This piece started life as a Twitter thread – ironically – which I posted recently. Read the thread here, or read on.
I’ve recently come to the conclusion that many of Twitter’s toxicity issues essentially come as a consequence of their own software’s behaviour.
Algorithms affect so many systems and businesses these days, that it can be very hard to determine whether making a change to a system is beneficial overall.
Look at YouTube and their ongoing crises. YouTube’s recommendation engine has become notorious for leading people down rabbit holes of more extreme videos – be they for neo-Nazis, conspiracy theorists or paedophiles.
Their algorithms are built to point you to another video with which you will engage with, respond to in some way, and ideally keep you watching.
While that should be harmless – delivering me a constant set of videos reviewing new gadgets, showing me people fixing bikes and a whole host of DIY-style videos – the evidence seems pretty clear that if you stray off the straight and narrow, you can easily get dragged into a black hole of increasingly outlandish or just creepy videos.
No programmer meant for their algorithm do this, any more than Amazon, who once told someone buying a baseball bat, that “you might also like” a balaclava. Nor did the authors of a Microsoft AI chatbot intend for it to become racist.
It’s incredible hard for algorithms to be fully tested and for developers to ensure that the algorithms are not misused or deliver other unintended consequences. That becomes even more that case when AI – or machine learning – is placed into the mix.
Which brings us to Twitter.
Twitter’s history is long and varied, including widespread use of SMS in its early days. These days it seems likely that most Twitter usage is on a mobile platform. And over time, Twitter has tweaked what we see quite a lot.
Much of that tweaking was done to make life easier for novice Twitter users. Because we forget, if you’re starting from scratch, using Twitter can have a very steep learning curve.
You need to find people to follow; perhaps work out which friends and colleagues you should be following. And then there’s the issue that with a “reverse chronological” timeline as Twitter launched with, users may miss important Tweets from people they follow.
To make things |”easier”, Twitter messed around with the timeline. No longer was it just the most recent Tweets. They highlighted things you might have missed that seemed to be popular with others.
At times, this went horribly wrong. During a breaking news event, there might be lots of information being shared that was only relevant at the precise moment it was sense. Maybe the situation wasn’t yet clear – casualty numbers were wrong, or important details were still unknown. When, 24 hours later, those early erroneous or irrelevant Tweets were appearing in timelines, it led to frustration.
For a while Twitter offered choices. Even until quite recently, users of Twitter’s beta software on Android could limit themselves to just a reverse chronological feed. But the non-chronological feed was the default.
As I write now, the default option in the official Twitter mobile apps is the non-chronological version. Users can choose to switch to a “Latest” view. But that’s temporary. You will be switched back a day or so later. You can’t permanently switch it off. (It also seems that the iOS app is faster to return you to the non-chronological version than the Android app). Twitter’s new redesign of their desktop view similarly only gives you the chronological view if you click on the “sparkle” button near the top.
Once upon a time, Twitter’s own mobile app was distinctly sub-optimal. There were better third party apps. And Twitter had a fairly generous API that allowed other developers to write their own apps and embed the Twitter feed into them. But Twitter fell out of love with that ideal, and while there remains a decent selection of third party apps, Twitter is either not adding, or is removing features from those apps.
If you want the full Twitter experience, Twitter wants you to use its own apps.
But aside from a number of third-party developers perhaps losing their livelihoods, what’s the problem?
Let me tell you a little about how I use Twitter. I follow quite a lot of accounts. They are many and varied .
Sometimes they’re work related – think radio, audio, podcasts, media.
Sometimes they’re related to my hobbies and interests – cycling, photography, videography, politics.
Sometimes they’re just people I’ve met or who I find to be interesting for any number of reasons.
In general, I don’t follow firebrands or shouty people. That’s not to say that some people I follow don’t get especially exercised about particular issues. But if someone is too overbearing, I’ll quietly unfollow them. I’m not really interested in the thoughts of some keyboard warrior somewhere who sits watching TV and reading Twitter all day, retweeting all the appalling things they’ve found. It’s all thoroughly reductive.
I don’t necessarily agree with all the views of everyone I follow, but I don’t out and out follow people I actively dislike. I’m not a journalist covering politics.
So here’s the real thing I’ve noticed with Twitter:
Twitter is a much nicer place when you’re reading it in reverse chronological order
That is to say, the non default version of Twitter is much nicer than the default view!
It comes down to ‘engagement.’ In the regular view of Twitter on mobile, as well as getting a few recent Tweets, you’re presented with Tweets from accounts that you follow – but taken from across the last 24 hours. These are essentially supposed to be Tweets that you might have missed. However, Twitter is picking them based on the engagement or response that they got.
Invariably that means the most “shouty” Tweets get shared. These Tweets may come from people I follow, but they’re not necessarily representative of my entire timeline.
A good example would be someone I follow, quoting something someone else has said, which in turn got a lot of response. While these are indeed the Tweets that have had the most “engagement” they’re not necessarily the most helpful.
A specific instance. I follow a number of political journalists, including those from the BBC. In these heightened “Brexit” times, there is a lot of Twitter shouting about who gets featured on Question Time or Today and in what slots. After the recent Peterborough by-election, Nigel Farage was interviewed at 7.10am on Today. A lot of people were indignant. Farage being interviewed by the BBC! He hadn’t even won the by-election. Someone politely pointed out that his party came within 700 votes of the Labour winner. But there was a stream of anti-Brexit Party responses about the BBC being biased. Of course, Labour did get the much more prestigious 8.10am interview slot that day, but let’s not let prejudice get in the way of things. (NB. I hold few candles for any political parties right now.)
That’s a single example. But what Twitter tends to highlight are Tweets that drive “conversation” – except that you can’t really have a conversation on Twitter. It’s not a place for nuanced debate.
Here’s another example. I’ve only blocked one person on Twitter. They happen to be a quite famous presenter of a certain breakfast TV show. That person I happen to find quite obnoxious. Because they’re blocked, Twitter makes that person’s Tweets unavailable to me. But because that person says lots of goading things, they get quoted quite a lot, and people get annoyed by them. Twitter then highlights those responses, despite the fact that I can’t read what the person said in the first place!
And the thing is, Twitter is very good at finding just the right stories to rile me up. A cycling campaigner posting some monstrous piece of road design that makes things worse for cyclists rather than better; something provably false that the President of the United States is saying (NB. I don’t follow Trump’s Twitter account); something really stupid and provocative that a speech radio host has said.
Despite yourself, you click the link. Read the stupid things. Then read a bunch of responses that all take the same sentiment. You get annoyed by what you’ve read.
You know you should step away from the phone, but you can’t help yourself.
What a great way to start the day!
If you instead just view a reverse chronological feed, then most of that nonsense gets buried. Some of it will show up – depending on how many people you follow, how often you check Twitter and how often the nonsense gets retweeted. But the chances of seeing it are significantly lower.
Twitter suddenly becomes a much nicer place.
There’s much less outrage and much less provocation. If that’s the sort of thing that floats your boat, then “engage” away. Enter those debates and you’ll be copied into the responses. And of course, you can check trends if you want to.
But Twitter becomes much more useful when its chronological.
Tweetdeck, the Twitter desktop app that works in a browser tab, is only offered in reverse chronological order – you don’t get ads either! For a hardcore Twitter user – and especially journalists – it can be essential. But it’s not available on mobile, and wouldn’t really work to its full functionality in a mobile environment anyway.
But all I can say in the meantime is that if you want your blood pressure to ease when you’re looking at Twitter, switch to the Latest view, and keep switching to it when Twitter insists on returning you to the non-chronological Home.