really? honestly?

The ABC of crowdsourcing in a crisis

Miscommunication is as old as communication. The dark side of all interconnectivity is its power to transmit the salacious, the fictitious, the misguided – often virally, often uncontrollably.

From the very first email: “virus warning, tell all your friends” there’s been natural – if unfortunate – exploitation of the very human wish to help, collaborate and communicate.

When this happens on Twitter, one big difference is that the explosion of communication happens much more quickly. Real-time response can of course be highly desirable. If you’ve just lost your child in a crowded train station, for example.

Yesterday, this tweet sparked an avalanche of Re-Tweets, alerting hundreds (maybe even thousands?) that a 7 year old girl had been lost.

If you’re a parent, it’s probably happened to you. The most awful feeling. That paralysis, that fear: Where do I get help? Do I stay put? Do I run wildly around searching? In which direction? Who should I tell? Should I make a sudden, very un-British public demonstration of the situation?

There are no perfect answers – but clearly in this case the report was serious enough to have already got to SE1, and the tweeting began.

I was heading across the river towards Waterloo anyway that afternoon, and kept my eyes open for a child as described. Probably many others did, or gave it some thought. Crowdsourcing at its best – unorganised, viral, organic, with a unifying purpose, but nothing else by way of structure to get in the way… Remember #uksnow? 😉

As I walked, I thought about whether there was a ‘best practice’ to using social media like this – every instinct telling me that ‘central places to report’, a #lostchild hashtag convention, a systematic urban-grid-search-plan with real-time mapping (thanks to @adrianshort for that) probably all had as many drawbacks and impracticalities as they’d offer by way of benefit. Nice intellectual exercise though.

Eventually I asked one of the British Transport Police on the station if “the child was still lost” – and got the answer: “oh, the 7 year old, no she’s been found”. Which was enough to assure me that we were communicating about the same thing, and I had enough confidence to tweet this as an update, (which did get RT’d but probably with less gusto than the original alert). And I notified SE1 so that they could update their site (which they did in a slightly curious way).

So, digesting all this, I offer the following suggestion on ‘good’ – not ‘perfect’ – handling of incidents like this.

A: Authority. What authority are you drawing on for your information? “a friend of a friend says that this new virus threat is…” wasn’t good enough to spam all your friends, and it’s not good enough for a RT, imho. So, rule of thumb: if your source is more ‘official’ or evidently better connected on the ground than you are (yellow jackets, radios or established websites are pretty good indicators here), then this becomes your Authority; just make sure you reference it.

B: Broadcast. If you have confidence in your source, tell your networks. That’s what they’re there for.

C: Close the loops. Perhaps the most important bit, but guaranteed to be the one that gets missed the most. With your broadcasting comes a responsibility: either to follow up and update yourself, or to transmit an update that you hear of (based on a suitable Authority, of course) to your network in just the same way as you’d broadcast the alert. In some ways the closure is just as important as the alert – it builds credibility around the whole communication process.

With crowdsourcing, no one’s in charge. No one ‘owns’ an incident. All information has some inaccuracy, and risk. Fictitious children will be searched for, and sacks of postcards delivered to an address down the road from where a child recovered from cancer five years ago.

But think A, B, C next time you pass on something. Particularly if it’s as emotive and real as a lost little girl.

And if you have a great idea to managing distributed information and agents in situations like this, I’d really love to hear it.

Filed under: Uncategorized, , , , , , ,

Why Andrew Stott shouldn’t just blindly follow everyone in return

A few inevitable comments are buzzing out there on @dirdigeng. As with any high-profile newcomer to the twitterverse, the most obvious being: “but why aren’t they following anyone?” Which is usually shorthand for: “why aren’t they following me?”

So here’s Andrew – one of the most experienced government/change/technology experts there is. Someone who not only has real technology heavy-lifting experience, but has been brave enough to challenge orthodoxy in trying to make something real out of this ‘Transformational Government’ concept. Brave enough to stand personally at the centre of massive re-engineering of thousands of unnecessary government websites. Brave enough to take on this new role.

This point about technology heavy-lifting is quite important. There are massive differences of scale between a here-today-gone-tomorrow social media enterprise set up in someone’s back bedroom using a bit of WordPress and an old server nobody else wanted, and a service that has to be designed to serve hundreds of thousands of people, without fail, for perhaps ten years. A period of time over which technology change will certainly test the relevance of any solution that seemed good to start with. This is not to say that great and small government services should always inhabit separate worlds – they do have to be brought together more closely, with more agility, and flexibility around monolithic technology contracts – and this could well be a useful priority for DirDigEng to work on.

Andrew gets this. Look at what is still seen as one of the most transformed services, using online delivery to stop us all queuing in the Post Office with bags of paper to get our cars taxed. Andrew was right at the heart of making that happen.

The greater initial challenge will be engaging with networks and communities that really get digital engagement. One of these, but only one, being the twitterverse. And so to the exam question – who to follow? Let’s look at the options:

1. No one. No. Just looks too weird. Forget that.

2. The boss, in terms of leadership of government’s agenda in this area: @tom_watson?

3. Those that think they should be followed (and say so)?

4. An inner cabal of thirty or so erudite figures?

5. The few hundred most active voices on digital engagement and open government around the world?

6. Everyone who follows?

All of these are possible models, and it’s very much a matter of personal taste, but two I’d advise against in particular are 4 and 6. 4 will reinforce a sense of digital dis-engagement in those who feel it’s all about the usual crowd cosily talking to each other; 6 will be a meaningless exercise as Andrew’s stream will become unintelligible. The ‘blessing’ of getting a follow from DirDigEng will drive all sorts of strange behaviours and attitudes.

Eventually, 5 might emerge organically, but give it time. You have to go through 4 first, which is a risky stage.

So 2 is the choice now, and it’s not a bad one. It’s largely symbolic, but symbols are ok.

Because the real exam question (and one so many twitterers miss) is not: “Who should I follow?” It’s “How should I listen?”, closely followed by “How should I interact?”

And here – as I’ve explained before here, Twitter is built for the listening. It’s not all about the following. So Andrew will no doubt be browsing, reading, dipping and filtering through finely tuned searches – and on finding relevance, interacting through ‘@’ messaging, #tagging, retweeting and so on, just like the rest of us. If he’s any sense. Which he has.

I will now log on to Twitter to find that he’s followed everyone back first thing this morning… Life’s like that.

Filed under: Uncategorized

How Twitter works

Perhaps I’ve got this entirely wrong. And if so, set me straight, here.

But I’ve been analysing Twitter – what it is, why it works, what makes it unique. It’s just the way my brain likes to do things.

And I broke it down like this. There are two principal things going on here:

1 – the maintenance of a user registration system. This allows users to sign up, create a profile, set a contact email, control visibility to others, follow/unfollow others, manage direct messages, block other users, and generally perform a whole suite of tasks that (in my simple world) can be described as:

user profile administration

…and once you’ve played ball in setting up a user profile, you are allowed access to:

The Main Event…

2 – a vast, completely freeform soup of tweets. Literally billions of <=140 character messages that have a couple of little code characters tucked into them: ‘@’ and ‘#’. Both let you (or to be more accurate, your client application*) parse** the soup in ways that make sense to you.

e.g.1. select all the messages beginning ‘@yourname’ and you have your ‘@replies’.

e.g.2. select all messages containing ‘@yourname’ and you have ‘mentions’

[Twitter has buggered around splendidly between terminology over ‘replies’ and ‘mentions’, but these just mask some very simple concepts.]

e.g.3. select all messages containing ‘#hashtag’ and you get a nice user-customised feed relating to a particular hashtag topic… and joy of joys, any user can create a hashtag on a whim, and see where it gets to.

e.g.4. select all messages originating from a particular user, and you have their Twitter feed.

And. That’s. Just. About. It.

What makes it so special is the purity of that soup. Ideally it should be unadulterated – served from the tureen neat so that your particular client application (however you’ve set it up) can season your bowl of soup as you desire.

Your client application may carry out these functions in different ways. It may even restrict you from seeing certain ingredients. But the point, as the #fixreplies issue has shown, is that the USP was Twitter’s completely open approach to offering the soup for seasoning.

Once you cut out all the carrots (to push the analogy just one more time) and removed all ‘half-conversations’ – go and read #fixreplies links if you need more detail – you’ve adulterated the recipe and created something different.

I see tonight that Twitter’s blog says there are tech issues that led them to make the #fixreplies change. I don’t buy that. If Twitter operates as it does in my head, and described above, that soup doesn’t need processing – it’s the sum of billions of short character strings. It’s how various client applications operate that generate technical issues, for each client, if any…

So #fixreplies is a big issue. It mucks about with what was a completely open approach to freeform, user-generated messaging. The old cliché: “you get out what you put in” was never more true than of Twitter. Although personally, I find the Rizla strapline: “it’s what you make of it” more accurate.

Let’s hear it for keeping it open, pure, interpretable, subvertible, hannibal, cannibal, mandible. But that all begs the question of where the business model lies.

I welcome your thoughts…


*this could arguably be seen as the third part of the Twitter mix. As well as user admin and a big bowl of tweet soup, you need to be able to access and process the soup. That’s where clients come in. Twitter has its own client applications, web and mobile (and they’re pretty basic) but the Twitter masterstroke was an open API (application program interface) that allowed any other chancer to write software to process the soup as its users desired. I’d argue that Twitter’s job is to run the user admin and the API and leave everything else to evolve organically.

**to parse – for those not familiar with the word, my definition is: to interpret a bit of information (due to its formatting or context) in a particular way that makes more sense than just reading it literally.

Filed under: Uncategorized