How to stop online Twitter mobs from ruining lives
It’s time for me to make a confession.
I have said and done some truly dumb things in my life.
Especially when I was younger. Sometimes I said things to people which make me cringe in recollection.
Luckily for me, none were written in the indelible, cloud-stored, 99.99% uptime world of the internet. They were just said – and the words disappeared as soon as they were done. I didn’t ever stand in front of a stadium crowd or a TV camera and insult someone, recorded for eternity. I’m probably the only one who remembers the dumb things I said - and even then my memory of the actual words is likely corrupted by time.
Actually, I did write some dumb stuff and let it loose on the internet. I once used a work email address to send what I thought was a funny story, but in immediate hindsight was something that reflected badly on me (it was about myself, sent to friends). Thankfully that one has disappeared into the maw of long-deleted inboxes and won’t be turning up in any internet searches.
Lately, increasing amounts of people have made the mistake of putting their dumb thoughts onto Twitter. Even though the majority of Twitter users has only a handful of followers, the mathematical power of 6 degrees of separation means you’re never far from your thoughts being amplified in the social media megaphone.
A recurring theme of our interconnected time is sports stars, politicians and well known people making a horror tweet and losing their sponsors, jobs and sometimes careers as a result. But increasingly it is beyond just the circle of ‘famous’ people that this problem is starting to bite.
A pair of recent tweets from Paul Graham of Y Combinator got met thinking about this problem more.
Both refer to the case of Justine Sacco, who is not a famous person – or, I should say, wasn’t famous before. She made a tweet before boarding a plane, and that tweet made it into the social media megaphone – so by the time she landed, her life as she knew it was over. An online mob had arisen and decided her fate as judge and jury before she even knew that someone managed to find the tweet manufacture outrage and offense. The Tweet wasn’t funny but if telling unfunny jokes means losing your job I’m lucky I’m not permanently unemployed.
The first tweet above links to an article which suggests a few ways the issue can be dealt with – ultimately there are no easy answers. The author suggests ‘standing up’ for the shamed is the way to go. That’s something I can wholeheartedly stand behind, but taking offence seems to be a way of life for many people and the silent majority is usually, well, silent.
The second tweet shown above was where Paul Graham was suggesting a different design of Twitter would be in a good place to take over where Twitter is weak – probably hinting at a Facebook eliminates MySpace scenario.
But I’d like to think there is another way, fixing the problem in Twitter itself. Ultimately technology enables the formation of online mobs so the answer can be within the technology itself.
The danger of mobs in other walks of life is well known – irrational mob behaviour and the madness of crowds is well documented. Solutions have been developed in many different areas to circuit-break the action of crowds and mobs.
In Stadiums and other large, physical crowds, barriers are placed to prevent a crush from happening. By placing horizontal fencing across the crowd, a large critical mass cannot be formed. These designs are as a result of several tragedies where large crowds unintentionally harmed people by crushing into an area.
Above : Terrace with Crowd Crush barriers at Morecambe Football Club : Source Wikipedia
The applicability of physical barriers has only conceptual value when it comes to online mobs – taking the concept further requires borrowing from a different sphere, where information flow is more important than physical movement.
My second example is from the financial markets. Some markets establish price movement limits to prevent large intra-day moves. An event may trigger a rush for the exits of the market (or market entries, though as the saying goes the bull goes up the stairs while the bear goes out the window). Particularly in the case where fear is dominating over rational thinking – or even automated trading systems reacting to inputs – the rush to sell creates a critical mass which may drive a price much further than it otherwise should. These selling mobs (whether real people or automated systems) can create havoc with margin calls and market writedowns, as well as destabilizing confidence in the market.
To prevent large price moves, exchanges implement price limits based on the prior trading day closing price, defining a maximum range that the price can move. No sales are allowed on that day outside that price range, known as the ‘limit’ price. When a market moves and hits the limit, it is termed as ‘locked limit down’ and the market essentially ceases trading unless a buyer can be found at the limit price. It's also possible to lock a market limit up, where a buying frenzy hits an upper limit.
How Twitter can protect Ordinary People from themselves
The commonality between the two examples above is the placement of barriers to prevent mobs running out of control and going further than they may do as rational, thinking individuals.
The essential problem with the formation of social media mobs is the bandwagon effect where a trending topic is enthusiastically taken up by geometrically more people as it trends. This is the valued ‘going viral’ that nearly everyone wants, except when it is about their mistake rather than their triumph.
To prevent something going viral, it needs a barrier to limit the spread. Real viruses are contained with quarantine. That’s what my idea is. Much like the way credit card fraud is detected by a change in spending pattern, a sudden change in retweets or mentions of a specific user would indicate something has changed around that user. A previous nobody who is suddenly trending is an abnormal pattern that should be easy to detect. Let’s call it the ‘Unexpected Retweet Algorithm’.
Above : A hypothetical Retweet/Mention graph for 4 Twitter users
My fictitious graph tracks the Mention/Retweet for 4 users over a 12 month period. You can see there are 3 famous people – a singer who is constantly talked about, a politician who manages to slowly get more people to talk about them (but perhaps loses an election in December), and a sports star who gets talked about during the season but gets few mentions in the off season.
And then there is a ‘Regular Joe’ who manages to catch the attention of a Twitter mob and goes from a few mentions amongst friends to more mentions than a singer with an active PR team.
The Unexpected Retweet Algorithm would seem to be trivially easy to write and discover Regular Joe in this scenario. Once a person like Joe is found, their tweet(s) would be flagged for review by the algorithm. A live review team would then check the account, and the retweets and mentions would be quarantined from audiences until actual human review had taken place. Any trending hashtag about that person would hit a barrier and a critical mob would not form as the trending subject was quarantined.
The purpose of the algorithm is to find things that become suddenly popular, and the job of the review team would be to prevent online mobs forming to blow some small thing out of proportion. If ‘Regular Joe’ had made a crass or insensitive tweet that was then being taken up enthusiastically by people outside their network, they could quarantine the tweet and Joe himself off air until Joe was contacted and asked if he wanted to ‘untweet’ it or go ahead and publish it. If the former is the case, the tweet is deleted and Joe’s account goes quiet for a week. If it’s the latter, then normal service resumes.
As we saw in the Justine Sacco case, the time taken to go from poorly conceived tweet to global meltdown was shorter than a flight from London to South Africa, so the detection software would have to move fast and the review team faster. These are all solvable problems that have been dealt with in other industries. I once bought a drink from a vending machine after getting out of a plane and finding myself short of local currency, used my credit card. The call from my credit card company came before I even got the cap off. Was I in this location? Did I just buy a drink from a vending machine? They knew this wasn’t my usual location or behaviour and it was flagged immediately.
The Spectre of Private Censorship?
The immediate reaction in any true internet enthusiast from this proposal should be one of shock and indignation that such censorship could even be considered, especially one administered by the host of the platform. I understand that position, and I’m a true believer in free speech and the need to say and be heard. I love the bawdy free exchanges that can be found on everything from programming language religious wars to sports and political discussions. People like to expose themselves to real arguments and debate, there’s no doubting that. May such places continue to thrive and expand.
A second reaction would realize an intervention in the viral system would make it harder for breakout hits to reach the world in a flash. This is one of the great things about platforms like Twitter – the ability for the unknown to become known. Care should be taken before meddling with the power of the platform.
Both criticisms are valid but less important than the very real damage that self-appointed and self-forming mobs can do to an ordinary person. An 18 wheeler is useful because of the load it carries, yet that load needs to be handled carefully, and safety systems are a necessity to ensure the roads can be shared.
The answer for the first objection (censorship) is that Twitter would be providing users an ‘untweet’ function if they realized in hindsight of the error of their ways. A tweet caught and held in review would enable the untweet function. If someone really wants to push a tweet out there – Twitter would let them. And viral hits – of the positive kind – would be all too easy to identify by a review team and left to go on their merry way. Determining a hilarious cat photo tweet is the work of seconds for a human being. Determining what is an offensive tweet that the person might regret is not that difficult – the team would merely flag it for review, not banish it from the internet forever. It’s a review team, not a judge and jury which will brook no challenge.
The future without an effective Untweet button
The very real future that Twitter faces is that ordinary people will start to shy away from the platform, wary of unseen and unknown people cruising through, looking for an offense to take or an issue to amplify. As any advocate of free speech will tell you, it’s not the punishment that is the problem, there is usually someone willing to stand up for their beliefs. It’s the climate of fear which squelches free exchange that is the issue. That's likely why so few people speak out in support of the mob victims. This is not about going out and deliberately being offensive, it’s about walking a tightrope and having to over think every single thing published. Ultimately many will decided that playing with fire is just too hard and extinguish the flame forever. I am already leaning this way. I have removed the Twitter app from my phone – I have to be sitting at my desk thinking to tweet.
For any community to thrive, it needs continuous engagement by existing users and a steady flow of new users. Any measure which makes people more comfortable is likely to be beneficial for the long term of that platform anyway. When any technology goes mainstream - Twitter is well past that point – a level of safety for users is not only desired by inevitable. Saving users from themselves is a killer feature to add to any technology.