additional context, much like the “in reply to” links today.
Viewing the retweet
The final difference would be how the retweeted message is displayed. The comment of the person you are following should the primary thing visible, with the original message shown underneath in context. (clicking “retweeting” would show/collapse the retweet history)
This would allow content-centric conversations to happen on Twitter, much like they do currently on Facebook wall items, forum threads or Flickr photos.
Does it solve the problems?
I believe so, and it does so in a way that preserves current user behavior. Let’s look at each problem in turn:
Attribution – the originator of the tweet and the originator of the comment are separated and clearly identified, and credit is given equally to originator and retweeter.
Identity protection – this model would allow old style RTs to be blocked, eliminating the possibility of false attribution.
Messy – the original content, the retweeter, the originator, and the comment are all separate and distinct
Untrackable – this method allows just the same sort of tracking of the spread of the message and the influence of different tweeters
Noise – I think this will make Twitter streams much easier to digest, because every message will have a context you can dig into if you wish
Redundancy – Much like Gmail’s conversation threads reorder mails to show history, the RT threads would group related comments on the same retweet together, avoiding redundancy while still preserving each retweeter’s right to add their own perspective.
Freedom to comment – The retweeter can not only add a comment to the retweet, they now have the full 140 characters available to do so.
I would be interested to know why Twitter did not come up with a design like this, which seems a fairly obvious all-round win. Did they really not think of this or not notice how people use retweets? Or do they want to change Twitter into more of an advertising network? What do you think? Add you comments below.
Photo credit: birds on wire image by TarikB on Flickr.
Social comments and analytics for this post…
This post was mentioned on Twitter by zoonini: A thoughtful rethinking of retweeting by @alexbfree – http://is.gd/4Z5ws…
I totally agree. I think that your suggestion would totally solve a lot of the current ”problems” with retweets.
I hope Twitter reads this, and implements this.
[…] independent of this, Alex Bowyer over on Bitcurrent wrote a thoughtful piece on how Twitter should have formalized Retweeting, and some of the issues with the current model. Unfortunately, there’s some strangeness going […]
[…] encourage you to check out this post by Alex Bowyer on the Bitcurrent blog speaking to a better design proposition for Twitter’s […]
Interestingly, you can now get a greasemonkey script which hacks the avatar display to show both the original poster and the retweeter:
http://bit.ly/5AxJ9F
This seems to support the idea that what Twitter have modeled isn’t what people want – some people want to see the retweeter not the original poster.
It also shows that if you force a community down a particular path, they will work against you to do continue to do what they want to do.
That appears to be an elegant solution to get Retweets working correctly. This isn’t much different than seeing tweets connected to replies via Twitter Seach so I don’t see any technical limitations. Ev’s post did imply that supporting Comments *were* planned for an upcoming version so hopefully they keep some of your key points in mind.~@JesseLuna
I started to write a comment here but it turned into a blog post.
http://www.scripting.com/stories/2009/11/25/whatsSoSacredAbout140.html
Dave
Over the last couple weeks, I have also noted and tweeted the limitations I’ve found in the new retweet format. Instinctively, I resist allowing someone to have this kind of control over what I “publish.” More and more, I’m moving to a twitter client like TweetDeck, so that I can edit a tweet before retweeting and replying.
Perhaps Chris Messina’s post about microsyntax (http://is.gd/53JyP) was a bit tongue-in-cheek, it does have merit. I have, in fact, begun to adopt it. For example, in the case of retweeting a retweet the format would be “/via @name2 @name1.” The order of attribution denoting the chronology. Regardless, your article here is full of thought provoking ideas.
This is an excellent article. The problems are clearly laid out and the solution is elegant. This has probably already been tweeted many times, but have not seen it, so I will post it again.
I hope that Twitter is listening!
Cheers,RLM.
Thank you. Thank you. Thank you. I thought all the people who should have been objecting had become pod people. I am deeply unhappy about the new RT. I love your solution. How do we get Twitter folks to listen? Thanks for this. I was feeling so alone! And worse, I was feeling like a marginalized malcontent. And you know how painful that can be. 🙂
[…] new retweet feature on Twitter and issues about attribution. He links to an article by Alex Bowyer, A better design for twitter retweets, which is a very well thought out post that deserves to be […]
[…] lets you look for specific occurrences even before they happen. Consider @alexbfree’s recent post on Twitter Retweeting, which got picked up by Dave Winer. You can set up an alert to see if Dave sends you […]
Agreed, that does seem like a nice alternative, but what would a Luddite like me with SMS updates get? “Agreed” … er, agreed what? Plus it does seem almost identical to a reply, which you can do right now- same metadata just no @alexbfree in front.
Hi James, you’re absolutely right, there is a problem with SMS updates, which are essentially individual updates where almost no context is possible (due to the 160 character limit).
But I don’t think this problem is unique to my proposal. As Dave Winer observes, there are many kinds of metadata about a tweet, of which retweet is just one. And none of them can easily be included in the SMS.
You can’t easily find out which message a tweet is “in reply to” via SMS either – making conversations already hard to follow by SMS – nor can you browse the rest of the hashtag which the message might reference. Depending on your connectivity you may not be able to follow a URL included. Unfortunately the SMS limits the scope of what can be done.
I think this is a separate technical problem to be solved – how do we communicate context/metadata alongside an SMS tweet. It’s not a reason to try and force metadata inside the 140 character limit or discard paradigms for the main interfaces. Perhaps a single link in an SMS could include all the metadata. Facebook solves it by letting you “request more” by replying with a “n” for next message. Or perhaps MMS could be used to attach additional metadata and context.
As for the similarity to reply, perhaps you’re right, I didn’t choose the best of examples. Here’s a better example. I tweeted something, and @bookoven opened it up to a question to their audience. Really a retweet is a forward not a reply… but critically it’s a forward with comment.
Really like the ‘request more’ idea to get round the SMS character limit.
As for the retweeting, I still think that the Twitter design covers a straight retweet pretty well. If you’re commenting on a tweet (agreeing/disagreeing/expanding on/asking for views on/etc.) a reply does seem to cover the metadata side already. Perhaps just rename ‘Reply’ to ‘Respond’? (And maybe display things as you suggest above.)
Nice description RT, of the problem and an elegant solution. Good job!
What’s nice about the new retweet feature is all retweets collapse down into a single item in my timeline. That way I don’t have to see the same retweet a dozen times. Unfortunately there’s not a great design for collapsing and allowing commenting.
Joseph, you’re right, and there’s probably a big advantage for Twitter too in data storage savings as well.
Having the tweet only show up once may not be a great thing for people who just dip into Twitter every now and then – it means you might miss something which is actually being talked about by a lot of people, because it only shows up once in your stream.
I’d prefer to see it once per mention, and have a “don’t show me this again” option, rather than only show it once by default. But again I suspect it’s a personal preference.
Yes, please.
[…] get an idea of what’s possible, I recommend reading A better design for Twitter retweets. Wouldn’t it be great if we didn’t have to wait for Twitter Corp to try this out? […]
[…] lets you look for specific occurrences even before they happen. Consider @alexbfree’s recent post on Twitter Retweeting, which got picked up by Dave Winer. You can set up an alert to see if Dave sends you […]
[…] in new ways, and all the things that encompasses – user-centric design, productivity, human-computer interfaces and exploring social trends. He used to work for IBM UK, specializing in Voice systems, Java and […]
Twitter can’t even thread normal tweets in a meaningful way. They have a parent ID, “tweet message ID”, reply tweet message ID, user iD’s on all ends, and dates to pull it all together.
Usenet has been threading on message-id since long before http
was even an nacronym most kids know. ( STOP HIJACKING MY THREADS!!! ) 🙂
I have no confidence Twitter could thread based on your suggestions. Oddly, Threading has been solved in mostall coment interactive systems. Twitter is the largest interactive comment system we’ve seen to date. People just like to call it other more fancy things.
Twitter is just a large 140 char message board.
I like your ideas, there are no shortages of methods to solve these problems. The problem is; the problem is not the technology it’s the Twitter technologists.
But I like that. As long as neutrality does not get fubar’d, anyone who can learn one of 20 various scripting languages is free to give it a go their own way.
Twitters API even gets you right up against their system if you want to piggyback. Too bad Twitter uses only a fraction of their own API for their operations. If Twitter ate 100% of their own API dogfood, we may not be talking about this.
If their tools are identical to our tools (API), nothing but feirce competition would result.
Nice post. Well said. Now go hit up that JSON and Restful API and make it happen. Use a firefox plug-in. According to my stats, the lions share of Twitter is still web based ( twatagent.com ) for metrics on twiter user agents.
[…] 2009, the year of Twitter (including two of our most popular posts, Twitter’s not a site, it’s a protocol and A better design for Twitter retweets). […]