Is Twitter Forcing Me To Scroll More And More?
Small interaction changes are contributing to an unhealthy habit
I’m scrolling more on the Twitter timeline than I used to.
I felt this in one morning in 2017 when I was browsing the Twitter timeline as usual in the bed, waiting for my brain to wake up (or until I start hating myself for spending so long on Twitter). Though I don’t track how many pixels my finger swipes, this strange feeling of increased scrolling seemed real. It wasn’t like this before. I’m definitely scrolling more on the timeline. I started wondering why. And this is the summary.
I begin with analysing the basic user behaviour in browsing the Twitter timeline by using a habit-forming interaction design model, and then look into recent changes Twitter made in their timeline UI, aiming to reveal what (or who?) is causing this strange feeling.
The Hook Model in Browsing the Twitter Timeline
Why am I so habitually browsing the Twitter timeline in the first place? I signed up in 2008 which is 10 years ago, and I’m still logging in, scrolling the timeline, every single day, multiple times a day. There is literally nothing else that I can think of me doing so such a long time. How am I hooked into this?
The Hook Model introduced by Nir Eyal explains it well. In his book Hooked from 2013, he details how popular digital services use this habit-forming user interaction model. The model illustrates how a user of a digital service forms a habit in using it, in other words, how a user starts using the service with little or no conscious thought. It’s summarised in the diagram below.
The point of this model is in creating a self-reinforcing interaction between a product and a user. Once the user goes through a sufficient number of cycles in a sufficient frequency, the brain forms a circuit between their internal desire and the reward given when the product fulfils the desire. There is no need for £££ advertising. Users spontaneously come back to the product simply by following their internal trigger. Zero-cost retention gain.
Twitter as a service appears many times in the book. The focus here, however, is a micro interaction of scrolling its timeline that I want to analyse through this Hook Model.
Trigger is the first step of the Hook Model. It directs the user’s desire towards the solution provided by the product. There is external one and internal one. For potential users or users who are new to the product, external triggers such as advert, notification, and word of month can be considered. On the other hand for ones who have already formed a habitual usage of the product, internal triggers suffice.
Me: moving finger to Twitter’s app icon even before consciously thinking “I don’t want to wake up just yet. Let’s see what’s happened while I was asleep”
Here the vaguest feeling of not wanting to wake up just yet is linked to the following action.
Action is a step done by a user in anticipation of receiving a reward. Launching an app or a browser to see a social media feed, tapping some buttons, scrolling the feed, etc, are all Actions. What’s important here is that the user interface for the required Action is carefully designed in order not to demotivate or distract the user’s desire. UI design techniques such as simple and unmistakable buttons and gesture manners following UI guidelines, are all to shine here.
Me: scrolling the Twitter timeline looking for interesting tweets, images, articles, threads, people etc…
Variable Reward is a widely used technique in gambling. It’s based on researches done in early 20th century which showed that the frequency of a particular action becomes significantly higher when the probability of a reward given for the action varies than when the reward is given every time the action takes place.
Me: one tweet appears after another from the bottom of the screen. Some are interesting but mostly boring, which nudges me for another action of scrolling…
Investment is done by a user after receiving the reward, to reinforce the user’s engagement to the service. Typical investments include liking a social media tweets, completing the user profile, and curating the following list. Successful investments increase the value of the product as well as nudge the user to overvalue the product, ultimately closing the circuit between their internal desire and the product fulfilling it.
Me: Liking and retweeting interesting ones and following newly found interesting people, expecting the whole new tweets I will see in the timeline tomorrow (or an hour later)
…well, this shows how happily I am hooked!
Why Am I Scrolling More Recently?
What has changed?
Well, I changed…
My behaviour in internet in general has certainly changed. In the last couple of years I experienced some big changes both at work and at home. How I use Twitter must have been affected by the surrounding environment at any given time. I’m following English-speaking accounts more, tweeting and replying less, browsing more; less personal tweets and more work-related tweets. However, these trends have been persistent since the time before I started feeling the increase in scrolling on the timeline.
But Twitter Changed More
The user interface of Twitter changed significantly. There are many more but here are the ones that I can think of potentially influencing my scrolling behaviour.
- Twitter Cards got expanded in the timeline.
- Max number of characters one can put in a tweet became 280.
- The order of tweets in the timeline became un-chronological.
Through The Hook Model, we can understand how each of these changes affect our behaviour at a very micro level.
Twitter Cards in the Timeline
Twitter Cards is a graphical element included within a tweet which often contains an image, title, excerpt of the webpage linked from the tweet. Example of Twitter Cards is shown below.
This used be only expanded and visible when we click a tweet to show its detail. However, sometime in 2016, this started showing in the general timeline too. This change, of course, increases the area that one tweet occupies. Now if I open my Twitter timeline on my 15-inch laptop, I can only see three tweets at a time. Although now I can make decision of whether a page linked from a tweet is worth clicking by simply looking at images and titles, this naturally increases the amount and time of scrolling required to encounter an interesting tweet. In the Hook Model’s cycle, the amount of Actions got bigger to received the Reward.
280 Characters in One Tweet
Twitter has been restricting the number of letters people can input in one tweet with the maximum of 140 characters since the beginning of the service. However they recently doubled it up to 280 characters to enable everyone to “express themselves easily in a Tweet”. Japanese, Korean, and Chinese remain to have the 140 characters cap for their dense writing semantics, but users following English-speaking accounts are affected by this change — another increase in the area that a tweet occupies. Even though this doesn’t mean that every single tweet becomes twice as long as they used to be as their research tells, it’s still an increase. Especially for people like me who doesn’t speak English as their first language but has their timeline filled with English, small increase in the volume does matter. Here again, the Action required to reach the Reward became slightly more difficult to execute.
Before, the Twitter timeline was listing every post in reverse-chronological order (new one comes on top of older ones). In 2016, Twitter abandoned this and made the algorithm-based timeline as the default setting. Although they still provide a chance to opt out by unticking “Show the best Tweets first” setting, the number of users who do this must be very small.
Around the same time, a series of replies to one tweet became a combined ‘thread’, pushing the original tweet, which is sometimes days-old, up to the top of the timeline as soon as someone reply to it. Moreover, Likes done by people you’re following are now flowing into the follower’s timeline. Combined with the retweet function, now many many tweets that are not necessarily happening now are shown in my timeline, even though I do opt out from the algorithm-based timeline.
If we apply The Hook Model to this one, it’s the variability of the Reward that’s increased. In a chronological timeline, the probability of one tweet being rewarding decreases as you scroll down to the past, and reaches zero as soon as you hit the tweet you’ve already read. On the other hand with the algorithm-based timeline, or non-chronological timeline with a lot of interrupt, your scrolling Action never stops because you cannot know when to stop and how interesting the next tweet will be.
In short, these three changes are influencing our scrolling behaviour in the following ways:
- Increased amount of Action required to gain a Reward
- Increased Variability in Reward that triggers further Actions
This seems like a set of changes that make one behaviour easier while another more difficult and somehow cancel each other out. But there is one parameter that will be boosted with the combination of them — the time spent on the timeline.
For users like me who have already formed the timeline-browsing habit, there is a set amount of stimulus that the brain needs to be fed with before closing the app. These changes increase the labour required to gain a small part of that stimulus, and at the same time, encourage further labour even if the labour doesn’t result in a success. It is obvious that the time and amount spent for the labour are made greater by these changes. I am forced to scroll more and more, and this influence is happening at a very micro, tacit, and unconscious level of interaction.
Why is Twitter doing this?
Although this influence may well be a byproduct of some changes intended to improve some other parts of the service, there is a clear benefit for Twitter to do this — the more people scroll the timeline, the more likely they see ads.
The 280-characters change mentioned above was symbolic in that Twitter now officially regards users who tweet and express as their primary users. Before, the experience of tweeting was different depending on the language you speak in, as Edward Snowden also pointed out. The 280 characters cap evens out the difference and provides with a universal tweeting experience. Twitter want people to tweet more rather than just browsing and doing nothing. But this should not mean that they can neglect, or worse, exploit users who are browsing the timeline, especially when that behaviour is habitual and difficult to quit.
One of the chapters in Hooked suggests designers to keep the application of The Hook Model within a situation where users have the sense of ownership and don’t feel manipulated by the service. Once people start pausing in the middle of the mindless scrolling and questioning if they’re forced to do so, they can be easily unhooked. Though I don’t have a plan to quit Twitter anytime soon, I hope that this habit is a healthy one.