Carl Miller: Twitcidents contain vital information about how people are reacting.

This Is Your Brain Online: The challenge of the twitcident

This Is Your Brain Online: The challenge of the twitcident

By Carl Miller

Major events, from terrorist attacks to football matches, are now accompanied by a surge of activity on Twitter. These are 'twitcidents', complex clouds of tweeted reaction, containing questions, interpretations, jokes, rumours, insults, even calls to violence. They are a new kind of event aftermath and, indeed, an increasingly important part of the event itself.

Whether you're a train company, a supermarket, a government or a political Party, twitcidents matter. Amidst the train crash, scandal, recall of a product, spread of food poisoning or flooding or a debate, twitcidents contain vital information about how people are reacting.

We collected and broke an important and tragic twitcident down to try to get a better idea of what they are, and what they mean. On May 22nd last year, Fusilier Lee Rigby was tragically and brutally murdered in South London. That day, the number of tweets mentioning the Metropolitan police's Twitter account leapt from its usual average of a few thousand to over 14,000.

Using coding techniques familiar to sociologists, we found ten different kinds of tweet within the twitcident:

Bots: Roughly half of the tweets were fake – sent from non-human, automated 'bot' accounts. One message was especially highly propagated: “Half the things people are tweeting should put them in jail @metpoliceuk The news hasnt even got confirmed stories yet! #Woolwich #Racism." We removed these from the analysis.

Offline evidence (two per cent): A small number of tweets were eyewitness accounts of crimes they had witnessed or were aware of happening offline: "@metpoliceuk yesterday in coach leaving from London Victoria at 10pm to Birmingham a person was openly racist towards another individual." This included, against the advice of the police, information that should have been sent in on 999: "@metpoliceuk #Stalkers with #listening devices threatening #Jamaican lady near Meadowbrook High Sch St Andrew #Jamaica 11:48am 23.5.2013."

Online evidence (20%): More common was the referral of social media content itself as evidence about alleged or supposed online and offline crimes. The majority were referrals of heated tweets about the Woolwich killing, but other tweets contained possible evidence of alleged driving infractions, fraud, involvement in riots, paedophilia, child abuse, drug-taking, cyber-bullying and animal abuse. Very often they came attached with an investigation, typically in the form of a twitpic or other photographic 'evidence'.  "@metpoliceuk stop these people http://t.co/xxx"

Indirect mention (34%):  A large part of the twitcident used the @metpoliceuk handle indirectly, as a way of identifying the police, but not apparently requiring or demanding an answer from them. People did this for a wide array of purposes, including comments on performance – both criticism and (more commonly) support: – "Called @metpoliceuk over an hour ago to report 2 pissed fellas using the square park as a toilet, bin and bed – no sign of 'em typical."

Conversation/Question (15%): The twitcident also contained tweets that did expect a response. This included for simple information: "@metpoliceuk Hi who do i contact about a recent fire?" and people resorting to Twitter due to other failed attempts to reach the police: "@metpoliceuk every number i've been given to call you back regarding a crime, either "Unavailable" or Rings Out! Please Help! #Frustrating."

Petition (four per cent): There was also strong evidence of systematic cooperation by large bodies of people to concertedly appeal and petition the police on Twitter to influence their policy. This included a campaign calling for the arrest of Altaf Hussin, the leader of the Pakistan political party Muttahida Quami Movement, and for the police to release more information on their investigation into the disappearance of Madeleine McCann.

Rumour/trolling (five per cent): Of course, there were also tweets that were intentionally spreading misinformation and trying to get a rise out of the police. Some users appears to be trying to link the Woolwich killing, in the immediate aftermath, to the Pakistani group PTI: "There are NO Good Taliban They All Are Bad & London Incident Is Occurred Today Shame #PTIBehindLondonAttack @David_Brown @MetPoliceUK."

Sousveillance (three per cent): The police themselves were watched on Twitter: "@BBCNews Why @metpoliceuk can unmarked police car BX5XXXX choose to go through a red light on hanger lane wiv no blues & 2's on? #1rule4us."

The rest were either retweets or (a common category on Twitter generally) utterly indecipherable. 

Of course, twitcidents will be as varied as the events that provoke them. Whilst the contours of the Woolwich twitcident is specific to the aftermath of a brutal, highly publicised and incendiary crime, I think it also tells us something broader: that they will throw tweets at organizations of radically different utility and urgency, requiring different kinds of responses, probably from different people.

Some tweets are digital-tip offs – useful as evidence for investigations that need to be stored in a tamper-proof way. Some are words from correspondents at the scene – valuable glimpses of a dynamic scene or context – and these need to go to decision-makers in a way that they can trust and assess. Some are digital knocks on the door – opportunities to engage with the enraged, disaffected or just plain confused by setting the record straight, providing reassurance and answering questions. Some are junk mail – bots, spam, deception, trolling and nonsense that need to be thrown away.

Listening to twitcidents will become more important and urgent. They will make our reactions smarter and quicker, make the boundaries between organizations and the people they serve more porous and open, and, ultimately, even save lives. It's time we started doing it.

Carl Miller is an associate of Demos working on social media, the internet and security.

The opinions in politics.co.uk's Comment and Analysis section are those of the author and are no reflection of the views of the website or its owners.